Applications of radiomics and machine learning for radiotherapy of malignant brain tumors

Impacts of Nutlin-3a and exercise on murine double minute 2-enriched glioma treatment

JOURNAL/nrgr/04.03/01300535-202504000-00029/figure1/v/2024-07-06T104127Z/r/image-tiff Recent research has demonstrated the impact of physical activity on the prognosis of glioma patients, with evidence suggesting exercise may reduce mortality risks and aid neural regeneration. The role of the small ubiquitin-like modifier (SUMO) protein, especially post-exercise, in cancer progression, is gaining attention, as are the potential anti-cancer effects of SUMOylation. We used machine learning to create the exercise and SUMO-related gene signature (ESLRS). This signature shows how physical activity might help improve the outlook for low-grade glioma and other cancers. We demonstrated the prognostic and immunotherapeutic significance of ESLRS markers, specifically highlighting how murine double minute 2 (MDM2), a component of the ESLRS, can be targeted by nutlin-3. This underscores the intricate relationship between natural compounds such as nutlin-3 and immune regulation. Using comprehensive CRISPR screening, we validated the effects of specific ESLRS genes on low-grade glioma progression. We also revealed insights into the effectiveness of Nutlin-3a as a potent MDM2 inhibitor through molecular docking and dynamic simulation. Nutlin-3a inhibited glioma cell proliferation and activated the p53 pathway. Its efficacy decreased with MDM2 overexpression, and this was reversed by Nutlin-3a or exercise. Experiments using a low-grade glioma mouse model highlighted the effect of physical activity on oxidative stress and molecular pathway regulation. Notably, both physical exercise and Nutlin-3a administration improved physical function in mice bearing tumors derived from MDM2-overexpressing cells. These results suggest the potential for Nutlin-3a, an MDM2 inhibitor, with physical exercise as a therapeutic approach for glioma management. Our research also supports the use of natural products for therapy and sheds light on the interaction of exercise, natural products, and immune regulation in cancer treatment.

The clinical implications and interpretability of computational medical imaging (radiomics) in brain tumors

Radiomics has widespread applications in the field of brain tumor research. However, radiomic analyses often function as a 'black box' due to their use of complex algorithms, which hinders the translation of brain tumor radiomics into clinical applications. In this review, we will elaborate extensively on the application of radiomics in brain tumors. Additionally, we will address the interpretability of handcrafted-feature radiomics and deep learning-based radiomics by integrating biological domain knowledge of brain tumors with interpretability methods. Furthermore, we will discuss the current challenges and prospects concerning the interpretability of brain tumor radiomics. Enhancing the interpretability of radiomics may make it more understandable for physicians, ultimately facilitating its translation into clinical practice. CRITICAL RELEVANCE STATEMENT: The interpretability of brain tumor radiomics empowers neuro-oncologists to make well-informed decisions from radiomic models. KEY POINTS: Radiomics makes a significant impact on the management of brain tumors in several key clinical areas. Transparent models, habitat analysis, and feature attribute explanations can enhance the interpretability of traditional handcrafted-feature radiomics in brain tumors. Various interpretability methods have been applied to explain deep learning-based models; however, there is a lack of biological mechanisms underlying these models.

Efficacy and safety of simultaneous integrated boost intensity-modulated radiotherapy combined with temozolomide for the postoperative chemotherapy treatment of multifocal high-grade glioma

Background

The multifocal manifestation of high-grade glioma is a rare disease with an unfavorable prognosis. The pathogenesis of multifocal gliomas and pathophysiological differences in unifocal gliomas are not fully understood. The optimal treatment for patients with multifocal high-grade glioma is not defined in the current guidelines; therefore, individual case series may be helpful as guidance for clinical decision-making.

Methods

Patients with multifocal high-grade glioma treated with simultaneous integrated boost intensity-modulated radiotherapy combined with temozolomide for postoperative treatment at our institution between January 2020 and December 2023 were retrospectively analyzed. Multifocality was neuroradiologically assessed and defined as at least two independent contrast-enhancing foci in the MRI T1 contrast-enhanced sequence. Overall and progression-free survival were calculated from the diagnosis until death and from the start of radiation therapy until the diagnosis of disease progression on MRI for all patients.

Results

A total of 42 patients with multifocal high-grade glioma were examined, of which 16 were female and 26 were male. The median age of all patients was 57 years (range: 23-77 years). The median KPS score was 80 (range: 50-100). Complete resection was performed in 10 cases, and partial resection was performed in 32 cases before the start of radiation therapy. The prescription schedule was 54 Gy (1.8 Gy × 30) with an SIB of 60 Gy (2 Gy × 30). Concomitant temozolomide chemotherapy was administered to 40 patients. Median survival was 19 months (95% CI 14.1-23.8 months) and median progression free survival after initiation of RT 13 months (95% CI 9.2-16.7 months). Five patients experienced grade 3 toxicity, none experienced grade 4 toxicity, and no treatment-related deaths occurred.

Conclusion

Multifocal high-grade gliomas can be treated safely and efficiently with simultaneous integrated boost intensity-modulated radiotherapy with concomitant and adjuvant TMZ chemotherapy.

Development and Validation of a Non-Invasive Prediction Model for Glioma-Associated Epilepsy: A Comparative Analysis of Nomogram and Decision Tree

Objective

Glioma-associated epilepsy (GAE) is a common neurological symptom in glioma patients, which can worsen the condition and increase the risk of death on the basis of primary injury. Given this, accurate prediction of GAE is crucial, and this study aims to develop and validate a GAE warning recognition prediction model.

Methods

We retrospectively collected MRI scan imaging data and urine samples from 566 glioma patients at the Xiangyang Central Hospital, Affiliated Hospital of Hubei University of Arts and Science from August 2016 to December 2023. Least Absolute Shrinkage and Selection Operator (LASSO) regression and multivariate logistic regression analysis are used to determine independent risk factors for GAE. The nomogram and decision tree GAE visualization prediction model were constructed based on independent risk factors. The discrimination, calibration, and clinical usefulness of GAE prediction models were evaluated through receiver operating characteristic (ROC) curve, calibration curve, and decision curve analysis (DCA), respectively.

Results

In the training and validation datasets, the incidence of GAE was 34.50% and 33.00%, respectively. Nomogram and decision tree were composed of five independent radiomic predictors and three differential protein molecules derived from urine. The discrimination rate of area under the curve (AUC) was 0.897 (95% CI: 0.840-0.954), slightly decreased in the validation data set, reaching 0.874 (95% CI: 8.817-0.931). The calibration curve showed a high degree of consistency between the predicted GAE probability and the actual probability. In addition, DCA analysis showed that in machine learning prediction models, decision trees have higher overall net returns within the threshold probability range.

Conclusion

We have introduced a machine learning prediction model for GAE detection in glioma patients based on multiomics data. This model can improve the prognosis of GAE by providing early warnings and actionable feedback and prevent or reduce pathological damage and neurobiochemical changes by implementing early interventions.

Ensemble learning-based radiomics model for discriminating brain metastasis from glioblastoma

OBJECTIVE

Differentiating between brain metastasis (BM) and glioblastoma (GBM) preoperatively is challenging due to their similar imaging features on conventional brain MRI. This study aimed to enhance diagnostic accuracy through a machine learning model based on MRI radiomics data.

METHODS

This retrospective study included 235 patients with confirmed solitary BM and 273 patients with GBM. Patients were randomly assigned to the training (n = 356) or the validation (n = 152) cohort. Conventional brain MRI sequences including T1-weighted imaging (T1WI), contrast-enhanced_T1WI, and T2-weighted imaging (T2WI) were acquired. Brain tumors were delineated on all three sequences and segmented. Features were selected from demographic, clinical, and radiomic data. An integrated ensemble machine learning model, i.e., the elastic regression-SVM-SVM model (ERSS) and a multivariable logistic regression (LR) model combining demographic, clinical, and radiomic data were built for predictive modeling. Model efficiency was evaluated using discrimination, calibration, and decision curve analyses. Additionally, external validation was performed using an independent cohort consisting of 47 patients with GBM and 43 patients with isolated BM to assess the ERSS model generalizability.

RESULTS

The ERSS model demonstrated more optimal classification performance (AUC: 0.9548, 95% CI: 0.9337-0.9734 in training cohort; AUC: 0.9716, 95% CI: 0.9485-0.9895 in validation cohort) as compared to the LR model according to the receiver operating characteristic (ROC) curve and decision curve for the internal cohort. The external validation cohort had less optimal but still robust performance (AUC: 0.7174, 95% CI: 0.6172-0.8024). The ERSS model with integration of multiple classifiers, including elastic net, random forest and support vector machine, produced robust predictive performance and outperformed the LR method.

CONCLUSION

The results suggested that the integrated machine learning model, i.e., the ERSS model, had the potential for efficient and accurate preoperative differentiation of BM from GBM, which may improve clinical decision-making and outcomes of patients with brain tumors.

Early characterization and prediction of glioblastoma and brain metastasis treatment efficacy using medical imaging-based radiomics and artificial intelligence algorithms

Among brain tumors, glioblastoma (GBM) is the most common and the most aggressive type, and brain metastases (BMs) occur in 20%-40% of cancer patients. Even with intensive treatment involving radiotherapy and surgery, which frequently leads to cognitive decline due to doses on healthy brain tissue, the median survival is 15 months for GBM and about 6 to 9 months for BM. Despite these treatments, GBM patients respond heterogeneously as do patients with BM. Following standard of care, some patients will respond and have an overall survival of more than 30 months and others will not respond and will die within a few months. Differentiating non-responders from responders as early as possible in order to tailor treatment in a personalized medicine fashion to optimize tumor control and preserve healthy brain tissue is the most pressing unmet therapeutic challenge. Innovative computer solutions recently emerged and could provide help to this challenge. This review will focus on 52 published research studies between 2013 and 2024 on (1) the early characterization of treatment efficacy with biomarker imaging and radiomic-based solutions, (2) predictive solutions with radiomic and artificial intelligence-based solutions, (3) interest in other biomarkers, and (4) the importance of the prediction of new treatment modalities' efficacy.

AI-guided virtual biopsy: Automated differentiation of cerebral gliomas from other benign and malignant MRI findings using deep learning

Background

This study aimed to develop an automated algorithm to noninvasively distinguish gliomas from other intracranial pathologies, preventing misdiagnosis and ensuring accurate analysis before further glioma assessment.

Methods

A cohort of 1280 patients with a variety of intracranial pathologies was included. It comprised 218 gliomas (mean age 54.76 ± 13.74 years; 136 males, 82 females), 514 patients with brain metastases (mean age 59.28 ± 12.36 years; 228 males, 286 females), 366 patients with inflammatory lesions (mean age 41.94 ± 14.57 years; 142 males, 224 females), 99 intracerebral hemorrhages (mean age 62.68 ± 16.64 years; 56 males, 43 females), and 83 meningiomas (mean age 63.99 ± 13.31 years; 25 males, 58 females). Radiomic features were extracted from fluid-attenuated inversion recovery (FLAIR), contrast-enhanced, and noncontrast T1-weighted MR sequences. Subcohorts, with 80% for training and 20% for testing, were established for model validation. Machine learning models, primarily XGBoost, were trained to distinguish gliomas from other pathologies.

Results

The study demonstrated promising results in distinguishing gliomas from various intracranial pathologies. The best-performing model consistently achieved high area-under-the-curve (AUC) values, indicating strong discriminatory power across multiple distinctions, including gliomas versus metastases (AUC = 0.96), gliomas versus inflammatory lesions (AUC = 1.0), gliomas versus intracerebral hemorrhages (AUC = 0.99), gliomas versus meningiomas (AUC = 0.98). Additionally, across all these entities, gliomas had an AUC of 0.94.

Conclusions

The study presents an automated approach that effectively distinguishes gliomas from common intracranial pathologies. This can serve as a quality control upstream to further artificial-intelligence-based genetic analysis of cerebral gliomas.

Artificial intelligence and omics in malignant gliomas

Glioblastoma multiforme (GBM) is one of the most common and aggressive type of malignant glioma with an average survival time of 12-18 mo. Despite the utilization of extensive surgical resections using cutting-edge neuroimaging, and advanced chemotherapy and radiotherapy, the prognosis remains unfavorable. The heterogeneity of GBM and the presence of the blood-brain barrier further complicate the therapeutic process. It is crucial to adopt a multifaceted approach in GBM research to understand its biology and advance toward effective treatments. In particular, omics research, which primarily includes genomics, transcriptomics, proteomics, and epigenomics, helps us understand how GBM develops, finds biomarkers, and discovers new therapeutic targets. The availability of large-scale multiomics data requires the development of computational models to infer valuable biological insights for the implementation of precision medicine. Artificial intelligence (AI) refers to a host of computational algorithms that is becoming a major tool capable of integrating large omics databases. Although the application of AI tools in GBM-omics is currently in its early stages, a thorough exploration of AI utilization to uncover different aspects of GBM (subtype classification, prognosis, and survival) would have a significant impact on both researchers and clinicians. Here, we aim to review and provide database resources of different AI-based techniques that have been used to study GBM pathogenesis using multiomics data over the past decade. We summarize different types of GBM-related omics resources that can be used to develop AI models. Furthermore, we explore various AI tools that have been developed using either individual or integrated multiomics data, highlighting their applications and limitations in the context of advancing GBM research and treatment.

Correlation of Edema/Tumor Index With Histopathological Outcomes According to the WHO Classification of Cranial Tumors

BACKGROUND

Metastatic brain tumors are a prevalent challenge in neurosurgery, with vasogenic edema being a significant consequence of these lesions. Despite the critical role of peritumoral edema in prognosis and patient outcomes, few studies have quantified its diagnostic and prognostic implications. This study aims to evaluate the correlation between the edema/tumor index (ETI) and histopathological outcomes according to the 2021 WHO classification of cranial tumors.

METHODOLOGY

We conducted a retrospective analysis of Digital Imaging and Communications in Medicine (DICOM)-format magnetic resonance imaging (MRI) data from May 2023 to May 2024, applying manual 3D volumetric segmentation using Image Tool Kit-SNAP (ITK-SNAP, version 3.8.0, University of Pennsylvania) software. The ETI was calculated by dividing the volume of peritumoral edema by the tumor volume. The study included 60 patients, and statistical analyses were performed to assess the correlation between ETI and tumor histopathology, including Receiver Operating Characteristic (ROC) curve analysis for cutoff points.

RESULTS

A total of 60 patients were included in the study, with 27 males (45%) and 33 females (55%). The average tumor volume measured by 3D segmentation was 46.9 cubic centimeters (cc) (standard deviation [SD] ± 25.6), and the average peritumoral edema volume was 79 cc (SD ± 37.5) for malignant tumors. The ETI was calculated for each case. Malignant tumors (WHO grades 3 and 4) had a mean ETI of 1.6 (SD ± 1.2), while non-malignant tumors (WHO grades 1 and 2) had a mean ETI of 1.2 (SD ± 1.1), but this difference was not statistically significant (P = 0.51). ROC curve analysis for the ETI did not provide a reliable cutoff point for predicting tumor malignancy (area under the curve [AUC] = 0.59, P = 0.20). Despite the larger edema volume observed in malignant tumors, the ETI did not correlate significantly with the histopathological grade.

CONCLUSIONS

This study found no significant correlation between the ETI and the histopathological grade of brain tumors according to the 2021 WHO classification. While malignant tumors were associated with larger volumes of both tumor and peritumoral edema, the ETI did not prove to be a reliable predictor of tumor malignancy. Therefore, the ETI should not be used as a standalone metric for determining tumor aggressiveness or guiding clinical decision-making. Further studies with larger cohorts are required to better understand the potential prognostic value of the ETI in brain tumors.

The emerging role of artificial intelligence in neuropathology: Where are we and where do we want to go?

The field of neuropathology, a subspecialty of pathology which studies the diseases affecting the nervous system, is experiencing significant changes due to advancements in artificial intelligence (AI). Traditionally reliant on histological methods and clinical correlations, neuropathology is now experiencing a revolution due to the development of AI technologies like machine learning (ML) and deep learning (DL). These technologies enhance diagnostic accuracy, optimize workflows, and enable personalized treatment strategies. AI algorithms excel at analyzing histopathological images, often revealing subtle morphological changes missed by conventional methods. For example, deep learning models applied to digital pathology can effectively differentiate tumor grades and detect rare pathologies, leading to earlier and more precise diagnoses. Progress in neuroimaging is another helpful tool of AI, as enhanced analysis of MRI and CT scans supports early detection of neurodegenerative diseases. By identifying biomarkers and progression patterns, AI aids in timely therapeutic interventions, potentially slowing disease progression. In molecular pathology, AI's ability to analyze complex genomic data helps uncover the genetic and molecular basis of neuropathological conditions, facilitating personalized treatment plans. AI-driven automation streamlines routine diagnostic tasks, allowing pathologists to focus on complex cases, especially in settings with limited resources. This review explores AI's integration into neuropathology, highlighting its current applications, benefits, challenges, and future directions.

[Strategies for prevention and treatment of vascular and nerve injuries in mandibular anterior implant surgery]

Important anatomical structures such as mandibular incisive canal, tongue foramen, and mouth floor vessels may be damaged during implant surgery in the mandibular anterior region, which may lead to mouth floor hematoma, asphyxia, pain, paresthesia and other symptoms. In severe cases, this can be life-threatening. The insufficient alveolar bone space and the anatomical variation of blood vessels and nerves in the mandibular anterior region increase the risk of blood vessel and nerve injury during implant surgery. In case of vascular injury, airway control and hemostasis should be performed, and in case of nerve injury, implant removal and early medical treatment should be performed. To avoid vascular and nerve injury during implant surgery in the mandibular anterior region, it is necessary to be familiar with the anatomical structure, take cone-beam computed tomography, design properly before surgery, and use digital technology during surgery to achieve accurate implant placement. This article summarizes the anatomical structure of the mandibular anterior region, discusses the prevention strategies of vascular and nerve injuries in this region, and discusses the treatment methods after the occurrence of vascular and nerve injuries, to provide clinical reference.

Multimodal deep learning radiomics model for predicting postoperative progression in solid stage I non-small cell lung cancer

PURPOSE

To explore the application value of a multimodal deep learning radiomics (MDLR) model in predicting the risk status of postoperative progression in solid stage I non-small cell lung cancer (NSCLC).

MATERIALS AND METHODS

A total of 459 patients with histologically confirmed solid stage I NSCLC who underwent surgical resection in our institution from January 2014 to September 2019 were reviewed retrospectively. At another medical center, 104 patients were reviewed as an external validation cohort according to the same criteria. A univariate analysis was conducted on the clinicopathological characteristics and subjective CT findings of the progression and non-progression groups. The clinicopathological characteristics and subjective CT findings that exhibited significant differences were used as input variables for the extreme learning machine (ELM) classifier to construct the clinical model. We used the transfer learning strategy to train the ResNet18 model, used the model to extract deep learning features from all CT images, and then used the ELM classifier to classify the deep learning features to obtain the deep learning signature (DLS). A MDLR model incorporating clinicopathological characteristics, subjective CT findings and DLS was constructed. The diagnostic efficiencies of the clinical model, DLS model and MDLR model were evaluated by the area under the curve (AUC).

RESULTS

Univariate analysis indicated that size (p = 0.004), neuron-specific enolase (NSE) (p = 0.03), carbohydrate antigen 19 - 9 (CA199) (p = 0.003), and pathological stage (p = 0.027) were significantly associated with the progression of solid stage I NSCLC after surgery. Therefore, these clinical characteristics were incorporated into the clinical model to predict the risk of progression in postoperative solid-stage NSCLC patients. A total of 294 deep learning features with nonzero coefficients were selected. The DLS in the progressive group was (0.721 ± 0.371), which was higher than that in the nonprogressive group (0.113 ± 0.350) (p < 0.001). The combination of size、NSE、CA199、pathological stage and DLS demonstrated the superior performance in differentiating postoperative progression status. The AUC of the MDLR model was 0.885 (95% confidence interval [CI]: 0.842-0.927), higher than that of the clinical model (0.675 (95% CI: 0.599-0.752)) and DLS model (0.882 (95% CI: 0.835-0.929)). The DeLong test and decision in curve analysis revealed that the MDLR model was the most predictive and clinically useful model.

CONCLUSION

MDLR model is effective in predicting the risk of postoperative progression of solid stage I NSCLC, and it is helpful for the treatment and follow-up of solid stage I NSCLC patients.

Construction and validation of a machine learning-based immune-related prognostic model for glioma

BACKGROUND

Glioma stands as the most prevalent primary brain tumor found within the central nervous system, characterized by high invasiveness and treatment resistance. Although immunotherapy has shown potential in various tumors, it still faces challenges in gliomas. This study seeks to develop and validate a prognostic model for glioma based on immune-related genes, to provide new tools for precision medicine.

METHODS

Glioma samples were obtained from a database that includes the ImmPort database. Additionally, we incorporated ten machine learning algorithms to assess the model's performance using evaluation metrics like the Harrell concordance index (C-index). The model genes were further studied using GSCA, TISCH2, and HPA databases to understand their role in glioma pathology at the genomic, molecular, and single-cell levels, and validate the biological function of IKBKE in vitro experiments.

RESULTS

In this study, a total of 199 genes associated with prognosis were identified using univariate Cox analysis. Subsequently, a consensus prognostic model was developed through the application of machine learning algorithms. In which the Lasso + plsRcox algorithm demonstrated the best predictive performance. The model showed a good ability to distinguish two groups in both the training and test sets. Additionally, the model genes were closely related to immunity (oligodendrocytes and macrophages), and mutation burden. The results of in vitro experiments showed that the expression level of the IKBKE gene had a significant effect on the apoptosis and migration of GL261 glioma cells. Western blot analysis showed that down-regulation of IKBKE resulted in increased expression of pro-apoptotic protein Bax and decreased expression of anti-apoptotic protein Bcl-2, which was consistent with increased apoptosis rate. On the contrary, IKBKE overexpression caused a decrease in Bax expression an increase in Bcl-2 expression, and a decrease in apoptosis rate. Tunel results further confirmed that down-regulation of IKBKE promoted apoptosis, while overexpression of IKBKE reduced apoptosis. In addition, cells with down-regulated IKBKE had reduced migration in scratch experiments, while cells with overexpression of IKBKE had increased migration.

CONCLUSION

This study successfully constructed a glioma prognosis model based on immune-related genes. These findings provide new perspectives for glioma prognosis assessment and immunotherapy.

Multicenter privacy-preserving model training for deep learning brain metastases autosegmentation

OBJECTIVES

This work aims to explore the impact of multicenter data heterogeneity on deep learning brain metastases (BM) autosegmentation performance, and assess the efficacy of an incremental transfer learning technique, namely learning without forgetting (LWF), to improve model generalizability without sharing raw data.

MATERIALS AND METHODS

A total of six BM datasets from University Hospital Erlangen (UKER), University Hospital Zurich (USZ), Stanford, UCSF, New York University (NYU), and BraTS Challenge 2023 were used. First, the performance of the DeepMedic network for BM autosegmentation was established for exclusive single-center training and mixed multicenter training, respectively. Subsequently privacy-preserving bilateral collaboration was evaluated, where a pretrained model is shared to another center for further training using transfer learning (TL) either with or without LWF.

RESULTS

For single-center training, average F1 scores of BM detection range from 0.625 (NYU) to 0.876 (UKER) on respective single-center test data. Mixed multicenter training notably improves F1 scores at Stanford and NYU, with negligible improvement at other centers. When the UKER pretrained model is applied to USZ, LWF achieves a higher average F1 score (0.839) than naive TL (0.570) and single-center training (0.688) on combined UKER and USZ test data. Naive TL improves sensitivity and contouring accuracy, but compromises precision. Conversely, LWF demonstrates commendable sensitivity, precision and contouring accuracy. When applied to Stanford, similar performance was observed.

CONCLUSION

Data heterogeneity (e.g., variations in metastases density, spatial distribution, and image spatial resolution across centers) results in varying performance in BM autosegmentation, posing challenges to model generalizability. LWF is a promising approach to peer-to-peer privacy-preserving model training.

Research on the application of radiomics in breast cancer: A bibliometrics and visualization analysis

Breast cancer is the most prevalent form of cancer worldwide. Therefore, improved disease detection has emerged as a focal point in clinical studies. At the forefront of innovation, radiomics has the capability to extract comprehensive insights from medical images, ultimately enhancing the accuracy of diagnostic procedures. There has been rapid growth in the field of radiomics research on breast cancer in the past few years. We explored pertinent research articles in the Web of Science Core Collection database to gain a thorough understanding of breast cancer radiomics. We used CiteSpace to conduct a bibliometric analysis of the annual distribution of different nations, institutions, journals, authors, keywords, and references in the field of breast cancer radiomics. GraphPad Prism software was used to examine and graph yearly and country-specific trends and the proportions of publications. The tools utilized for the visualization of science mapping included CiteSpace and VOSviewer. Of the 891 publications, most were original articles (731, 91.09%) and a few were reviews (160, 8.91%). Most academic research has been published in China and the United States. The study centers predominantly consisted of major academic institutions, such as Fudan University and the Chinese Academy of Sciences, with some of their members being prominent figures in the field. Pinker, Katja has published the largest number of research papers. The majority of these studies have been published in medical journals focusing on radiology and oncology in recent years. In the realm of cutting-edge medical research, the top two keywords, magnetic resonance imaging and machine learning stand at the forefront as current areas of intense focus. Breast cancer radiomics is advancing rapidly, presenting numerous opportunities and obstacles. Our study of the literature in this academic area aimed to pinpoint the primary themes addressed in the studies and anticipate prospective avenues for research.

Predicting post-surgical functional status in high-grade glioma with resting state fMRI and machine learning

PURPOSE

High-grade glioma (HGG) is the most common and deadly malignant glioma of the central nervous system. The current standard of care includes surgical resection of the tumor, which can lead to functional and cognitive deficits. The aim of this study is to develop models capable of predicting functional outcomes in HGG patients before surgery, facilitating improved disease management and informed patient care.

METHODS

Adult HGG patients (N = 102) from the neurosurgery brain tumor service at Washington University Medical Center were retrospectively recruited. All patients completed structural neuroimaging and resting state functional MRI prior to surgery. Demographics, measures of resting state network connectivity (FC), tumor location, and tumor volume were used to train a random forest classifier to predict functional outcomes based on Karnofsky Performance Status (KPS < 70, KPS ≥ 70).

RESULTS

The models achieved a nested cross-validation accuracy of 94.1% and an AUC of 0.97 in classifying KPS. The strongest predictors identified by the model included FC between somatomotor, visual, auditory, and reward networks. Based on location, the relation of the tumor to dorsal attention, cingulo-opercular, and basal ganglia networks were strong predictors of KPS. Age was also a strong predictor. However, tumor volume was only a moderate predictor.

CONCLUSION

The current work demonstrates the ability of machine learning to classify postoperative functional outcomes in HGG patients prior to surgery accurately. Our results suggest that both FC and the tumor's location in relation to specific networks can serve as reliable predictors of functional outcomes, leading to personalized therapeutic approaches tailored to individual patients.

Machine learning applications in craniosynostosis diagnosis and treatment prediction: a systematic review

Craniosynostosis refers to the premature fusion of one or more of the fibrous cranial sutures connecting the bones of the skull. Machine learning (ML) is an emerging technology and its application to craniosynostosis detection and management is underexplored. This systematic review aims to evaluate the application of ML techniques in the diagnosis, severity assessment, and predictive modeling of craniosynostosis. A comprehensive search was conducted on the PubMed and Google Scholar databases using predefined keywords related to craniosynostosis and ML. Inclusion criteria encompassed peer-reviewed studies in English that investigated ML algorithms in craniosynostosis diagnosis, severity assessment, or treatment outcome prediction. Three independent reviewers screened the search results, performed full-text assessments, and extracted data from selected studies using a standardized form. Thirteen studies met the inclusion criteria and were included in the review. Of the thirteen papers examined on the application of ML to the identification and treatment of craniosynostosis, two papers were dedicated to sagittal craniosynostosis, five papers utilized several different types of craniosynostosis in the training and testing of their ML models, and six papers were dedicated to metopic craniosynostosis. ML models demonstrated high accuracy in identifying different types of craniosynostosis and objectively quantifying severity using innovative metrics such as metopic severity score and cranial morphology deviation. The findings highlight the significant strides made in utilizing ML techniques for craniosynostosis diagnosis, severity assessment, and predictive modeling. Predictive modeling of treatment outcomes following surgical interventions showed promising results, aiding in personalized treatment strategies. Despite methodological diversities among studies, the collective evidence underscores ML's transformative potential in revolutionizing craniosynostosis management.

The Differentiation between Progressive Disease and Treatment-Induced Effects with Perfusion-Weighted Arterial Spin-Labeling in High-Grade Gliomas

BACKGROUND AND PURPOSE

Treatment-induced effects are difficult to differentiate from progressive disease in radiologically progressing diffuse gliomas after treatment. This retrospective, single-center cohort study investigated the diagnostic value of arterial spin-labeling perfusion in differentiating progressive disease from treatment-induced effects in irradiated patients with a high-grade glioma.

MATERIALS AND METHODS

Adults with a high-grade glioma diagnosed between January 1, 2012, and December 31, 2018, with a new or increasing contrast-enhancing lesion after radiotherapy with or without chemotherapy and arterial spin-labeling were consecutively included. Arterial spin-labeling is part of the routine follow-up examinations of patients with a high-grade glioma. The outcomes of progressive disease or treatment-induced effects were defined after histologic or >6 weeks radiologic follow-up. Two neuroradiologists graded the arterial spin-labeling visually as negative (hypointense to gray matter) or positive (iso-/hyperintense). Additionally, the arterial spin-labeling signal intensity in the enhancing lesion was compared quantitatively with that in the contralateral normal brain. Diagnostic test properties and the Cohen κ inter- and intrarater reliability were determined. We present data according to the time after radiation therapy.

RESULTS

We included 141 patients with 173 lesions (median age, 63 years). Ninety-four (54%) lesions showed treatment-induced effects, and 79 (46%), progressive disease. For visual analysis, the ORs of an arterial spin-labeling positive for progressive disease in the group with progression within 3, between 3 and 6, and after 6 months after radiation therapy were 0.65 (95% CI, 0.28-1.51; P = .319), 3.5 (95% CI, 0.69-17.89; P = .132), and 6.8 (95% CI, 1.48-32; P = .014). The areas under the curve were 0.456, 0.652, and 0.719. In quantitative analysis, the areas under the curve were 0.520, 0.588, and 0.587 in these groups. Inter- and intrarater reliability coefficients were 0.67 and 0.62.

CONCLUSIONS

Arterial spin-labeling performed poorly in differentiating progressive disease from treatment-induced effects in high-grade gliomas within 6 months after radiation therapy, with fair performance after this period. Arterial spin-labeling may need to be combined with other imaging features and clinical information for better performance.

Diagnosing Progression in Glioblastoma-Tackling a Neuro-Oncology Problem Using Artificial-Intelligence-Derived Volumetric Change over Time on Magnetic Resonance Imaging to Examine Progression-Free Survival in Glioblastoma

Glioblastoma (GBM) is the most aggressive and the most common primary brain tumor, defined by nearly uniform rapid progression despite the current standard of care involving maximal surgical resection followed by radiation therapy (RT) and temozolomide (TMZ) or concurrent chemoirradiation (CRT), with an overall survival (OS) of less than 30% at 2 years. The diagnosis of tumor progression in the clinic is based on clinical assessment and the interpretation of MRI of the brain using Response Assessment in Neuro-Oncology (RANO) criteria, which suffers from several limitations including a paucity of precise measures of progression. Given that imaging is the primary modality that generates the most quantitative data capable of capturing change over time in the standard of care for GBM, this renders it pivotal in optimizing and advancing response criteria, particularly given the lack of biomarkers in this space. In this study, we employed artificial intelligence (AI)-derived MRI volumetric parameters using the segmentation mask output of the nnU-Net to arrive at four classes (background, edema, non-contrast enhancing tumor (NET), and contrast-enhancing tumor (CET)) to determine if dynamic changes in AI volumes detected throughout therapy can be linked to PFS and clinical features. We identified associations between MR imaging AI-generated volumes and PFS independently of tumor location, MGMT methylation status, and the extent of resection while validating that CET and edema are the most linked to PFS with patient subpopulations separated by district rates of change throughout the disease. The current study provides valuable insights for risk stratification, future RT treatment planning, and treatment monitoring in neuro-oncology.

A Review of Machine Learning Approaches for Brain Positron Emission Tomography Data Analysis

Positron emission tomography (PET) imaging has moved forward the development of medical diagnostics and research across various domains, including cardiology, neurology, infection detection, and oncology. The integration of machine learning (ML) algorithms into PET data analysis has further enhanced their capabilities of including disease diagnosis and classification, image segmentation, and quantitative analysis. ML algorithms empower researchers and clinicians to extract valuable insights from complex big PET datasets, which enabling automated pattern recognition, predictive health outcome modeling, and more efficient data analysis. This review explains the basic knowledge of PET imaging, statistical methods for PET image analysis, and challenges of PET data analysis. We also discussed the improvement of analysis capabilities by combining PET data with machine learning algorithms and the application of this combination in various aspects of PET image research. This review also highlights current trends and future directions in PET imaging, emphasizing the driving and critical role of machine learning and big PET image data analytics in improving diagnostic accuracy and personalized medical approaches. Integration between PET imaging will shape the future of medical diagnosis and research.

Modeling of brain tumors using in vitro, in vivo, and microfluidic models: A review of the current developments

Brain cancers are some of the most complex diseases to treat, despite the numerous advances science has made in cancer chemotherapy and research. One of the key obstacles to identifying potential cures for this disease is the difficulty in emulating the complexity of the brain and the surrounding microenvironment to understand potential therapeutic approaches. This paper discusses some of the most important in vitro, in vivo, and microfluidic brain tumor models that aim to address these challenges.

Global trends and hotspots of ChatGPT in medical research: a bibliometric and visualized study

Objective

With the rapid advancement of Chat Generative Pre-Trained Transformer (ChatGPT) in medical research, our study aimed to identify global trends and focal points in this domain.

Method

All publications on ChatGPT in medical research were retrieved from the Web of Science Core Collection (WoSCC) by Clarivate Analytics from January 1, 2023, to January 31, 2024. The research trends and focal points were visualized and analyzed using VOSviewer and CiteSpace.

Results

A total of 1,239 publications were collected and analyzed. The USA contributed the largest number of publications (458, 37.145%) with the highest total citation frequencies (2,461) and the largest H-index. Harvard University contributed the highest number of publications (33) among all full-time institutions. The Cureus Journal of Medical Science published the most ChatGPT-related research (127, 10.30%). Additionally, Wiwanitkit V contributed the majority of publications in this field (20). "Artificial Intelligence (AI) and Machine Learning (ML)," "Education and Training," "Healthcare Applications," and "Data Analysis and Technology" emerged as the primary clusters of keywords. These areas are predicted to remain hotspots in future research in this field.

Conclusion

Overall, this study signifies the interdisciplinary nature of ChatGPT research in medicine, encompassing AI and ML technologies, education and training initiatives, diverse healthcare applications, and data analysis and technology advancements. These areas are expected to remain at the forefront of future research, driving continued innovation and progress in the field of ChatGPT in medical research.

Artificial intelligence-based MRI radiomics and radiogenomics in glioma

The specific genetic subtypes that gliomas exhibit result in variable clinical courses and the need to involve multidisciplinary teams of neurologists, epileptologists, neurooncologists and neurosurgeons. Currently, the diagnosis of gliomas pivots mainly around the preliminary radiological findings and the subsequent definitive surgical diagnosis (via surgical sampling). Radiomics and radiogenomics present a potential to precisely diagnose and predict survival and treatment responses, via morphological, textural, and functional features derived from MRI data, as well as genomic data. In spite of their advantages, it is still lacking standardized processes of feature extraction and analysis methodology among different research groups, which have made external validations infeasible. Radiomics and radiogenomics can be used to better understand the genomic basis of gliomas, such as tumor spatial heterogeneity, treatment response, molecular classifications and tumor microenvironment immune infiltration. These novel techniques have also been used to predict histological features, grade or even overall survival in gliomas. In this review, workflows of radiomics and radiogenomics are elucidated, with recent research on machine learning or artificial intelligence in glioma.

Development of preoperative and postoperative models to predict recurrence in postoperative glioma patients: a longitudinal cohort study

BACKGROUND

Glioma recurrence, subsequent to maximal safe resection, remains a pivotal challenge. This study aimed to identify key clinical predictors influencing recurrence and develop predictive models to enhance neurological diagnostics and therapeutic strategies.

METHODS

This longitudinal cohort study with a substantial sample size (n = 2825) included patients with non-recurrent glioma who were pathologically diagnosed and had undergone initial surgical resection between 2010 and 2018. Logistic regression models and stratified Cox proportional hazards models were established with the top 15 clinical variables significantly influencing outcomes screened by the least absolute shrinkage and selection operator (LASSO) method. Preoperative and postoperative models predicting short-term (within 6 months) postoperative recurrence in glioma patients were developed to explore the risk factors associated with short- and long-term recurrence in glioma patients.

RESULTS

Preoperative and postoperative logistic models predicting short-term recurrence had accuracies of 0.78 and 0.87, respectively. A range of biological and early symptomatic characteristics linked to short- and long-term recurrence have been pinpointed. Age, headache, muscle weakness, tumor location and Karnofsky score represented significant odd ratios (t > 2.65, p < 0.01) in the preoperative model, while age, WHO grade 4 and chemotherapy or radiotherapy treatments (t > 4.12, p < 0.0001) were most significant in the postoperative period. Postoperative predictive models specifically targeting the glioblastoma and IDH wildtype subgroups were also performed, with an AUC of 0.76 and 0.80, respectively. The 50 combinations of distinct risk factors accommodate diverse recurrence risks among glioma patients, and the nomograms visualizes the results for clinical practice. A stratified Cox model identified many prognostic factors for long-term recurrence, thereby facilitating the enhanced formulation of perioperative care plans for patients, and glioblastoma patients displayed a median progression-free survival (PFS) of only 11 months.

CONCLUSION

The constructed preoperative and postoperative models reliably predicted short-term postoperative glioma recurrence in a substantial patient cohort. The combinations risk factors and nomograms enhance the operability of personalized therapeutic strategies and care regimens. Particular emphasis should be placed on patients with recurrence within six months post-surgery, and the corresponding treatment strategies require comprehensive clinical investigation.

AI-driven estimation of O6 methylguanine-DNA-methyltransferase (MGMT) promoter methylation in glioblastoma patients: a systematic review with bias analysis

BACKGROUND

Accurate and non-invasive estimation of MGMT promoter methylation status in glioblastoma (GBM) patients is of paramount clinical importance, as it is a predictive biomarker associated with improved overall survival (OS). In response to the clinical need, recent studies have focused on the development of non-invasive artificial intelligence (AI)-based methods for MGMT estimation. In this systematic review, we not only delve into the technical aspects of these AI-driven MGMT estimation methods but also emphasize their profound clinical implications. Specifically, we explore the potential impact of accurate non-invasive MGMT estimation on GBM patient care and treatment decisions.

METHODS

Employing a PRISMA search strategy, we identified 33 relevant studies from reputable databases, including PubMed, ScienceDirect, Google Scholar, and IEEE Explore. These studies were comprehensively assessed using 21 diverse attributes, encompassing factors such as types of imaging modalities, machine learning (ML) methods, and cohort sizes, with clear rationales for attribute scoring. Subsequently, we ranked these studies and established a cutoff value to categorize them into low-bias and high-bias groups.

RESULTS

By analyzing the 'cumulative plot of mean score' and the 'frequency plot curve' of the studies, we determined a cutoff value of 6.00. A higher mean score indicated a lower risk of bias, with studies scoring above the cutoff mark categorized as low-bias (73%), while 27% fell into the high-bias category.

CONCLUSION

Our findings underscore the immense potential of AI-based machine learning (ML) and deep learning (DL) methods in non-invasively determining MGMT promoter methylation status. Importantly, the clinical significance of these AI-driven advancements lies in their capacity to transform GBM patient care by providing accurate and timely information for treatment decisions. However, the translation of these technical advancements into clinical practice presents challenges, including the need for large multi-institutional cohorts and the integration of diverse data types. Addressing these challenges will be critical in realizing the full potential of AI in improving the reliability and accessibility of MGMT estimation while lowering the risk of bias in clinical decision-making.

Beyond invasive biopsies: using VASARI MRI features to predict grade and molecular parameters in gliomas

BACKGROUND

Gliomas present a significant economic burden and patient management challenge. The 2021 WHO classification incorporates molecular parameters, which guide treatment decisions. However, acquiring these molecular data involves invasive biopsies, prompting a need for non-invasive diagnostic methods. This study aims to assess the potential of Visually AcceSAble Rembrandt Images (VASARI) MRI features to predict glioma characteristics such as grade, IDH mutation, and MGMT methylation status.

METHODS

This study enrolled 107 glioma patients treated between 2017 and 2022, meeting specific criteria including the absence of prior chemotherapy/radiation therapy, and the presence of molecular and MRI data. Images were assessed using the 27 VASARI MRI features by two blinded radiologists. Pathological and molecular assessments were conducted according to WHO 2021 CNS Tumor classification. Cross-validation Least Absolute Shrinkage and Selection Operator (CV-LASSO) logistic regression was applied for statistical analysis to identify significant VASARI features in determining glioma grade, IDH mutation, and MGMT methylation status.

RESULTS

The study demonstrated substantial observer agreement in VASARI feature evaluation (inter- and intra-observer κ = 0.714 - 0.831 and 0.910, respectively). Patient imaging characteristics varied significantly with glioma grade, IDH mutation, and MGMT methylation. A predictive model was established using VASARI features for glioma grade prediction, exhibiting an AUC of 0.995 (95% CI = 0.986 - 0.998), 100% sensitivity, and 92.86% specificity. IDH mutation status was predicted with AUC 0.930 (95% CI = 0.882 - 0.977), and improved slightly to 0.933 with 'age-at-diagnosis' added. A model predicting MGMT methylation had a satisfactory performance (AUC 0.757, 95% CI = 0.645 - 0.868), improving to 0.791 when 'age-at-diagnosis' was added.

CONCLUSIONS

The T1/FLAIR ratio, enhancement quality, hemorrhage, and proportion enhancing predict glioma grade with excellent accuracy. The proportion enhancing, thickness of enhancing margin, and T1/FLAIR ratio are significant predictors for IDH mutation status. Lastly, MGMT methylation is related to the longest diameter of the lesion, edema crossing the midline, and the proportion of the non-enhancing lesion. VASARI MRI features offer non-invasive and accurate predictive models for glioma grade, IDH mutation, and MGMT methylation status, enhancing glioma patient management.

Comprehensive multimodal deep learning survival prediction enabled by a transformer architecture: A multicenter study in glioblastoma

Background

This research aims to improve glioblastoma survival prediction by integrating MR images, clinical, and molecular-pathologic data in a transformer-based deep learning model, addressing data heterogeneity and performance generalizability.

Methods

We propose and evaluate a transformer-based nonlinear and nonproportional survival prediction model. The model employs self-supervised learning techniques to effectively encode the high-dimensional MRI input for integration with nonimaging data using cross-attention. To demonstrate model generalizability, the model is assessed with the time-dependent concordance index (Cdt) in 2 training setups using 3 independent public test sets: UPenn-GBM, UCSF-PDGM, and Rio Hortega University Hospital (RHUH)-GBM, each comprising 378, 366, and 36 cases, respectively.

Results

The proposed transformer model achieved a promising performance for imaging as well as nonimaging data, effectively integrating both modalities for enhanced performance (UCSF-PDGM test-set, imaging Cdt 0.578, multimodal Cdt 0.672) while outperforming state-of-the-art late-fusion 3D-CNN-based models. Consistent performance was observed across the 3 independent multicenter test sets with Cdt values of 0.707 (UPenn-GBM, internal test set), 0.672 (UCSF-PDGM, first external test set), and 0.618 (RHUH-GBM, second external test set). The model achieved significant discrimination between patients with favorable and unfavorable survival for all 3 datasets (log-rank P 1.9 × 10-8, 9.7 × 10-3, and 1.2 × 10-2). Comparable results were obtained in the second setup using UCSF-PDGM for training/internal testing and UPenn-GBM and RHUH-GBM for external testing (Cdt 0.670, 0.638, and 0.621).

Conclusions

The proposed transformer-based survival prediction model integrates complementary information from diverse input modalities, contributing to improved glioblastoma survival prediction compared to state-of-the-art methods. Consistent performance was observed across institutions supporting model generalizability.

Image-localized biopsy mapping of brain tumor heterogeneity: A single-center study protocol

Brain cancers pose a novel set of difficulties due to the limited accessibility of human brain tumor tissue. For this reason, clinical decision-making relies heavily on MR imaging interpretation, yet the mapping between MRI features and underlying biology remains ambiguous. Standard (clinical) tissue sampling fails to capture the full heterogeneity of the disease. Biopsies are required to obtain a pathological diagnosis and are predominantly taken from the tumor core, which often has different traits to the surrounding invasive tumor that typically leads to recurrent disease. One approach to solving this issue is to characterize the spatial heterogeneity of molecular, genetic, and cellular features of glioma through the intraoperative collection of multiple image-localized biopsy samples paired with multi-parametric MRIs. We have adopted this approach and are currently actively enrolling patients for our 'Image-Based Mapping of Brain Tumors' study. Patients are eligible for this research study (IRB #16-002424) if they are 18 years or older and undergoing surgical intervention for a brain lesion. Once identified, candidate patients receive dynamic susceptibility contrast (DSC) perfusion MRI and diffusion tensor imaging (DTI), in addition to standard sequences (T1, T1Gd, T2, T2-FLAIR) at their presurgical scan. During surgery, sample anatomical locations are tracked using neuronavigation. The collected specimens from this research study are used to capture the intra-tumoral heterogeneity across brain tumors including quantification of genetic aberrations through whole-exome and RNA sequencing as well as other tissue analysis techniques. To date, these data (made available through a public portal) have been used to generate, test, and validate predictive regional maps of the spatial distribution of tumor cell density and/or treatment-related key genetic marker status to identify biopsy and/or treatment targets based on insight from the entire tumor makeup. This type of methodology, when delivered within clinically feasible time frames, has the potential to further inform medical decision-making by improving surgical intervention, radiation, and targeted drug therapy for patients with glioma.

Automated Prediction of Neoadjuvant Chemoradiotherapy Response in Locally Advanced Cervical Cancer Using Hybrid Model-Based MRI Radiomics

BACKGROUND

This study aimed to develop a model that automatically predicts the neoadjuvant chemoradiotherapy (nCRT) response for patients with locally advanced cervical cancer (LACC) based on T2-weighted MR images and clinical parameters.

METHODS

A total of 138 patients were enrolled, and T2-weighted MR images and clinical information of the patients before treatment were collected. Clinical information included age, stage, pathological type, squamous cell carcinoma (SCC) level, and lymph node status. A hybrid model extracted the domain-specific features from the computational radiomics system, the abstract features from the deep learning network, and the clinical parameters. Then, it employed an ensemble learning classifier weighted by logistic regression (LR) classifier, support vector machine (SVM) classifier, K-Nearest Neighbor (KNN) classifier, and Bayesian classifier to predict the pathologic complete response (pCR). The area under the receiver operating characteristics curve (AUC), accuracy (ACC), true positive rate (TPR), true negative rate (TNR), and precision were used as evaluation metrics.

RESULTS

Among the 138 LACC patients, 74 were in the pCR group, and 64 were in the non-pCR group. There was no significant difference between the two cohorts in terms of tumor diameter (p = 0.787), lymph node (p = 0.068), and stage before radiotherapy (p = 0.846), respectively. The 109-dimension domain features and 1472-dimension abstract features from MRI images were used to form a hybrid model. The average AUC, ACC, TPR, TNR, and precision of the proposed hybrid model were about 0.80, 0.71, 0.75, 0.66, and 0.71, while the AUC values of using clinical parameters, domain-specific features, and abstract features alone were 0.61, 0.67 and 0.76, respectively. The AUC value of the model without an ensemble learning classifier was 0.76.

CONCLUSIONS

The proposed hybrid model can predict the radiotherapy response of patients with LACC, which might help radiation oncologists create personalized treatment plans for patients.

A multi-task deep learning model for EGFR genotyping prediction and GTV segmentation of brain metastasis

BACKGROUND

The precise prediction of epidermal growth factor receptor (EGFR) mutation status and gross tumor volume (GTV) segmentation are crucial goals in computer-aided lung adenocarcinoma brain metastasis diagnosis. However, these two tasks present continuous difficulties due to the nonuniform intensity distributions, ambiguous boundaries, and variable shapes of brain metastasis (BM) in MR images.The existing approaches for tackling these challenges mainly rely on single-task algorithms, which overlook the interdependence between these two tasks.

METHODS

To comprehensively address these challenges, we propose a multi-task deep learning model that simultaneously enables GTV segmentation and EGFR subtype classification. Specifically, a multi-scale self-attention encoder that consists of a convolutional self-attention module is designed to extract the shared spatial and global information for a GTV segmentation decoder and an EGFR genotype classifier. Then, a hybrid CNN-Transformer classifier consisting of a convolutional block and a Transformer block is designed to combine the global and local information. Furthermore, the task correlation and heterogeneity issues are solved with a multi-task loss function, aiming to balance the above two tasks by incorporating segmentation and classification loss functions with learnable weights.

RESULTS

The experimental results demonstrate that our proposed model achieves excellent performance, surpassing that of single-task learning approaches. Our proposed model achieves a mean Dice score of 0.89 for GTV segmentation and an EGFR genotyping accuracy of 0.88 on an internal testing set, and attains an accuracy of 0.81 in the EGFR genotype prediction task and an average Dice score of 0.85 in the GTV segmentation task on the external testing set. This shows that our proposed method has outstanding performance and generalization.

CONCLUSION

With the introduction of an efficient feature extraction module, a hybrid CNN-Transformer classifier, and a multi-task loss function, the proposed multi-task deep learning network significantly enhances the performance achieved in both GTV segmentation and EGFR genotyping tasks. Thus, the model can serve as a noninvasive tool for facilitating clinical treatment.

Radiogenomics of adult intracranial gliomas after the 2021 World Health Organisation classification: a review of changes, challenges and opportunities

The classification of diffuse gliomas has undergone substantial changes over the last decade, starting with the 2016 World Health Organisation (WHO) classification, which introduced the importance of molecular markers for glioma diagnosis, in particular, isocitrate dehydrogenase (IDH) status and 1p/19-codeletion. This has spurred research into the correlation of imaging features with the key molecular markers, known as "radiogenomics" or "imaging genomics". Radiogenomics has a variety of possible benefits, including supplementing immunohistochemistry to refine the histological diagnosis and overcoming some of the limitations of the histological assessment. The recent 2021 WHO classification has introduced a variety of changes and continues the trend of increasing the importance of molecular markers in the diagnosis. Key changes include a formal distinction between adult- and paediatric-type diffuse gliomas, the addition of new diagnostic entities, refinements to the nomenclature for IDH-mutant (IDHmut) and IDH-wildtype (IDHwt) gliomas, a shift to grading within tumour types, and the addition of molecular markers as a determinant of tumour grade in addition to phenotype. These changes provide both challenges and opportunities for the field of radiogenomics, which are discussed in this review. This includes implications for the interpretation of research performed prior to the 2021 classification, based on the shift to first classifying gliomas based on genotype ahead of grade, as well as opportunities for future research and priorities for clinical integration.

Treatment of Central Nervous System Tumors on Combination MR-Linear Accelerators: Review of Current Practice and Future Directions

Magnetic resonance imaging (MRI) provides excellent visualization of central nervous system (CNS) tumors due to its superior soft tissue contrast. Magnetic resonance-guided radiotherapy (MRgRT) has historically been limited to use in the initial treatment planning stage due to cost and feasibility. MRI-guided linear accelerators (MRLs) allow clinicians to visualize tumors and organs at risk (OARs) directly before and during treatment, a process known as online MRgRT. This novel system permits adaptive treatment planning based on anatomical changes to ensure accurate dose delivery to the tumor while minimizing unnecessary toxicity to healthy tissue. These advancements are critical to treatment adaptation in the brain and spinal cord, where both preliminary MRI and daily CT guidance have typically had limited benefit. In this narrative review, we investigate the application of online MRgRT in the treatment of various CNS malignancies and any relevant ongoing clinical trials. Imaging of glioblastoma patients has shown significant changes in the gross tumor volume over a standard course of chemoradiotherapy. The use of adaptive online MRgRT in these patients demonstrated reduced target volumes with cavity shrinkage and a resulting reduction in radiation dose to uninvolved tissue. Dosimetric feasibility studies have shown MRL-guided stereotactic radiotherapy (SRT) for intracranial and spine tumors to have potential dosimetric advantages and reduced morbidity compared with conventional linear accelerators. Similarly, dosimetric feasibility studies have shown promise in hippocampal avoidance whole brain radiotherapy (HA-WBRT). Next, we explore the potential of MRL-based multiparametric MRI (mpMRI) and genomically informed radiotherapy to treat CNS disease with cutting-edge precision. Lastly, we explore the challenges of treating CNS malignancies and special limitations MRL systems face.

The importance of planning CT-based imaging features for machine learning-based prediction of pain response

Patients suffering from painful spinal bone metastases (PSBMs) often undergo palliative radiation therapy (RT), with an efficacy of approximately two thirds of patients. In this exploratory investigation, we assessed the effectiveness of machine learning (ML) models trained on radiomics, semantic and clinical features to estimate complete pain response. Gross tumour volumes (GTV) and clinical target volumes (CTV) of 261 PSBMs were segmented on planning computed tomography (CT) scans. Radiomics, semantic and clinical features were collected for all patients. Random forest (RFC) and support vector machine (SVM) classifiers were compared using repeated nested cross-validation. The best radiomics classifier was trained on CTV with an area under the receiver-operator curve (AUROC) of 0.62 ± 0.01 (RFC; 95% confidence interval). The semantic model achieved a comparable AUROC of 0.63 ± 0.01 (RFC), significantly below the clinical model (SVM, AUROC: 0.80 ± 0.01); and slightly lower than the spinal instability neoplastic score (SINS; LR, AUROC: 0.65 ± 0.01). A combined model did not improve performance (AUROC: 0,74 ± 0,01). We could demonstrate that radiomics and semantic analyses of planning CTs allowed for limited prediction of therapy response to palliative RT. ML predictions based on established clinical parameters achieved the best results.

Development and validation of a lung graph-based machine learning model to predict acute pulmonary thromboembolism on chest noncontrast computed tomography

Background

Computed tomography pulmonary angiography (CTPA) is a first-line noninvasive method to diagnose acute pulmonary thromboembolism (APE); however, whether chest noncontrast CT (NC-CT) could aid in the diagnosis of APE remains unknown. The aim of this study was to build and evaluate a holistic lung graph-based machine learning (HLG-ML) using NC-CT for the diagnosis of APE and to compare its performance with that of radiologists and the YEARS algorithm.

Methods

This study enrolled 178 cases (77 males; age 63.9±16.7 years) who underwent NC-CT and CTPA in the same day from January 2019 to December 2020. Of these patients, 133 (75% of cases; 58 males; age 65.4±15.6 years) were placed into a training group and 45 (25% of cases; 19 males; age 59.6±19.2 years) into a testing group. The other 43 cases (18 males; age 62.8±20.0 years) were used to externally validate the model between January 2021 and March 2022. A HLG was developed with a pulmonary radiomics descriptor derived from NC-CT images. The approach extracted local radiomics features and encoded these local features into a radiomics descriptor as a characterization of global radiomics feature distribution. Subsequently, 8 ML models were trained and compared based on the radiomics descriptor. In the validation group, area under the curves (AUCs) of the HLG-ML model in the diagnosis of APE were compared with those of the 3 radiologists and the YEARS algorithm.

Results

Among the 8 ML models, gradient boosting decision tree demonstrated the best classification performance (AUC =0.772) on the training set. In the testing set, the AUC of gradient boosting decision trees was 0.857 [95% confidence intervals (CIs): 0.699-0.951]. In the validation set, the performance of gradient boosting decision tree (AUC =0.810; 95% CI: 0.669-0.952; Youden index =0.621) outperformed 3 radiologists (AUC =0.508, 95% CI: 0.335-0.681, Youden index =0.016; AUC =0.504, 95% CI: 0.354-0.654, Youden index =0.008; AUC =0.527, 95% CI: 0.363-0.691, Youden index =0.050) and the YEARS algorithm (AUC =0.618; 95% CI: 0.469-0.767; Youden index =0.237).

Conclusions

Compared to all 3 radiologists and the YEARS algorithm, the proposed HLG-based gradient boosting decision tree model achieved a superior performance in the diagnosis of APE on the NC-CT and may thus serve as a valuable tool for physicians in the diagnosis of APE.

Radiomics: The New Promise for Differentiating Progression, Recurrence, Pseudoprogression, and Radionecrosis in Glioma and Glioblastoma Multiforme

Glioma and glioblastoma multiform (GBM) remain among the most debilitating and life-threatening brain tumors. Despite advances in diagnosing approaches, patient follow-up after treatment (surgery and chemoradiation) is still challenging for differentiation between tumor progression/recurrence, pseudoprogression, and radionecrosis. Radiomics emerges as a promising tool in initial diagnosis, grading, and survival prediction in patients with glioma and can help differentiate these post-treatment scenarios. Preliminary published studies are promising about the role of radiomics in post-treatment glioma/GBM. However, this field faces significant challenges, including a lack of evidence-based solid data, scattering publication, heterogeneity of studies, and small sample sizes. The present review explores radiomics's capabilities in following patients with glioma/GBM status post-treatment and to differentiate tumor progression, recurrence, pseudoprogression, and radionecrosis.

AI Evaluation of Imaging Factors in the Evolution of Stage-Treated Metastases Using Gamma Knife

BACKGROUND

The study investigated whether three deep-learning models, namely, the CNN_model (trained from scratch), the TL_model (transfer learning), and the FT_model (fine-tuning), could predict the early response of brain metastases (BM) to radiosurgery using a minimal pre-processing of the MRI images. The dataset consisted of 19 BM patients who underwent stereotactic-radiosurgery (SRS) within 3 months. The images used included axial fluid-attenuated inversion recovery (FLAIR) sequences and high-resolution contrast-enhanced T1-weighted (CE T1w) sequences from the tumor center. The patients were classified as responders (complete or partial response) or non-responders (stable or progressive disease).

METHODS

A total of 2320 images from the regression class and 874 from the progression class were randomly assigned to training, testing, and validation groups. The DL models were trained using the training-group images and labels, and the validation dataset was used to select the best model for classifying the evaluation images as showing regression or progression.

RESULTS

Among the 19 patients, 15 were classified as "responders" and 4 as "non-responders". The CNN_model achieved good performance for both classes, showing high precision, recall, and F1-scores. The overall accuracy was 0.98, with an AUC of 0.989. The TL_model performed well in identifying the "progression" class, but could benefit from improved precision, while the "regression" class exhibited high precision, but lower recall. The overall accuracy of the TL_model was 0.92, and the AUC was 0.936. The FT_model showed high recall for "progression", but low precision, and for the "regression" class, it exhibited a high precision, but lower recall. The overall accuracy for the FT_model was 0.83, with an AUC of 0.885.

CONCLUSIONS

Among the three models analyzed, the CNN_model, trained from scratch, provided the most accurate predictions of SRS responses for unlearned BM images. This suggests that CNN models could potentially predict SRS prognoses from small datasets. However, further analysis is needed, especially in cases where class imbalances exist.

The predictive value of a computed tomography-based radiomics model for the surgical separability of thymic epithelial tumors from the superior vena cava and the left innominate vein

Background

The aim of this study was to develop a radiomics machine learning model based on computed tomography (CT) that can predict whether thymic epithelial tumors (TETs) can be separated from veins during surgery and to compare the accuracy of the radiomics model to that of radiologists.

Methods

Patients who underwent thymectomy at our hospital from 2009 to 2017 were included in the screening process. After the selection of patients according to the inclusion and exclusion criteria, the cohort was randomly divided into training and testing groups, and CT images of these patients were collected. Subsequently, two-dimensional (2D) and three-dimensional (3D) regions of interest were labelled using ITK-SNAP 3.8.0 software, and Radiomics features were extracted using Python software (Python Software Foundation) and selected through the least absolute shrinkage and selection operator (LASSO) regression model. To construct the classifier, a support vector machine (SVM) was employed, and a nomogram was created using logistic regression to predict vascular inseparable TETs based on the radiomics score (radscore) and image features. To assess the accuracy of these models, area under receiver operating characteristic (ROC) curves of these models were calculated, and differences among the models were identified using the Delong test.

Results

In this retrospective study, 204 patients with TETs were included, among whom 21 were diagnosed with surgical vascularly inseparable TETs. The area under ROC curve (AUC) of the 2D model, 3D model, 2D + 3D model, and radiologist diagnoses were 0.94, 0.92, 0.95, and 0.87 in the training cohort and 0.95, 0.92, 0.98, and 0.78 in testing cohort, respectively. The Delong test revealed a significant improvement in the performance of the radiomics models compared to radiologists' diagnoses. The logistic regression selected 3 image features, namely maximum diameter of the tumor, degree of abutment of vessel circumference >50%, and absence of the mediastinal fat layer or space between the tumor and surrounding structures. These features, along with the radscore, were included to develop a nomogram. The AUCs of this nomogram were 0.99 in both the training set and testing set, and the Delong test did not find a significant difference between ROC plots of the nomogram and radiomics models.

Conclusions

The proposed radiomics model could accurately predict surgical vascularly inseparable TETs preoperatively and was shown to have a higher predictive value than the radiologists.

Multiparametric MRI radiomics for the differentiation of brain glial cell hyperplasia from low-grade glioma

BACKGROUND

Differentiating between low-grade glioma and brain glial cell hyperplasia is crucial for the customized clinical treatment of patients.

OBJECTIVE

Based on multiparametric MRI imaging and clinical risk factors, a radiomics-clinical model and nomogram were constructed for the distinction of brain glial cell hyperplasia from low-grade glioma.

METHODS

Patients with brain glial cell hyperplasia and low-grade glioma who underwent surgery at the First Affiliated Hospital of Soochow University from March 2016 to March 2022 were retrospectively included. In this study, A total of 41 patients of brain glial cell hyperplasia and 87 patients of low-grade glioma were divided into training group and validation group randomly at a ratio of 7:3. Radiomics features were extracted from T1-weighted imaging (T1WI), T2-weighted imaging (T2WI), diffusion-weighted imaging (DWI), contrast-enhanced T1-weighted imaging (T1-enhanced). Then, LASSO, SVM, and RF models were created in order to choose a model with a greater level of efficiency for calculating each patient's Rad-score (radiomics score). The independent risk factors were identified via univariate and multivariate logistic regression analysis to filter the Rad-score and clinical risk variables in turn. A radiomics-clinical model was next built of which effectiveness was assessed.

RESULTS

Brain glial cell hyperplasia and low-grade gliomas from the 128 cases were randomly divided into 10 groups, of which 7 served as training group and 3 as validation group. The mass effect and Rad-score were two independent risk variables used in the construction of the radiomics-clinical model, and their respective AUCs for the training group and validation group were 0.847 and 0.858. The diagnostic accuracy, sensitivity, and specificity of the validation group were 0.821, 0.750, and 0.852 respectively.

CONCLUSION

Combining with radiomics constructed by multiparametric MRI images and clinical features, the radiomics-clinical model and nomogram that were developed to distinguish between brain glial cell hyperplasia and low-grade glioma had a good performance.

Machine Learning and AI in Cancer Prognosis, Prediction, and Treatment Selection: A Critical Approach

Cancer is a leading cause of morbidity and mortality worldwide. While progress has been made in the diagnosis, prognosis, and treatment of cancer patients, individualized and data-driven care remains a challenge. Artificial intelligence (AI), which is used to predict and automate many cancers, has emerged as a promising option for improving healthcare accuracy and patient outcomes. AI applications in oncology include risk assessment, early diagnosis, patient prognosis estimation, and treatment selection based on deep knowledge. Machine learning (ML), a subset of AI that enables computers to learn from training data, has been highly effective at predicting various types of cancer, including breast, brain, lung, liver, and prostate cancer. In fact, AI and ML have demonstrated greater accuracy in predicting cancer than clinicians. These technologies also have the potential to improve the diagnosis, prognosis, and quality of life of patients with various illnesses, not just cancer. Therefore, it is important to improve current AI and ML technologies and to develop new programs to benefit patients. This article examines the use of AI and ML algorithms in cancer prediction, including their current applications, limitations, and future prospects.

Identifying miRNA biomarkers for breast cancer and ovarian cancer: a text mining perspective

BACKGROUND

microRNA (miRNAs) are small, non-coding RNAs that mediate post-transcriptional gene silencing. Numerous studies have demonstrated the critical role of miRNAs in the development of breast cancer and ovarian cancer. To reduce potential bias from individual studies, a more comprehensive approach of exploring miRNAs in cancer research is essential. This study aims to explore the role of miRNAs in the development of breast cancer and ovarian cancer.

METHODS

Abstracts of the publications were tokenized and the biomedical terms (miRNA, gene, disease, species) were identified and extracted for vectorization. Predictive analyses were conducted with four machine learning models: K-Nearest Neighbors (KNN), Support Vector Machines (SVM), Random Forest (RF), and Naïve Bayes. Both holdout validation and cross-validation were utilized. Feature importance will be identified for miRNA-cancer networks construction.

RESULTS

We found that miR-182 is highly specific to female cancers. miR-182 targets different genes in regulating breast cancer and ovarian cancer. Naïve Bayes provided a promising prediction model for breast cancer and ovarian cancer with miRNAs and genes combination, with an accuracy score greater than 60%. Feature importance identified miR-155 and miR-199 are critical for breast cancer and ovarian cancer prediction, with miR-155 being highly related to breast cancer, whereas miR-199 being more associated with ovarian cancer.

CONCLUSION

Our approach effectively identified potential miRNA biomarkers associated with breast cancer and ovarian cancer, providing a solid foundation for generating novel research hypotheses and guiding future experimental studies.

Assessment of brain cancer atlas maps with multimodal imaging features

BACKGROUND

Glioblastoma Multiforme (GBM) is a fast-growing and highly aggressive brain tumor that invades the nearby brain tissue and presents secondary nodular lesions across the whole brain but generally does not spread to distant organs. Without treatment, GBM can result in death in about 6 months. The challenges are known to depend on multiple factors: brain localization, resistance to conventional therapy, disrupted tumor blood supply inhibiting effective drug delivery, complications from peritumoral edema, intracranial hypertension, seizures, and neurotoxicity.

MAIN TEXT

Imaging techniques are routinely used to obtain accurate detections of lesions that localize brain tumors. Especially magnetic resonance imaging (MRI) delivers multimodal images both before and after the administration of contrast, which results in displaying enhancement and describing physiological features as hemodynamic processes. This review considers one possible extension of the use of radiomics in GBM studies, one that recalibrates the analysis of targeted segmentations to the whole organ scale. After identifying critical areas of research, the focus is on illustrating the potential utility of an integrated approach with multimodal imaging, radiomic data processing and brain atlases as the main components. The templates associated with the outcome of straightforward analyses represent promising inference tools able to spatio-temporally inform on the GBM evolution while being generalizable also to other cancers.

CONCLUSIONS

The focus on novel inference strategies applicable to complex cancer systems and based on building radiomic models from multimodal imaging data can be well supported by machine learning and other computational tools potentially able to translate suitably processed information into more accurate patient stratifications and evaluations of treatment efficacy.

Development of Predictive Models for the Response of Vestibular Schwannoma Treated with Cyberknife®: A Feasibility Study Based on Radiomics and Machine Learning

PURPOSE

to predict vestibular schwannoma (VS) response to radiosurgery by applying machine learning (ML) algorithms on radiomic features extracted from pre-treatment magnetic resonance (MR) images.

METHODS

patients with VS treated with radiosurgery in two Centers from 2004 to 2016 were retrospectively evaluated. Brain T1-weighted contrast-enhanced MR images were acquired before and at 24 and 36 months after treatment. Clinical and treatment data were collected contextually. Treatment responses were assessed considering the VS volume variation based on pre- and post-radiosurgery MR images at both time points. Tumors were semi-automatically segmented and radiomic features were extracted. Four ML algorithms (Random Forest, Support Vector Machine, Neural Network, and extreme Gradient Boosting) were trained and tested for treatment response (i.e., increased or non-increased tumor volume) using nested cross-validation. For training, feature selection was performed using the Least Absolute Shrinkage and Selection Operator, and the selected features were used as input to separately build the four ML classification algorithms. To overcome class imbalance during training, Synthetic Minority Oversampling Technique was used. Finally, trained models were tested on the corresponding held out set of patients to evaluate balanced accuracy, sensitivity, and specificity.

RESULTS

108 patients treated with Cyberknife^® were retrieved; an increased tumor volume was observed at 24 months in 12 patients, and at 36 months in another group of 12 patients. The Neural Network was the best predictive algorithm for response at 24 (balanced accuracy 73% ± 18%, specificity 85% ± 12%, sensitivity 60% ± 42%) and 36 months (balanced accuracy 65% ± 12%, specificity 83% ± 9%, sensitivity 47% ± 27%).

CONCLUSIONS

radiomics may predict VS response to radiosurgery avoiding long-term follow-up as well as unnecessary treatment.

Machine learning in metastatic cancer research: Potentials, possibilities, and prospects

Cancer has received extensive recognition for its high mortality rate, with metastatic cancer being the top cause of cancer-related deaths. Metastatic cancer involves the spread of the primary tumor to other body organs. As much as the early detection of cancer is essential, the timely detection of metastasis, the identification of biomarkers, and treatment choice are valuable for improving the quality of life for metastatic cancer patients. This study reviews the existing studies on classical machine learning (ML) and deep learning (DL) in metastatic cancer research. Since the majority of metastatic cancer research data are collected in the formats of PET/CT and MRI image data, deep learning techniques are heavily involved. However, its black-box nature and expensive computational cost are notable concerns. Furthermore, existing models could be overestimated for their generality due to the non-diverse population in clinical trial datasets. Therefore, research gaps are itemized; follow-up studies should be carried out on metastatic cancer using machine learning and deep learning tools with data in a symmetric manner.

Investigation of radiomics and deep convolutional neural networks approaches for glioma grading

Purpose.To determine glioma grading by applying radiomic analysis or deep convolutional neural networks (DCNN) and to benchmark both approaches on broader validation sets.Methods.Seven public datasets were considered: (1) low-grade glioma or high-grade glioma (369 patients, BraTS'20) (2) well-differentiated liposarcoma or lipoma (115, LIPO); (3) desmoid-type fibromatosis or extremity soft-tissue sarcomas (203, Desmoid); (4) primary solid liver tumors, either malignant or benign (186, LIVER); (5) gastrointestinal stromal tumors (GISTs) or intra-abdominal gastrointestinal tumors radiologically resembling GISTs (246, GIST); (6) colorectal liver metastases (77, CRLM); and (7) lung metastases of metastatic melanoma (103, Melanoma). Radiomic analysis was performed on 464 (2016) radiomic features for the BraTS'20 (others) datasets respectively. Random forests (RF), Extreme Gradient Boosting (XGBOOST) and a voting algorithm comprising both classifiers were tested. The parameters of the classifiers were optimized using a repeated nested stratified cross-validation process. The feature importance of each classifier was computed using the Gini index or permutation feature importance. DCNN was performed on 2D axial and sagittal slices encompassing the tumor. A balanced database was created, when necessary, using smart slices selection. ResNet50, Xception, EficientNetB0, and EfficientNetB3 were transferred from the ImageNet application to the tumor classification and were fine-tuned. Five-fold stratified cross-validation was performed to evaluate the models. The classification performance of the models was measured using multiple indices including area under the receiver operating characteristic curve (AUC).Results.The best radiomic approach was based on XGBOOST for all datasets; AUC was 0.934 (BraTS'20), 0.86 (LIPO), 0.73 (LIVER), (0.844) Desmoid, 0.76 (GIST), 0.664 (CRLM), and 0.577 (Melanoma) respectively. The best DCNN was based on EfficientNetB0; AUC was 0.99 (BraTS'20), 0.982 (LIPO), 0.977 (LIVER), (0.961) Desmoid, 0.926 (GIST), 0.901 (CRLM), and 0.89 (Melanoma) respectively.Conclusion.Tumor classification can be accurately determined by adapting state-of-the-art machine learning algorithms to the medical context.

A Deep Learning-Based Computer Aided Detection (CAD) System for Difficult-to-Detect Brain Metastases

PURPOSE

We sought to develop a computer-aided detection (CAD) system that optimally augments human performance, excelling especially at identifying small inconspicuous brain metastases (BMs), by training a convolutional neural network on a unique magnetic resonance imaging (MRI) data set containing subtle BMs that were not detected prospectively during routine clinical care.

METHODS AND MATERIALS

Patients receiving stereotactic radiosurgery (SRS) for BMs at our institution from 2016 to 2018 without prior brain-directed therapy or small cell histology were eligible. For patients who underwent 2 consecutive courses of SRS, treatment planning MRIs from their initial course were reviewed for radiographic evidence of an emerging metastasis at the same location as metastases treated in their second SRS course. If present, these previously unidentified lesions were contoured and categorized as retrospectively identified metastases (RIMs). RIMs were further subcategorized according to whether they did (+DC) or did not (-DC) meet diagnostic imaging-based criteria to definitively classify them as metastases based upon their appearance in the initial MRI alone. Prospectively identified metastases (PIMs) from these patients, and from patients who only underwent a single course of SRS, were also included. An open-source convolutional neural network architecture was adapted and trained to detect both RIMs and PIMs on thin-slice, contrast-enhanced, spoiled gradient echo MRIs. Patients were randomized into 5 groups: 4 for training/cross-validation and 1 for testing.

RESULTS

One hundred thirty-five patients with 563 metastases, including 72 RIMS, met criteria. For the test group, CAD sensitivity was 94% for PIMs, 80% for +DC RIMs, and 79% for PIMs and +DC RIMs with diameter <3 mm, with a median of 2 false positives per patient and a Dice coefficient of 0.79.

CONCLUSIONS

Our CAD model, trained on a novel data set and using a single common MR sequence, demonstrated high sensitivity and specificity overall, outperforming published CAD results for small metastases and RIMs - the lesion types most in need of human performance augmentation.

Lowering the computational barrier: Partially Bayesian neural networks for transparency in medical imaging AI

Deep Neural Networks (DNNs) can provide clinicians with fast and accurate predictions that are highly valuable for high-stakes medical decision-making, such as in brain tumor segmentation and treatment planning. However, these models largely lack transparency about the uncertainty in their predictions, potentially giving clinicians a false sense of reliability that may lead to grave consequences in patient care. Growing calls for Transparent and Responsible AI have promoted Uncertainty Quantification (UQ) to capture and communicate uncertainty in a systematic and principled manner. However, traditional Bayesian UQ methods remain prohibitively costly for large, million-dimensional tumor segmentation DNNs such as the U-Net. In this work, we discuss a computationally-efficient UQ approach via the partially Bayesian neural networks (pBNN). In pBNN, only a single layer, strategically selected based on gradient-based sensitivity analysis, is targeted for Bayesian inference. We illustrate the effectiveness of pBNN in capturing the full uncertainty for a 7.8-million parameter U-Net. We also demonstrate how practitioners and model developers can use the pBNN's predictions to better understand the model's capabilities and behavior.

Glioma radiogenomics and artificial intelligence: road to precision cancer medicine

Radiogenomics refers to the study of the relationship between imaging phenotypes and gene expression patterns/molecular characteristics, which might allow improved diagnosis, decision-making, and predicting patient outcomes in the context of multiple diseases. Central nervous system (CNS) tumours contribute to significant cancer-related mortality in the present age. Although historically CNS neoplasms were classified and graded based on microscopic appearance, there was discordance between two histologically similar tumours that showed varying prognosis and behaviour, attributable to their molecular signatures. These led to the incorporation of molecular markers in the classification of CNS neoplasms. Meanwhile, advancements in imaging technology such as diffusion-based imaging (including tractography), perfusion, and spectroscopy in addition to the conventional imaging of glial neoplasms, have opened an avenue for radiogenomics. This review touches upon the schema of the current classification of gliomas, concepts behind molecular markers, and parameters that are used in radiogenomics to characterise gliomas and the role of artificial intelligence for the same. Further, the role of radiomics in the grading of brain tumours, prediction of treatment response and prognosis has been discussed. Use of automated and semi-automated tumour segmentation for radiotherapy planning and follow-up has also been discussed briefly.

Radiomics-based evaluation and possible characterization of dynamic contrast enhanced (DCE) perfusion derived different sub-regions of Glioblastoma

BACKGROUND

Glioblastoma (GB) is among the most devastative brain tumors, which usually comprises sub-regions like enhancing tumor (ET), non-enhancing tumor (NET), edema (ED), and necrosis (NEC) as described on MRI. Semi-automated algorithms to extract these tumor subpart volumes and boundaries have been demonstrated using dynamic contrast-enhanced (DCE) perfusion imaging. We aim to characterize these sub-regions derived from DCE perfusion MRI using routine 3D post-contrast-T1 (T1GD) and FLAIR images with the aid of Radiomics analysis. We also explored the possibility of separating edema from tumor sub-regions by extracting the most influential radiomics features.

METHODS

A total of 89 patients with histopathological confirmed IDH wild type GB were considered, who underwent the MR imaging with DCE perfusion-MRI. Perfusion and kinetic indices were computed and further used to segment tumor sub-regions. Radiomics features were extracted from FLAIR and T1GD images with PyRadiomics tool. Statistical analysis of the features was carried out using two approaches as well as machine learning (ML) models were constructed separately, i) within different tumor sub-regions and ii) ED as one category and the remaining sub-regions combined as another category. ML based predictive feature maps was also constructed.

RESULTS

Seven features found to be statistically significant to differentiate tumor sub-regions in FLAIR and T1GD images, with p-value < 0.05 and AUC values in the range of 0.72 to 0.93. However, the edema features stood out in the analysis. In the second approach, the ML model was able to categorize the ED from the rest of the tumor sub-regions in FLAIR and T1GD images with AUC of 0.95 and 0.89 respectively.

CONCLUSION

Radiomics-based specific feature values and maps help to characterize different tumor sub-regions. However, the GLDM_DependenceNonUniformity feature appears to be most specific for separating edema from the remaining tumor sub-regions using conventional FLAIR images. This may be of value in the segmentation of edema from tumors using conventional MRI in the future.

Radiomics in Determining Tumor-to-Normal Brain SUV Ratio Based on 11C-Methionine PET/CT in Glioblastoma

Modern methodology of PET/CT quantitative analysis in patients with glioblastomas is not strictly standardized in clinic settings and does not exclude the influence of the human factor. Methods of radiomics may facilitate unification, and improve objectivity and efficiency of the medical image analysis. The aim of the study is to evaluate the potential of radiomics in the analysis of PET/CT glioblastoma images identifying the relationship between the radiomic features and the 11С-methionine tumor-to-normal brain uptake ratio (TNR) determined by an expert in routine.

MATERIALS AND METHODS

PET/CT data (2018-2020) from 40 patients (average age was 55±12 years; 77.5% were males) with a histologically confirmed diagnosis of "glioblastoma" were included in the analysis. TNR was calculated as a ratio of the standardized uptake value of 11C-methionine measured in the tumor and intact tissue. Calculation of radiomic features for each PET was performed in the specified volumetric region of interest, capturing the tumor with the surrounding tissues. The relationship between TNR and the radiomic features was determined using the linear regression model. Predictors were included in the model following correlation analysis and LASSO regularization. The experiment with machine learning was repeated 300 times, splitting the training (70%) and test (30%) subsets randomly. The model quality metrics and predictor significance obtained in 300 tests were summarized.

RESULTS

Of 412 PET/CT radiomic parameters significantly correlated with TNR (p<0.05), the regularization procedure left no more than 30 in each model (the median number of predictors was 9 [7; 13]). The experiment has demonstrated a non-random linear correlation (the Spearman correlation coefficient was 0.58 [0.43; 0.74]) between TNR and separate radiomic features, primarily fractal dimensions, characterizing the geometrical properties of the image.

CONCLUSION

Radiomics enabled an objective determination of PET/CT image texture features reflecting the biological activity of glioblastomas. Despite the existing limitations in the application, the first results provide a good perspective of these methods in neurooncology.

Development and external validation of an MRI-based neural network for brain metastasis segmentation in the AURORA multicenter study

BACKGROUND

Stereotactic radiotherapy is a standard treatment option for patients with brain metastases. The planning target volume is based on gross tumor volume (GTV) segmentation. The aim of this work is to develop and validate a neural network for automatic GTV segmentation to accelerate clinical daily routine practice and minimize interobserver variability.

METHODS

We analyzed MRIs (T1-weighted sequence ± contrast-enhancement, T2-weighted sequence, and FLAIR sequence) from 348 patients with at least one brain metastasis from different cancer primaries treated in six centers. To generate reference segmentations, all GTVs and the FLAIR hyperintense edematous regions were segmented manually. A 3D-U-Net was trained on a cohort of 260 patients from two centers to segment the GTV and the surrounding FLAIR hyperintense region. During training varying degrees of data augmentation were applied. Model validation was performed using an independent international multicenter test cohort (n = 88) including four centers.

RESULTS

Our proposed U-Net reached a mean overall Dice similarity coefficient (DSC) of 0.92 ± 0.08 and a mean individual metastasis-wise DSC of 0.89 ± 0.11 in the external test cohort for GTV segmentation. Data augmentation improved the segmentation performance significantly. Detection of brain metastases was effective with a mean F1-Score of 0.93 ± 0.16. The model performance was stable independent of the center (p = 0.3). There was no correlation between metastasis volume and DSC (Pearson correlation coefficient 0.07).

CONCLUSION

Reliable automated segmentation of brain metastases with neural networks is possible and may support radiotherapy planning by providing more objective GTV definitions.