A review of original articles published in the emerging field of radiomics

Radiomic Feature Characteristics of Ovine Pulmonary Adenocarcinoma

Radiomic feature (RF) analysis of computed tomography (CT) images may aid the diagnosis and staging of ovine pulmonary adenocarcinoma (OPA). We assessed the RF characteristics of OPA tumours in JSRV-infected sheep compared to non-tumour lung tissues, examined their stability over time, and analysed RF variations in the nascent tumour field (NTF) and nascent tumour margin field (NTmF). In monthly CT scans, lung tissues were automatically segmented by density, and lung tumours were manually segmented. RFs were calculated for each imaging session, selected according to stability and reproducibility, and adjusted for volume dependence where appropriate. Comparisons between scans within sheep were facilitated through fiducial registration and spatial transformations. Initially, 9/36 RFs differed significantly from non-tumour lung tissue of similar density. Predominant RF changes included ngtdm_Complexity, glrlm_RunLNUnif_VN, and gldm_SmDHGLE. RFs in lung tumour segments showed time-dependent changes, whereas non-tumour lung tissue of similar density remained consistent. OPA lung tumour RF characteristics are distinct from those of other lung tissues of similar density and evolve as the tumour develops. Such characteristics suggest that radiomic analysis offers potential for the early detection and management of JSRV-related lung tumours. This research enhances the understanding of OPA imaging, potentially informing better diagnosis and control measures for naturally occurring infections.

Breast cancer detection from ultrasound computed tomography imaging using radiomic analysis: in silico trial

BACKGROUND

Ultrasound computed tomography (USCT) is an imaging modality currently under development for its clinical use in breast imaging. In order to justify clinical trials on imaging prototypes, further research is required to investigate uses and limitations of USCT.

PURPOSE

We investigate the potential of USCT for the detection of breast lesions through the computerized analysis of speed-of-sound (SOS) images of the breast.

METHODS

We conducted an in silico study with a set of 116 virtual breast phantoms (VBPs). We simulated US acquisition and reconstructed 2D SOS slices of the breast via the full waveform inversion (FWI) technique. Subsequently, we conducted breast lesion detection based on computerized texture features (i.e., radiomic features) of the SOS slices. We compare the performance in cancer detection against radiomic analysis of mammograms in terms of the area under the curve (AUC) of the receiver operating characteristic (ROC) curve with 95% confidence intervals estimated using five-fold cross-validation. Statistical analysis involved the Wilcoxon rank-sum test to evaluate significant differences in detection scores, with a significance level ofp < 0.05 $p<0.05$ . AUCs were compared using DeLong's test, and the significance level was adjusted with Bonferroni's correction to account for multiple comparisons.

RESULTS

The AUC for lesion detection from reconstructed SOS images and mammography were 0.87 (95% CI: 0.81-0.94) and 0.77 (95% CI: 0.68-0.86), respectively. Detection of breast lesions using the multimodal approach combining SOS images and mammograms, yielded an AUC of 0.89 (95% CI: 0.83-0.95), with statistically significant differences with respect to the use of mammograms alone (p = 0.0112).

CONCLUSIONS

Our in silico experimental results demonstrate the feasibility of using USCT for breast lesion detection using fully automatic analysis of reconstructed SOS images. The multimodal approach, that combines radio-density and acoustic properties of the breast, outperforms the analysis using a single modality.

Comparison of 2D and 3D radiomics features with conventional features based on contrast-enhanced CT images for preoperative prediction the risk of thymic epithelial tumors

BACKGROUND

This study aimed to develop and validate 2-Dimensional (2D) and 3-Dimensional (3D) radiomics signatures based on contrast-enhanced computed tomography (CECT) images for preoperative prediction of the thymic epithelial tumors (TETs) risk and compare the predictive performance with conventional CT features.

PATIENTS AND METHODS

149 TET patients were retrospectively enrolled from January 2016 to December 2018, and divided into high-risk group (B2/B3/TCs, n = 103) and low-risk group (A/AB/B1, n = 46). All patients were randomly assigned into the training (n = 104) and testing (n = 45) set. 14 conventional CT features were collected, and 396 radiomic features were extracted from 2D and 3D CECT images, respectively. Three models including conventional, 2D radiomics and 3D radiomics model were established using multivariate logistic regression analysis. The discriminative performances of the models were demonstrated by receiver operating characteristic (ROC) curves.

RESULTS

In the conventional model, area under the curves (AUCs) in the training and validation sets were 0.863 and 0.853, sensitivity was 78% and 55%, and specificity was 88% and 100%, respectively. The 2D model yielded AUCs of 0.854 and 0.834, sensitivity of 86% and 77%, and specificity of 72% and 86% in the training and validation sets. The 3D model revealed AUC of 0.902 and 0.906, sensitivity of 75% and 68%, and specificity of 94% and 100% in the training and validation sets.

CONCLUSIONS

Radiomics signatures based on 3D images could distinguish high-risk from low-risk TETs and provide complementary diagnostic information.

Spatial encoding and growth-related change of sheep lung radiomic features

Introduction

Different regions of the small ruminant lung exhibit variable susceptibility to specific lung pathologies. Such susceptibility may be reflected in regional lung radiomic features extracted from computed tomography (CT) images. In this study, we investigated whether region-specific variation in radiomic features exists in ovine lungs and whether these features remain stable over time.

Methods

Thoracic CT image datasets from 30 young adult sheep were subject to an image segmentation protocol directed at partitioning the lung into individual lobar and sub-lobar segments for radiomic feature analysis. After identifying and removing unstable, non-reproducible, and highly correlated features, 22 features remained and were used as input for principal component (PC) analysis.

Results

The significance of segment-related influence on PC scores was determined and visualised. For six sheep, successive CT images were acquired at monthly intervals for a period of 9 months in order to assess time-dependent variation in radiomic features. The results indicated that there was a significant difference in radiomic features derived from different lung segments. Visualisation of PC scores highlighted differences between caudodorsal and cranioventral lung, between lobar and sub-lobar segments, and suggested a bias towards one lung or the other. Significant changes in PC scores occurred over time. With few exceptions, largely similar changes occurred across all segments in this regard.

Discussion

Overall, our results indicate that although sheep lung radiomic features are influenced by the lung segment of origin, their variation over time is largely consistent throughout the lung. Such influence should be borne in mind when interpreting radiomic features and their changes over time.

The application of radiomics in cancer imaging with a focus on lung cancer, renal cell carcinoma, gastrointestinal cancer, and head and neck cancer: A systematic review

PURPOSE

To study the application of radiomics in cancer imaging with a focus on lung cancer, renal cell carcinoma, gastrointestinal cancer, and head and neck cancer.

METHODS

Different electronic databases were considered. Articles published in the last five years were analyzed (January 2019 and December 2023). Papers were selected by two investigators with over 15 years of experience in Radiomics analysis in cancer imaging. The methodological quality of each radiomics study was performed using the Radiomic Quality Score (RQS) by two different readers in consensus and then by a third operator to solve disagreements between the two readers.

RESULTS

19 articles are included in the review. Among the analyzed studies, only one study achieved an RQS of 18 reporting multivariable analyzes with also non-radiomics features and using the validation phase considering two datasets from two distinct institutes and open science and data domain.

CONCLUSION

This informative review has brought attention to the increasingly consolidated potential of Radiomics, although there are still several aspects to be evaluated before the transition to routine clinical practice. There are several challenges to address, including the need for standardization at all stages of the workflow and the potential for cross-site validation using heterogeneous real-world datasets. It will be necessary to establish and promote an imaging data acquisition protocol, conduct multicenter prospective quality control studies, add scanner differences and vendor-dependent characteristics; to collect images of individuals at additional time points, to report calibration statistics.

Robustness of radiomics within photon-counting detector CT: impact of acquisition and reconstruction factors

OBJECTIVES

To assess the impact of acquisition and reconstruction factors on the robustness of radiomics within photon-counting detector CT (PCD-CT).

METHODS

A phantom with twenty-eight texture materials was scanned with different acquisition and reconstruction factors including reposition, scan mode (standard vs high-pitch), tube voltage (120 kVp vs 140 kVp), slice thickness (1.0 mm vs 0.4 mm), radiation dose level (0.5 mGy, 1.0 mGy, 3.0 mGy, 5.0 mGy, vs 10.0 mGy), quantum iterative reconstruction level (0/4, 2/4, vs 4/4), and reconstruction kernel (Qr40, Qr44, vs Qr48). Thirteen sets of virtual monochromatic images at 70-keV were reconstructed. The regions of interest were drawn with rigid registrations. Ninety-three radiomics features were extracted from each material. The reproducibility of radiomics features was evaluated using the intraclass correlation coefficient (ICC) and concordance correlation coefficient (CCC). The variability of radiomics features was assessed by coefficient of variation (CV) and quartile coefficient of dispersion (QCD).

RESULTS

The percentage of features with ICC > 0.90 and CCC > 0.90 were high when repositioned (88.2% and 88.2%) and tube voltage was changed (87.1% and 87.1%), but none of the features with ICC > 0.90 and CCC > 0.90 when high-pitch scan and different slice thickness were used. The percentage of features with CV < 10% and QCD < 10% were high when repositioned (47.3% and 68.8%) and tube voltage was changed (64.2% and 71.0%), but that with CV < 10% and QCD < 10% were low between standard and high-pitch scans (16.1% and 26.9%) and slice thickness (19.4% and 29.0%).

CONCLUSIONS

The PCD-CT radiomics was robust to tube voltage, radiation dose, reconstruction strength level, and kernel, but brittle to high-pitch scan and slice thickness.

KEY POINTS

Question The stability of radiomics features against acquisition and reconstruction factors within PCD-CT should be fully determined before academic research and clinical application. Findings The radiomics features are robust against tube voltage, radiation dose, reconstruction strength level, and kernel within PCD-CT but brittle to high-pitch scan and slice thickness. Clinical relevance The high-pitch scan and slice thickness that influence voxel size should be set with careful attention within PCD-CT, to allow a higher robustness of radiomics features before the implementation of radiomics analysis in clinical routine.

Radiomics for prediction of perineural invasion in colorectal cancer: a systematic review and meta-analysis

BACKGROUND

Perineural invasion (PNI) in colorectal cancer (CRC) is a significant prognostic factor associated with poor outcomes. Radiomics, which involves extracting quantitative features from medical imaging, has emerged as a potential tool for predicting PNI. This systematic review and meta-analysis aimed to evaluate the diagnostic accuracy of radiomics models in predicting PNI in CRC.

METHODS

A comprehensive literature search was conducted across PubMed, Embase, and Web of Science for studies published up to July 28, 2024. Inclusion criteria focused on studies using radiomics models to predict PNI in CRC with sufficient data to construct diagnostic accuracy metrics. The quality of the included studies was assessed using QUADAS-2 and METRICS tools. Pooled estimates of sensitivity, specificity, and area under the curve (AUC) were calculated. Subgroup analyses were performed based on imaging modalities, segmentation methods, and other variables.

RESULTS

Twelve studies comprising 2853 patients were included in the systematic review, with ten studies contributing to the meta-analysis. The pooled sensitivity and specificity for radiomics models in predicting PNI were 0.74 (95% CI: 0.63-0.82) and 0.85 (95% CI: 0.79-0.90), respectively, in the training cohorts. In the validation cohorts, the sensitivity was 0.65 (95% CI: 0.57-0.72), and specificity was 0.85 (95% CI: 0.81-0.89). The AUC was 0.87 (95% CI: 0.63-0.82) for the training cohorts and 0.84 (95% CI: 0.81-0.87) for the validation cohorts, indicating good diagnostic accuracy. The METRICS scores for the included studies ranged from 65.8 to 85.1%, with an overall average score of 67.25%, reflecting good methodological quality. However, significant heterogeneity was observed across studies, particularly in sensitivity and specificity estimates.

CONCLUSION

Radiomics models show promise as a non-invasive tool for predicting PNI in CRC, with moderate to good diagnostic accuracy. However, the current study's limitations, including reliance on retrospective data, geographic concentration in China, and methodological variability, suggest that further research is needed. Future studies should focus on prospective designs, standardization of methodologies, and the integration of advanced machine-learning techniques to improve the clinical applicability and reliability of radiomics models in CRC management.

Improving rupture status prediction for intracranial aneurysms using wall shear stress informatics

BACKGROUND

Wall shear stress (WSS) plays a crucial role in the natural history of intracranial aneurysms (IA). However, spatial variations among WSS have rarely been utilized to correlate with IAs' natural history. This study aims to establish the feasibility of using spatial patterns of WSS data to predict IAs' rupture status (i.e., ruptured versus unruptured).

METHODS

"Patient-specific" computational fluid dynamics (CFD) simulations were performed for 112 IAs; each IA's rupture status was known from medical records. Recall that CFD-simulated hemodynamics data (wall shear stress and its derivatives) are located on unstructured meshes. Hence, we mapped WSS data from an unstructured grid onto a unit disk (i.e., a uniformly sampled polar coordinate system); data in a uniformly sampled polar system is equivalent to image data. Mapped WSS data (onto the unit disk) were readily available for Radiomics analysis to extract spatial patterns of WSS data. We named this innovative technology "WSS-informatics" (i.e., using informatics techniques to analyze WSS data); the usefulness of WSS-informatics was demonstrated during the predictive modeling of IAs' rupture status.

RESULTS

None of the conventional WSS parameters correlated to IAs' rupture status. However, WSS-informatics metrics were discriminative (p-value < 0.05) to IAs' rupture status. Furthermore, predictive models with WSS-informatics features could significantly improve the prediction performance (area under the receiver operating characteristic curve [AUROC]: 0.78 vs. 0.85; p-value < 0.01).

CONCLUSION

The proposed innovations enabled the first study to use spatial patterns of WSS data to improve the predictive modeling of IAs' rupture status.

Radiomics in radiology: What the radiologist needs to know about technical aspects and clinical impact

Radiomics represents the science of extracting and analyzing a multitude of quantitative features from medical imaging, revealing the quantitative potential of radiologic images. This scientific review aims to provide radiologists with a comprehensive understanding of radiomics, emphasizing its principles, applications, challenges, limits, and prospects. The limitations of standardization in current scientific production are analyzed, along with possible solutions proposed by some of the referenced papers. As the continuous evolution of medical imaging is ongoing, radiologists must be aware of new perspectives to play a central role in patient management.

Bibliometric and visual analysis of radiomics for evaluating lymph node status in oncology

Background

Radiomics, which involves the conversion of digital images into high-dimensional data, has been used in oncological studies since 2012. We analyzed the publications that had been conducted on this subject using bibliometric and visual methods to expound the hotpots and future trends regarding radiomics in evaluating lymph node status in oncology.

Methods

Documents published between 2012 and 2023, updated to August 1, 2024, were searched using the Scopus database. VOSviewer, R Package, and Microsoft Excel were used for visualization.

Results

A total of 898 original articles and reviews written in English and be related to radiomics for evaluating lymph node status in oncology, published between 2015 and 2023, were retrieved. A significant increase in the number of publications was observed, with an annual growth rate of 100.77%. The publications predominantly originated from three countries, with China leading in the number of publications and citations. Fudan University was the most contributing affiliation, followed by Sun Yat-sen University and Southern Medical University, all of which were from China. Tian J. from the Chinese Academy of Sciences contributed the most within 5885 authors. In addition, Frontiers in Oncology had the most publications and transcended other journals in recent 4 years. Moreover, the keywords co-occurrence suggested that the interplay of "radiomics" and "lymph node metastasis," as well as "major clinical study" were the predominant topics, furthermore, the focused topics shifted from revealing the diagnosis of cancers to exploring the deep learning-based prediction of lymph node metastasis, suggesting the combination of artificial intelligence research would develop in the future.

Conclusion

The present bibliometric and visual analysis described an approximately continuous trend of increasing publications related to radiomics in evaluating lymph node status in oncology and revealed that it could serve as an efficient tool for personalized diagnosis and treatment guidance in clinical patients, and combined artificial intelligence should be further considered in the future.

Radiomics-driven personalized radiotherapy for primary and recurrent tumors: A general review with a focus on reirradiation

PURPOSE

This review systematically investigates the role of radiomics in radiotherapy, with a particular emphasis on the use of quantitative imaging biomarkers for predicting clinical outcomes, assessing toxicity, and optimizing treatment planning. While the review encompasses various applications of radiomics in radiotherapy, it particularly highlights its potential for guiding reirradiation of recurrent cancers.

METHODS

A systematic review was conducted based on a Medline search with the search engine PubMed using the keywords "radiomics or radiomic" and "radiotherapy or reirradiation". Out of 189 abstracts reviewed, 147 original articles were included in the analysis. These studies were categorized by tumor localization, imaging modality, study objectives, and performance metrics, with a particular emphasis on the inclusion of external validation and adherence to a standardized radiomics pipeline.

RESULTS

The review identified 14 tumor localizations, with the majority of studies focusing on lung (33 studies), head and neck (27 studies), and brain (15 studies) cancers. CT was the most frequently employed imaging modality (77 studies) for radiomics, followed by MRI (46 studies) and PET (13 studies). The overall AUC across all studies, primarily focused on predicting the risk of recurrence (94 studies) or toxicity (41 studies), was 0.80 (SD=0.08). However, only 24 studies (16.3%) included external validation, with a slightly lower AUC compared to those without it. For studies using CT versus MRI or PET, both had a median AUC of 0.79, with IQRs of 0.73-0.86 for CT and 0.76-0.855 for MRI/PET, showing no significant differences in performance. Five studies involving reirradiation reported a median AUC of 0.81 (IQR: 0.73-0.825).

CONCLUSION

Radiomics demonstrates considerable potential in personalizing radiotherapy by improving treatment precision through better outcome prediction and treatment planning. However, its clinical adoption is hindered by the lack of external validation and variability in study designs. Future research should focus on implementing rigorous validation methods and standardizing imaging protocols to enhance the reliability and generalizability of radiomics in clinical radiotherapy, with particular attention to its application in reirradiation.

CLEAR guideline for radiomics: Early insights into current reporting practices endorsed by EuSoMII

PURPOSE

This study aims to evaluate current reporting practices in radiomics research, with a focus on CheckList for EvaluAtion of Radiomics research (CLEAR).

METHODS

We conducted a citation search using Google Scholar to collect original research articles on radiomics citing the CLEAR guideline up to June 17, 2024. We examined the adoption of the guideline, adherence scores per publication, item-wise adherence rates, and self-reporting practices. An expert panel from the European Society of Medical Imaging Informatics Radiomics Auditing Group conducted a detailed item-by-item confirmation analysis of the self-reported CLEAR checklists.

RESULTS

Out of 100 unique citations from 104 records, 48 original research papers on radiomics were included. The overall adoption rate in the literature was 2 %. Among the citing articles, 94 % (45/48) adopted CLEAR for reporting purposes, applying it to both hand-crafted radiomics (89 %) and deep learning (24 %). Self-reported checklists were included in 58 % (26/45) of these papers. Median study-wise adherence score for self-reported data was 91 % (interquartile range = 18 %). Mean confirmed adherence score was 66 % (standard deviation = 14 %). Difference between these scores was statistically significant, (mean = 21 %; standard deviation = 11 %), p < 0.001. Using an arbitrary 50 % adherence cut-off, the number of items with poor adherence increased from 3 to 15 after confirmation analysis, mostly comprised of open science-related items. In addition, several items were frequently misreported.

CONCLUSION

This study revealed significant discrepancies between self-reported and confirmed adherence to the CLEAR guideline in radiomics research, indicating a need for improved reporting accuracy and verification practices.

Transforming Health Care Landscapes: The Lever of Radiology Research and Innovation on Emerging Markets Poised for Aggressive Growth

Advances in radiology are crucial not only to the future of the field but to medicine as a whole. Here, we present three emerging areas of medicine that are poised to change how health care is delivered-hospital at home, artificial intelligence, and precision medicine-and illustrate how advances in radiological tools and technologies are helping to fuel the growth of these markets in the United States and across the globe.

Improved Prediction of Epidermal Growth Factor Receptor Status by Combined Radiomics of Primary Nonsmall-Cell Lung Cancer and Distant Metastasis

OBJECTIVES

This study aimed to investigate radiomics based on primary nonsmall-cell lung cancer (NSCLC) and distant metastases to predict epidermal growth factor receptor (EGFR) mutation status.

METHODS

A total of 290 patients (mean age, 58.21 ± 9.28) diagnosed with brain (BM, n = 150) or spinal bone metastasis (SM, n = 140) from primary NSCLC were enrolled as a primary cohort. An external validation cohort, consisting of 69 patients (mean age, 59.87 ± 7.23; BM, n = 36; SM, n = 33), was enrolled from another center. Thoracic computed tomography-based features were extracted from the primary tumor and peritumoral area and selected using the least absolute shrinkage and selection operator regression to build a radiomic signature (RS-primary). Contrast-enhanced magnetic resonance imaging-based features were calculated and selected from the BM and SM to build RS-BM and RS-SM, respectively. The RS-BM-Com and RS-SM-Com were developed by integrating the most important features from the primary tumor, BM, and SM.

RESULTS

Six computed tomography-based features showed high association with EGFR mutation status: 3 from intratumoral and 3 from peritumoral areas. By combination of features from primary tumor and metastases, the developed RS-BM-Com and RS-SM-Com performed well with areas under curve in the training (RS-BM-Com vs RS-BM, 0.936 vs 0.885, P = 0.177; RS-SM-Com vs RS-SM, 0.929 vs 0.843, P = 0.003), internal validation (RS-BM-Com vs RS-BM, 0.920 vs 0.858, P = 0.492; RS-SM-Com vs RS-SM, 0.896 vs 0.859, P = 0.379), and external validation (RS-BM-Com vs RS-BM, 0.882 vs 0.805, P = 0.263; RS-SM-Com vs RS-SM, 0.865 vs 0.816, P = 0.312) cohorts.

CONCLUSIONS

This study indicates that the accuracy of detecting EGFR mutations significantly enhanced in the presence of metastases in primary NSCLC. The established radiomic signatures from this approach may be useful as new predictors for patients with distant metastases.

A Practical Roadmap to Implementing Deep Learning Segmentation in the Clinical Neuroimaging Research Workflow

BACKGROUND

Thanks to the proliferation of open-source tools, we are seeing an exponential growth of machine-learning applications, and its integration has become more accessible, particularly for segmentation tools in neuroimaging.

METHODS

This article explores a generalized methodology that harnesses these tools and aims/enables to expedite and enhance the reproducibility of clinical research. Herein, critical considerations include hardware, software, neural network training strategies, and data labeling guidelines. More specifically, we advocate an iterative approach to model training and transfer learning, focusing on internal validation and outlier handling early in the labeling process and fine-tuning later.

RESULTS

The iterative refinement process allows experts to intervene and improve model reliability while cutting down on their time spent in manual work. A seamless integration of the final model's predictions into clinical research is proposed to ensure standardized and reproducible results.

CONCLUSIONS

In short, this article provides a comprehensive framework for accelerating research using machine-learning techniques for image segmentation.

Global trends in the application of fluorescence imaging in pancreatic diseases: a bibliometric and knowledge graph analysis

Background

In recent years, with the continuous development of fluorescence imaging technology, research on its application in pancreatic diseases has surged. This area is currently of high research interest and holds the potential to become a non-invasive and effective tool in the diagnosis and treatment of pancreatic diseases. The objective of this study is to explore the hotspots and trends in the field of fluorescence imaging technology applications in pancreatic diseases from 2003 to 2023 through bibliometric and visual analysis.

Methods

This study utilized the Web of Science (core collection) to identify publications related to the application of fluorescence imaging technology in pancreatic diseases from 2003 to 2023. Tools such as CiteSpace (V 6.2.R6), VOSviewer (v1.6.20), and R Studio (Bibliometrix: R-tool version 4.1.4) were employed to analyze various dimensions including publication count, countries, institutions, journals, authors, co-cited references, keywords, burst words, and references.

Results

A comprehensive analysis was conducted on 913 papers published from January 1, 2003, to December 1, 2023, on the application of fluorescence imaging technology in pancreatic diseases. The number of publications in this field has rapidly increased, with the United States being the central hub. The University of California, San Diego emerged as the most active institution. "Biomaterials" was identified as the most influential journal. Authors with the most publications and the highest average citations per article are Hoffman, Robert M. and Luiken, George A., respectively. Keywords such as pancreatic cancer, cancer, expression, indocyanine green, and nanoparticles received widespread attention, with indocyanine green and nanoparticles being current active research hotspots in the field.

Conclusion

This study is the first bibliometric analysis in the field of fluorescence imaging technology applications in pancreatic diseases. Our data will facilitate a better understanding of the developmental trends, identification of research hotspots, and direction in this field. The findings provide practical information for other scholars to grasp key directions and cutting-edge insights.

Non-linear Logistic Regression applied to Radiomics

In the past decade, radiomics has gained huge popularity. This method has a significant potential in cancer detection, diagnostic or prediction of response to a treatment due to its capacity of reflecting tumor-phenotypic characteristics. Despite its qualities and the interest shown in it by the scientific community, radiomics is still not used for clinical care. Reliable Machine Learning (ML) approaches could promote the usage of radiomics by showing relevant results. In this context, we propose to improve the performance of the Logistic Regression (LR) which is the most used ML model in radiomics. We propose to represent the radiomics features with one-hot encoding. Then, we show that the resulting LR is non-linear and almost equivalent to the Naive Bayes classifier (NBC). Since our LR score function remains additive, the contribution of each feature to the decision can be easily measured. We illustrate the performance of the proposed non-linear LR with two radiomics datasets.

Radiomics for the identification of extraprostatic extension with prostate MRI: a systematic review and meta-analysis

OBJECTIVES

Extraprostatic extension (EPE) of prostate cancer (PCa) is predicted using clinical nomograms. Incorporating MRI could represent a leap forward, although poor sensitivity and standardization represent unsolved issues. MRI radiomics has been proposed for EPE prediction. The aim of the study was to systematically review the literature and perform a meta-analysis of MRI-based radiomics approaches for EPE prediction.

MATERIALS AND METHODS

Multiple databases were systematically searched for radiomics studies on EPE detection up to June 2022. Methodological quality was appraised according to Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) tool and radiomics quality score (RQS). The area under the receiver operating characteristic curves (AUC) was pooled to estimate predictive accuracy. A random-effects model estimated overall effect size. Statistical heterogeneity was assessed with I2 value. Publication bias was evaluated with a funnel plot. Subgroup analyses were performed to explore heterogeneity.

RESULTS

Thirteen studies were included, showing limitations in study design and methodological quality (median RQS 10/36), with high statistical heterogeneity. Pooled AUC for EPE identification was 0.80. In subgroup analysis, test-set and cross-validation-based studies had pooled AUC of 0.85 and 0.89 respectively. Pooled AUC was 0.72 for deep learning (DL)-based and 0.82 for handcrafted radiomics studies and 0.79 and 0.83 for studies with multiple and single scanner data, respectively. Finally, models with the best predictive performance obtained using radiomics features showed pooled AUC of 0.82, while those including clinical data of 0.76.

CONCLUSION

MRI radiomics-powered models to identify EPE in PCa showed a promising predictive performance overall. However, methodologically robust, clinically driven research evaluating their diagnostic and therapeutic impact is still needed.

CLINICAL RELEVANCE STATEMENT

Radiomics might improve the management of prostate cancer patients increasing the value of MRI in the assessment of extraprostatic extension. However, it is imperative that forthcoming research prioritizes confirmation studies and a stronger clinical orientation to solidify these advancements.

KEY POINTS

• MRI radiomics deserves attention as a tool to overcome the limitations of MRI in prostate cancer local staging. • Pooled AUC was 0.80 for the 13 included studies, with high heterogeneity (84.7%, p < .001), methodological issues, and poor clinical orientation. • Methodologically robust radiomics research needs to focus on increasing MRI sensitivity and bringing added value to clinical nomograms at patient level.

Imaging Genomics and Multiomics: A Guide for Beginners Starting Radiomics-Based Research

Radiomics uses advanced mathematical analysis of pixel-level information from radiologic images to extract existing information in traditional imaging algorithms. It is intended to find imaging biomarkers related to the genomics of tumors or disease patterns that improve medical care by advanced detection of tumor response patterns in tumors and to assess prognosis. Radiomics expands the paradigm of medical imaging to help with diagnosis, management of diseases and prognostication, leveraging image features by extracting information that can be used as imaging biomarkers to predict prognosis and response to treatment. Radiogenomics is an emerging area in radiomics that investigates the association between imaging characteristics and gene expression profiles. There are an increasing number of research publications using different radiomics approaches without a clear consensus on which method works best. We aim to describe the workflow of radiomics along with a guide of what to expect when starting a radiomics-based research project.

Applying oversampling before cross-validation will lead to high bias in radiomics

Class imbalance is often unavoidable for radiomic data collected from clinical routine. It can create problems during classifier training since the majority class could dominate the minority class. Consequently, resampling methods like oversampling or undersampling are applied to the data to class-balance the data. However, the resampling must not be applied upfront to all data because it would lead to data leakage and, therefore, to erroneous results. This study aims to measure the extent of this bias. Five-fold cross-validation with 30 repeats was performed using a set of 15 radiomic datasets to train predictive models. The training involved two scenarios: first, the models were trained correctly by applying the resampling methods during the cross-validation. Second, the models were trained incorrectly by performing the resampling on all the data before cross-validation. The bias was defined empirically as the difference between the best-performing models in both scenarios in terms of area under the receiver operating characteristic curve (AUC), sensitivity, specificity, balanced accuracy, and the Brier score. In addition, a simulation study was performed on a randomly generated dataset for verification. The results demonstrated that incorrectly applying the oversampling methods to all data resulted in a large positive bias (up to 0.34 in AUC, 0.33 in sensitivity, 0.31 in specificity, and 0.37 in balanced accuracy). The bias depended on the data balance, and approximately an increase of 0.10 in the AUC was observed for each increase in imbalance. The models also showed a bias in calibration measured using the Brier score, which differed by up to -0.18 between the correctly and incorrectly trained models. The undersampling methods were not affected significantly by bias. These results emphasize that any resampling method should be applied correctly only to the training data to avoid data leakage and, subsequently, biased model performance and calibration.

A meta-analysis of MRI radiomics-based diagnosis for BI-RADS 4 breast lesions

OBJECTIVE

The aim of this study is to conduct a systematic evaluation of the diagnostic efficacy of Breast Imaging Reporting and Data System (BI-RADS) 4 benign and malignant breast lesions using magnetic resonance imaging (MRI) radiomics.

METHODS

A systematic search identified relevant studies. Eligible studies were screened, assessed for quality, and analyzed for diagnostic accuracy. Subgroup and sensitivity analyses explored heterogeneity, while publication bias, clinical relevance and threshold effect were evaluated.

RESULTS

This study analyzed a total of 11 studies involving 1,915 lesions in 1,893 patients with BI-RADS 4 classification. The results showed that the combined sensitivity and specificity of MRI radiomics for diagnosing BI-RADS 4 lesions were 0.88 (95% CI 0.83-0.92) and 0.79 (95% CI 0.72-0.84). The positive likelihood ratio (PLR), negative likelihood ratio (NLR), and diagnostic odds ratio (DOR) were 4.2 (95% CI 3.1-5.7), 0.15 (95% CI: 0.10-0.22), and 29.0 (95% CI 15-55). The summary receiver operating characteristic (SROC) analysis yielded an area under the curve (AUC) of 0.90 (95% CI 0.87-0.92), indicating good diagnostic performance. The study found no significant threshold effect or publication bias, and heterogeneity among studies was attributed to various factors like feature selection algorithm, radiomics algorithms, etc. Overall, the results suggest that MRI radiomics has the potential to improve the diagnostic accuracy of BI-RADS 4 lesions and enhance patient outcomes.

CONCLUSION

MRI-based radiomics is highly effective in diagnosing BI-RADS 4 benign and malignant breast lesions, enabling improving patients' medical outcomes and quality of life.

Explanation and Elaboration with Examples for CLEAR (CLEAR-E3): an EuSoMII Radiomics Auditing Group Initiative

Overall quality of radiomics research has been reported as low in literature, which constitutes a major challenge to improve. Consistent, transparent, and accurate reporting is critical, which can be accomplished with systematic use of reporting guidelines. The CheckList for EvaluAtion of Radiomics research (CLEAR) was previously developed to assist authors in reporting their radiomic research and to assist reviewers in their evaluation. To take full advantage of CLEAR, further explanation and elaboration of each item, as well as literature examples, may be useful. The main goal of this work, Explanation and Elaboration with Examples for CLEAR (CLEAR-E3), is to improve CLEAR's usability and dissemination. In this international collaborative effort, members of the European Society of Medical Imaging Informatics-Radiomics Auditing Group searched radiomics literature to identify representative reporting examples for each CLEAR item. At least two examples, demonstrating optimal reporting, were presented for each item. All examples were selected from open-access articles, allowing users to easily consult the corresponding full-text articles. In addition to these, each CLEAR item's explanation was further expanded and elaborated. For easier access, the resulting document is available at https://radiomic.github.io/CLEAR-E3/ . As a complementary effort to CLEAR, we anticipate that this initiative will assist authors in reporting their radiomics research with greater ease and transparency, as well as editors and reviewers in reviewing manuscripts.Relevance statement Along with the original CLEAR checklist, CLEAR-E3 is expected to provide a more in-depth understanding of the CLEAR items, as well as concrete examples for reporting and evaluating radiomic research.Key points• As a complementary effort to CLEAR, this international collaborative effort aims to assist authors in reporting their radiomics research, as well as editors and reviewers in reviewing radiomics manuscripts.• Based on positive examples from the literature selected by the EuSoMII Radiomics Auditing Group, each CLEAR item explanation was further elaborated in CLEAR-E3.• The resulting explanation and elaboration document with examples can be accessed at  https://radiomic.github.io/CLEAR-E3/ .

Training and external validation of pre-treatment FDG PET-CT-based models for outcome prediction in anal squamous cell carcinoma

OBJECTIVES

The incidence of anal squamous cell carcinoma (ASCC) is increasing worldwide, with a significant proportion of patients treated with curative intent having recurrence. The ability to accurately predict progression-free survival (PFS) and overall survival (OS) would allow for development of personalised treatment strategies. The aim of the study was to train and external test radiomic/clinical feature derived time-to-event prediction models.

METHODS

Consecutive patients with ASCC treated with curative intent at two large tertiary referral centres with baseline FDG PET-CT were included. Radiomic feature extraction was performed using LIFEx software on the pre-treatment PET-CT. Two distinct predictive models for PFS and OS were trained and tuned at each of the centres, with the best performing models externally tested on the other centres' patient cohort.

RESULTS

A total of 187 patients were included from centre 1 (mean age 61.6 ± 11.5 years, median follow up 30 months, PFS events = 57/187, OS events = 46/187) and 257 patients were included from centre 2 (mean age 62.6 ± 12.3 years, median follow up 35 months, PFS events = 70/257, OS events = 54/257). The best performing model for PFS and OS was achieved using a Cox regression model based on age and metabolic tumour volume (MTV) with a training c-index of 0.7 and an external testing c-index of 0.7 (standard error = 0.4).

CONCLUSIONS

A combination of patient age and MTV has been demonstrated using external validation to have the potential to predict OS and PFS in ASCC patients.

CLINICAL RELEVANCE STATEMENT

A Cox regression model using patients' age and metabolic tumour volume showed good predictive potential for progression-free survival in external testing. The benefits of a previous radiomics model published by our group could not be confirmed on external testing.

KEY POINTS

• A predictive model based on patient age and metabolic tumour volume showed potential to predict overall survival and progression-free survival and was validated on an external test cohort. • The methodology used to create a predictive model from age and metabolic tumour volume was repeatable using external cohort data. • The predictive ability of positron emission tomography-computed tomography-derived radiomic features diminished when the influence of metabolic tumour volume was accounted for.

Delta Radiomic Features Predict Resection Margin Status and Overall Survival in Neoadjuvant-Treated Pancreatic Cancer Patients

BACKGROUND

Neoadjuvant therapy (NAT) emerged as the standard of care for patients with pancreatic ductal adenocarcinoma (PDAC) who undergo surgery; however, surgery is morbid, and tools to predict resection margin status (RMS) and prognosis in the preoperative setting are needed. Radiomic models, specifically delta radiomic features (DRFs), may provide insight into treatment dynamics to improve preoperative predictions.

METHODS

We retrospectively collected clinical, pathological, and surgical data (patients with resectable, borderline, locally advanced, and metastatic disease), and pre/post-NAT contrast-enhanced computed tomography (CT) scans from PDAC patients at the University of Texas Southwestern Medical Center (UTSW; discovery) and Humanitas Hospital (validation cohort). Gross tumor volume was contoured from CT scans, and 257 radiomics features were extracted. DRFs were calculated by direct subtraction of pre/post-NAT radiomic features. Cox proportional models and binary prediction models, including/excluding clinical variables, were constructed to predict overall survival (OS), disease-free survival (DFS), and RMS.

RESULTS

The discovery and validation cohorts comprised 58 and 31 patients, respectively. Both cohorts had similar clinical characteristics, apart from differences in NAT (FOLFIRINOX vs. gemcitabine/nab-paclitaxel; p < 0.05) and type of surgery resections (pancreatoduodenectomy, distal or total pancreatectomy; p < 0.05). The model that combined clinical variables (pre-NAT carbohydrate antigen (CA) 19-9, the change in CA19-9 after NAT (∆CA19-9), and resectability status) and DRFs outperformed the clinical feature-based models and other radiomics feature-based models in predicting OS (UTSW: 0.73; Humanitas: 0.66), DFS (UTSW: 0.75; Humanitas: 0.64), and RMS (UTSW 0.73; Humanitas: 0.69).

CONCLUSIONS

Our externally validated, predictive/prognostic delta-radiomics models, which incorporate clinical variables, show promise in predicting the risk of predicting RMS in NAT-treated PDAC patients and their OS or DFS.

Automated Assessment of T2-Weighted MRI to Differentiate Malignant and Benign Primary Solid Liver Lesions in Noncirrhotic Livers Using Radiomics

RATIONALE AND OBJECTIVES

Distinguishing malignant from benign liver lesions based on magnetic resonance imaging (MRI) is an important but often challenging task, especially in noncirrhotic livers. We developed and externally validated a radiomics model to quantitatively assess T2-weighted MRI to distinguish the most common malignant and benign primary solid liver lesions in noncirrhotic livers.

MATERIALS AND METHODS

Data sets were retrospectively collected from three tertiary referral centers (A, B, and C) between 2002 and 2018. Patients with malignant (hepatocellular carcinoma and intrahepatic cholangiocarcinoma) and benign (hepatocellular adenoma and focal nodular hyperplasia) lesions were included. A radiomics model based on T2-weighted MRI was developed in data set A using a combination of machine learning approaches. The model was internally evaluated on data set A through cross-validation, externally validated on data sets B and C, and compared to visual scoring of two experienced abdominal radiologists on data set C.

RESULTS

The overall data set included 486 patients (A: 187, B: 98, and C: 201). The radiomics model had a mean area under the curve (AUC) of 0.78 upon internal validation on data set A and a similar AUC in external validation (B: 0.74 and C: 0.76). In data set C, the two radiologists showed moderate agreement (Cohen's κ: 0.61) and achieved AUCs of 0.86 and 0.82.

CONCLUSION

Our T2-weighted MRI radiomics model shows potential for distinguishing malignant from benign primary solid liver lesions. External validation indicated that the model is generalizable despite substantial MRI acquisition protocol differences. Pending further optimization and generalization, this model may aid radiologists in improving the diagnostic workup of patients with liver lesions.

Radiomics for residual tumour detection and prognosis in newly diagnosed glioblastoma based on postoperative [11C] methionine PET and T1c-w MRI

Personalized treatment strategies based on non-invasive biomarkers have potential to improve patient management in patients with newly diagnosed glioblastoma (GBM). The residual tumour burden after surgery in GBM patients is a prognostic imaging biomarker. However, in clinical patient management, its assessment is a manual and time-consuming process that is at risk of inter-rater variability. Furthermore, the prediction of patient outcome prior to radiotherapy may identify patient subgroups that could benefit from escalated radiotherapy doses. Therefore, in this study, we investigate the capabilities of traditional radiomics and 3D convolutional neural networks for automatic detection of the residual tumour status and to prognosticate time-to-recurrence (TTR) and overall survival (OS) in GBM using postoperative [11C] methionine positron emission tomography (MET-PET) and gadolinium-enhanced T1-w magnetic resonance imaging (MRI). On the independent test data, the 3D-DenseNet model based on MET-PET achieved the best performance for residual tumour detection, while the logistic regression model with conventional radiomics features performed best for T1c-w MRI (AUC: MET-PET 0.95, T1c-w MRI 0.78). For the prognosis of TTR and OS, the 3D-DenseNet model based on MET-PET integrated with age and MGMT status achieved the best performance (Concordance-Index: TTR 0.68, OS 0.65). In conclusion, we showed that both deep-learning and conventional radiomics have potential value for supporting image-based assessment and prognosis in GBM. After prospective validation, these models may be considered for treatment personalization.

The effect of data resampling methods in radiomics

Radiomic datasets can be class-imbalanced, for instance, when the prevalence of diseases varies notably, meaning that the number of positive samples is much smaller than that of negative samples. In these cases, the majority class may dominate the model's training and thus negatively affect the model's predictive performance, leading to bias. Therefore, resampling methods are often utilized to class-balance the data. However, several resampling methods exist, and neither their relative predictive performance nor their impact on feature selection has been systematically analyzed. In this study, we aimed to measure the impact of nine resampling methods on radiomic models utilizing a set of fifteen publicly available datasets regarding their predictive performance. Furthermore, we evaluated the agreement and similarity of the set of selected features. Our results show that applying resampling methods did not improve the predictive performance on average. On specific datasets, slight improvements in predictive performance (+ 0.015 in AUC) could be seen. A considerable disagreement on the set of selected features was seen (only 28.7% of features agreed), which strongly impedes feature interpretability. However, selected features are similar when considering their correlation (82.9% of features correlated on average).

The effect of feature normalization methods in radiomics

OBJECTIVES

In radiomics, different feature normalization methods, such as z-Score or Min-Max, are currently utilized, but their specific impact on the model is unclear. We aimed to measure their effect on the predictive performance and the feature selection.

METHODS

We employed fifteen publicly available radiomics datasets to compare seven normalization methods. Using four feature selection and classifier methods, we used cross-validation to measure the area under the curve (AUC) of the resulting models, the agreement of selected features, and the model calibration. In addition, we assessed whether normalization before cross-validation introduces bias.

RESULTS

On average, the difference between the normalization methods was relatively small, with a gain of at most + 0.012 in AUC when comparing the z-Score (mean AUC: 0.707 ± 0.102) to no normalization (mean AUC: 0.719 ± 0.107). However, on some datasets, the difference reached + 0.051. The z-Score performed best, while the tanh transformation showed the worst performance and even decreased the overall predictive performance. While quantile transformation performed, on average, slightly worse than the z-Score, it outperformed all other methods on one out of three datasets. The agreement between the features selected by different normalization methods was only mild, reaching at most 62%. Applying the normalization before cross-validation did not introduce significant bias.

CONCLUSION

The choice of the feature normalization method influenced the predictive performance but depended strongly on the dataset. It strongly impacted the set of selected features.

CRITICAL RELEVANCE STATEMENT

Feature normalization plays a crucial role in the preprocessing and influences the predictive performance and the selected features, complicating feature interpretation.

KEY POINTS

• The impact of feature normalization methods on radiomic models was measured. • Normalization methods performed similarly on average, but differed more strongly on some datasets. • Different methods led to different sets of selected features, impeding feature interpretation. • Model calibration was not largely affected by the normalization method.

CT radiomics to differentiate between Wilms tumor and clear cell sarcoma of the kidney in children

BACKGROUND

To investigate the role of CT radiomics in distinguishing Wilms tumor (WT) from clear cell sarcoma of the kidney (CCSK) in pediatric patients.

METHODS

We retrospectively enrolled 83 cases of WT and 33 cases of CCSK. These cases were randomly stratified into a training set (n = 81) and a test set (n = 35). Several imaging features from the nephrographic phase were analyzed, including the maximum tumor diameter, the ratio of the maximum CT value of the tumor solid portion to the mean CT value of the contralateral renal vein (CTmax/CT renal vein), and the presence of dilated peritumoral cysts. Radiomics features from corticomedullary phase were extracted, selected, and subsequently integrated into a logistic regression model. We evaluated the model's performance using the area under the curve (AUC), 95% confidence interval (CI), and accuracy.

RESULTS

In the training set, there were statistically significant differences in the maximum tumor diameter (P = 0.021) and the presence of dilated peritumoral cysts (P = 0.005) between WT and CCSK, whereas in the test set, no statistically significant differences were observed (P > 0.05). The radiomics model, constructed using four radiomics features, demonstrated strong performance in the training set with an AUC of 0.889 (95% CI: 0.811-0.967) and an accuracy of 0.864. Upon evaluation using fivefold cross-validation in the training set, the AUC remained high at 0.863 (95% CI: 0.774-0.952), with an accuracy of 0.852. In the test set, the radiomics model achieved an AUC of 0.792 (95% CI: 0.616-0.968) and an accuracy of 0.857.

CONCLUSION

CT radiomics proves to be diagnostically valuable for distinguishing between WT and CCSK in pediatric cases.

Observer- and radiomics model-based computed tomography classification of suppurative versus tuberculous lymphadenitis complicated with nodal necrosis of the neck in children

BACKGROUND

Computed tomography (CT) can be used for the early detection of lymphadenitis. Radiomics is able to identify a large amount of hidden information from images. However, few CT-based radiomics studies on cervical lymphadenitis in children have been published.

OBJECTIVE

This study aimed to investigate the role of visual CT analysis and CT radiomics in differentiating cervical suppurative node necrosis from tuberculous node necrosis in pediatric patients.

MATERIALS AND METHODS

A total of 101 patients with cervical suppurative lymphadenitis (n=52) or cervical tuberculous lymphadenitis (n=49) were included. Clinical data and CT images were retrieved for analysis. For visual observation, 11 major CT features were identified for univariate and multivariate analyses. For radiomics analysis, image segmentation, feature value extraction, and dimension reduction, feature selection and the construction of radiomics-based models were performed through the RadCloud platform.

RESULTS

For the visual observation, significant differences were found between the two groups, including the short diameter of the largest necrotic lymph node (P=0.03), sharp border of the node (P=0.02), fusion of nodes (P=0.02), regular silhouette of the necrotic area (P=0.001), multilocular necrotic area (P=0.02), node calcification (P=0.004), and enhancement degree of the nodal nonnecrotic area (P=0.01). No feature was found to be an independent predictor for suppurative or tuberculous lymphadenitis (P>0.05 for all features). Concerning the radiomics analysis, after feature value extraction and dimension reduction, nine related features were selected. The support vector machine classifier achieved high diagnostic performance in distinguishing suppurative from tuberculous lymphadenitis. The area under the curve, accuracy, sensitivity, and specificity of the support vector machine model test set were 0.89 (95% confidence interval: 0.72-1.00), 0.88, 0.78, and 0.90, respectively.

CONCLUSION

Compared to observer-based CT image analyses, radiomics model-based CT image analyses exhibit better performance in the differential diagnosis of cervical suppurative and tuberculous lymphadenitis complicated with nodal necrosis in children.

Multivariable Risk Modelling and Survival Analysis with Machine Learning in SARS-CoV-2 Infection

AIM

To evaluate the performance of a machine learning model based on demographic variables, blood tests, pre-existing comorbidities, and computed tomography(CT)-based radiomic features to predict critical outcome in patients with acute respiratory syndrome coronavirus 2 (SARS-CoV-2).

METHODS

We retrospectively enrolled 694 SARS-CoV-2-positive patients. Clinical and demographic data were extracted from clinical records. Radiomic data were extracted from CT. Patients were randomized to the training (80%, n = 556) or test (20%, n = 138) dataset. The training set was used to define the association between severity of disease and comorbidities, laboratory tests, demographic, and CT-based radiomic variables, and to implement a risk-prediction model. The model was evaluated using the C statistic and Brier scores. The test set was used to assess model prediction performance.

RESULTS

Patients who died (n = 157) were predominantly male (66%) over the age of 50 with median (range) C-reactive protein (CRP) = 5 [1, 37] mg/dL, lactate dehydrogenase (LDH) = 494 [141, 3631] U/I, and D-dimer = 6.006 [168, 152.015] ng/mL. Surviving patients (n = 537) had median (range) CRP = 3 [0, 27] mg/dL, LDH = 484 [78, 3.745] U/I, and D-dimer = 1.133 [96, 55.660] ng/mL. The strongest risk factors were D-dimer, age, and cardiovascular disease. The model implemented using the variables identified using the LASSO Cox regression analysis classified 90% of non-survivors as high-risk individuals in the testing dataset. In this sample, the estimated median survival in the high-risk group was 9 days (95% CI; 9-37), while the low-risk group did not reach the median survival of 50% (p < 0.001).

CONCLUSIONS

A machine learning model based on combined data available on the first days of hospitalization (demographics, CT-radiomics, comorbidities, and blood biomarkers), can identify SARS-CoV-2 patients at risk of serious illness and death.

Multiregional radiomics of brain metastasis can predict response to EGFR-TKI in metastatic NSCLC

OBJECTIVES

To develop radiomics signatures from multiparametric magnetic resonance imaging (MRI) scans to detect epidermal growth factor receptor (EGFR) mutations and predict the response to EGFR-tyrosine kinase inhibitors (EGFR-TKIs) in non-small cell lung cancer (NSCLC) patients with brain metastasis (BM).

METHODS

We included 230 NSCLC patients with BM treated at our hospital between January 2017 and December 2021 and 80 patients treated at another hospital between July 2014 and October 2021 to form the primary and external validation cohorts, respectively. All patients underwent contrast-enhanced T1-weighted (T1C) and T2-weighted (T2W) MRI, and radiomics features were extracted from both the tumor active area (TAA) and peritumoral edema area (POA) for each patient. The least absolute shrinkage and selection operator (LASSO) was used to identify the most predictive features. Radiomics signatures (RSs) were constructed using logistic regression analysis.

RESULTS

For predicting the EGFR mutation status, the created RS-EGFR-TAA and RS-EGFR- POA showed similar performance. By combination of TAA and POA, the multi-region combined RS (RS-EGFR-Com) achieved the highest prediction performance, with AUCs of 0.896, 0.856, and 0.889 in the primary training, internal validation, and external validation cohort, respectively. For predicting response to EGFR-TKI, the multi-region combined RS (RS-TKI-Com) generated the highest AUCs in the primary training (AUC = 0.817), internal validation (AUC = 0.788), and external validation (AUC = 0.808) cohort, respectively.

CONCLUSIONS

Our findings suggested values of multiregional radiomics of BM for predicting EGFR mutations and response to EGFR-TKI.

CLINICAL RELEVANCE STATEMENT

The application of radiomic analysis of multiparametric brain MRI has proven to be a promising tool to stratify which patients can benefit from EGFR-TKI therapy and to facilitate the precise therapeutics of NSCLC patients with brain metastases.

KEY POINTS

• Multiregional radiomics can improve efficacy in predicting therapeutic response to EGFR-TKI therapy in NSCLC patients with brain metastasis. • The tumor active area (TAA) and peritumoral edema area (POA) may hold complementary information related to the therapeutic response to EGFR-TKI. • The developed multi-region combined radiomics signature achieved the best predictive performance and may be considered as a potential tool for predicting response to EGFR-TKI.

Quantitative radiomics approach to assess acute radiation dermatitis in breast cancer patients

PURPOSE

We applied a radiomics approach to skin surface images to objectively assess acute radiation dermatitis in patients undergoing radiotherapy for breast cancer.

METHODS

A prospective cohort study of 20 patients was conducted. Skin surface images in normal, polarized, and ultraviolet (UV) modes were acquired using a skin analysis device before starting radiotherapy ('Before RT'), approximately 7 days after the first treatment ('RT D7'), on 'RT D14', and approximately 10 days after the radiotherapy ended ('After RT D10'). Eighteen types of radiomic feature ratios were calculated based on the values acquired 'Before RT'. We measured skin doses in ipsilateral breasts using optically stimulated luminescent dosimeters on the first day of radiotherapy. Clinical evaluation of acute radiation dermatitis was performed using the Radiation Therapy Oncology Group scoring criteria on 'RT D14' and 'After RT D10'. Several statistical analysis methods were used in this study to test the performance of radiomic features as indicators of radiodermatitis evaluation.

RESULTS

As the skin was damaged by radiation, the energy for normal mode and sum variance for polarized and UV modes decreased significantly for ipsilateral breasts, whereas contralateral breasts exhibited a smaller decrease with statistical significance. The radiomic feature ratios at 'RT D7' had strong correlations to skin doses and those at 'RT D14' and 'after RT D10' with statistical significance.

CONCLUSIONS

The energy for normal mode and sum variance for polarized and UV modes demonstrated the potential to evaluate and predict acute radiation, which assists in its appropriate management.

Reducing the risk of hallucinations with interpretable deep learning models for low-dose CT denoising: comparative performance analysis

Objective.Reducing CT radiation dose is an often proposed measure to enhance patient safety, which, however results in increased image noise, translating into degradation of clinical image quality. Several deep learning methods have been proposed for low-dose CT (LDCT) denoising. The high risks posed by possible hallucinations in clinical images necessitate methods which aid the interpretation of deep learning networks. In this study, we aim to use qualitative reader studies and quantitative radiomics studies to assess the perceived quality, signal preservation and statistical feature preservation of LDCT volumes denoised by deep learning. We aim to compare interpretable deep learning methods with classical deep neural networks in clinical denoising performance.Approach.We conducted an image quality analysis study to assess the image quality of the denoised volumes based on four criteria to assess the perceived image quality. We subsequently conduct a lesion detection/segmentation study to assess the impact of denoising on signal detectability. Finally, a radiomic analysis study was performed to observe the quantitative and statistical similarity of the denoised images to standard dose CT (SDCT) images.Main results.The use of specific deep learning based algorithms generate denoised volumes which are qualitatively inferior to SDCT volumes(p< 0.05). Contrary to previous literature, denoising the volumes did not reduce the accuracy of the segmentation (p> 0.05). The denoised volumes, in most cases, generated radiomics features which were statistically similar to those generated from SDCT volumes (p> 0.05).Significance.Our results show that the denoised volumes have a lower perceived quality than SDCT volumes. Noise and denoising do not significantly affect detectability of the abdominal lesions. Denoised volumes also contain statistically identical features to SDCT volumes.

Radiomic Analysis of Intrahepatic Cholangiocarcinoma: Non-Invasive Prediction of Pathology Data: A Multicenter Study to Develop a Clinical-Radiomic Model

Standard imaging cannot assess the pathology details of intrahepatic cholangiocarcinoma (ICC). We investigated whether CT-based radiomics may improve the prediction of tumor characteristics. All consecutive patients undergoing liver resection for ICC (2009-2019) in six high-volume centers were evaluated for inclusion. On the preoperative CT, we segmented the ICC (Tumor-VOI, i.e., volume-of-interest) and a 5-mm parenchyma rim around the tumor (Margin-VOI). We considered two types of pathology data: tumor grading (G) and microvascular invasion (MVI). The predictive models were internally validated. Overall, 244 patients were analyzed: 82 (34%) had G3 tumors and 139 (57%) had MVI. For G3 prediction, the clinical model had an AUC = 0.69 and an Accuracy = 0.68 at internal cross-validation. The addition of radiomic features extracted from the portal phase of CT improved the model performance (Clinical data+Tumor-VOI: AUC = 0.73/Accuracy = 0.72; +Tumor-/Margin-VOI: AUC = 0.77/Accuracy = 0.77). Also for MVI prediction, the addition of portal phase radiomics improved the model performance (Clinical data: AUC = 0.75/Accuracy = 0.70; +Tumor-VOI: AUC = 0.82/Accuracy = 0.73; +Tumor-/Margin-VOI: AUC = 0.82/Accuracy = 0.75). The permutation tests confirmed that a combined clinical-radiomic model outperforms a purely clinical one (p < 0.05). The addition of the textural features extracted from the arterial phase had no impact. In conclusion, the radiomic features of the tumor and peritumoral tissue extracted from the portal phase of preoperative CT improve the prediction of ICC grading and MVI.

Can radiomics be used to detect hypoxic-ischemic encephalopathy in neonates without magnetic resonance imaging abnormalities?

BACKGROUND

No study has assessed normal magnetic resonance imaging (MRI) findings to predict potential brain injury in neonates with hypoxic-ischemic encephalopathy (HIE).

OBJECTIVE

We aimed to evaluate the efficacy of MRI-based radiomics models of the basal ganglia, thalami and deep medullary veins to differentiate between HIE and the absence of MRI abnormalities in neonates.

MATERIALS AND METHODS

In this study, we included 38  full-term neonates with HIE and normal MRI findings and 89 normal neonates. Radiomics features were extracted from T1-weighted images, T2-weighted images, diffusion-weighted imaging and susceptibility-weighted imaging (SWI). The different models were evaluated using receiver operating characteristic curve analysis. Clinical utility was evaluated using decision curve analysis.

RESULTS

The SWI model exhibited the best performance among the seven single-sequence models. For the training and validation cohorts, the area under the curves (AUCs) of the SWI model were 1.00 and 0.98, respectively. The combined nomogram model incorporating SWI Rad-scores and independent predictors of clinical characteristics was not able to distinguish HIE in patients without MRI abnormalities from the control group (AUC, 1.00). A high degree of fitting and favorable clinical utility was detected using the calibration curve with the Hosmer-Lemeshow test. Decision curve analysis was used for the SWI, clinical and combined nomogram models. The decision curve showed that the SWI and combined nomogram models had better predictive performance than the clinical model.

CONCLUSIONS

HIE can be detected in patients without MRI abnormalities using an MRI-based radiomics model. The SWI model performed better than the other models.

Performance of alternative manual and automated deep learning segmentation techniques for the prediction of benign and malignant lung nodules

Purpose

We aim to evaluate the performance of radiomic biopsy (RB), best-fit bounding box (BB), and a deep-learning-based segmentation method called no-new-U-Net (nnU-Net), compared to the standard full manual (FM) segmentation method for predicting benign and malignant lung nodules using a computed tomography (CT) radiomic machine learning model.

Materials and Methods

A total of 188 CT scans of lung nodules from 2 institutions were used for our study. One radiologist identified and delineated all 188 lung nodules, whereas a second radiologist segmented a subset (n = 20 ) of these nodules. Both radiologists employed FM and RB segmentation methods. BB segmentations were generated computationally from the FM segmentations. The nnU-Net, a deep-learning-based segmentation method, performed automatic nodule detection and segmentation. The time radiologists took to perform segmentations was recorded. Radiomic features were extracted from each segmentation method, and models to predict benign and malignant lung nodules were developed. The Kruskal-Wallis and DeLong tests were used to compare segmentation times and areas under the curve (AUC), respectively.

Results

For the delineation of the FM, RB, and BB segmentations, the two radiologists required a median time (IQR) of 113 (54 to 251.5), 21 (9.25 to 38), and 16 (12 to 64.25) s, respectively (p = 0.04 ). In dataset 1, the mean AUC (95% CI) of the FM, RB, BB, and nnU-Net model were 0.964 (0.96 to 0.968), 0.985 (0.983 to 0.987), 0.961 (0.956 to 0.965), and 0.878 (0.869 to 0.888). In dataset 2, the mean AUC (95% CI) of the FM, RB, BB, and nnU-Net model were 0.717 (0.705 to 0.729), 0.919 (0.913 to 0.924), 0.699 (0.687 to 0.711), and 0.644 (0.632 to 0.657).

Conclusion

Radiomic biopsy-based models outperformed FM and BB models in prediction of benign and malignant lung nodules in two independent datasets while deep-learning segmentation-based models performed similarly to FM and BB. RB could be a more efficient segmentation method, but further validation is needed.

An overview of meta-analyses on radiomics: more evidence is needed to support clinical translation

OBJECTIVE

To conduct an overview of meta-analyses of radiomics studies assessing their study quality and evidence level.

METHODS

A systematical search was updated via peer-reviewed electronic databases, preprint servers, and systematic review protocol registers until 15 November 2022. Systematic reviews with meta-analysis of primary radiomics studies were included. Their reporting transparency, methodological quality, and risk of bias were assessed by PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) 2020 checklist, AMSTAR-2 (A MeaSurement Tool to Assess systematic Reviews, version 2) tool, and ROBIS (Risk Of Bias In Systematic reviews) tool, respectively. The evidence level supporting the radiomics for clinical use was rated.

RESULTS

We identified 44 systematic reviews with meta-analyses on radiomics research. The mean ± standard deviation of PRISMA adherence rate was 65 ± 9%. The AMSTAR-2 tool rated 5 and 39 systematic reviews as low and critically low confidence, respectively. The ROBIS assessment resulted low, unclear and high risk in 5, 11, and 28 systematic reviews, respectively. We reperformed 53 meta-analyses in 38 included systematic reviews. There were 3, 7, and 43 meta-analyses rated as convincing, highly suggestive, and weak levels of evidence, respectively. The convincing level of evidence was rated in (1) T2-FLAIR radiomics for IDH-mutant vs IDH-wide type differentiation in low-grade glioma, (2) CT radiomics for COVID-19 vs other viral pneumonia differentiation, and (3) MRI radiomics for high-grade glioma vs brain metastasis differentiation.

CONCLUSIONS

The systematic reviews on radiomics were with suboptimal quality. A limited number of radiomics approaches were supported by convincing level of evidence.

CLINICAL RELEVANCE STATEMENT

The evidence supporting the clinical application of radiomics are insufficient, calling for researches translating radiomics from an academic tool to a practicable adjunct towards clinical deployment.

Radiomics and dosiomics-based prediction of radiotherapy-induced xerostomia in head and neck cancer patients

BACKGROUND AND AIM

Dose-response modeling for radiotherapy-induced xerostomia in head and neck cancer (HN) patients is a promising frontier for personalized therapy. Feature extraction from diagnostic and therapeutic images (radiomics and dosiomics features) can be used for data-driven response modeling. The aim of this study is to develop xerostomia predictive models based on radiomics-dosiomics features.

METHODS

Data from the cancer imaging archive (TCIA) for 31 HN cancer patients were employed. For all patients, parotid CT radiomics features were extracted, utilizing Lasso regression for feature selection and multivariate modeling. The models were developed by selected features from pretreatment (CT1), mid-treatment (CT2), post-treatment (CT3), and delta features (ΔCT2-1, ΔCT3-1, ΔCT3-2). We also considered dosiomics features extracted from the parotid dose distribution images (Dose model). Thus, combination models of radio-dosiomics (CT + dose & ΔCT + dose) were developed. Moreover, clinical, and dose-volume histogram (DVH) models were built. Nested 10-fold cross-validation was used to assess the predictive classification of patients into those with and without xerostomia, and the area under the receiver operative characteristic curve (AUC) was used to compare the predictive power of the models. The sensitivity and accuracy of models also were obtained.

RESULTS

In total, 59 parotids were assessed, and 13 models were developed. Our results showed three models with AUC of 0.89 as most predictive, namely ΔCT2-1 + Dose (Sensitivity 0.99, Accuracy 0.94 & Specificity 0.86), CT3 model (Sensitivity 0.96, Accuracy 0.94 & Specificity 0.86) and DVH (Sensitivity 0.93, Accuracy 0.89 & Specificity 0.84). These models were followed by Clinical (AUC 0.89, Sensitivity 0.81, Accuracy 0.97 & Specificity 0.89) and CT2 & Dose (AUC 0.86, Sensitivity 0.97, Accuracy 0.87 & Specificity 0.82). The Dose model (developed by dosiomics features only) had AUC, Sensitivity, Specificity, and Accuracy of 0.72, 0.98, 0.33, and 0.79 respectively.

CONCLUSION

Quantitative features extracted from diagnostic imaging during and after radiotherapy alone or in combination with dosiomics markers obtained from dose distribution images can be used for radiotherapy response modeling, opening up prospects for personalization of therapies toward improved therapeutic outcomes.

Radiomics and machine learning for predicting the consistency of benign tumors of the central nervous system: A systematic review

PURPOSE

Predicting the consistency of benign central nervous system (CNS) tumors prior to surgery helps to improve surgical outcomes. This review summarizes and analyzes the literature on using radiomics and/or machine learning (ML) for consistency prediction.

METHOD

The Medical Literature Analysis and Retrieval System Online (MEDLINE) database was screened for studies published in English from January 1st 2000. Data was extracted according to the PRISMA guidelines and quality of the studies was assessed in compliance with the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2).

RESULTS

Eight publications were included focusing on pituitary macroadenomas (n = 5), pituitary adenomas (n = 1), and meningiomas (n = 2) using a retrospective (n = 6), prospective (n = 1), and unknown (n = 1) study design with a total of 763 patients for the consistency prediction. The studies reported an area under the curve (AUC) of 0.71-0.99 for their respective best performing model regarding the consistency prediction. Of all studies, four articles validated their models internally whereas none validated their models externally. Two articles stated making data available on request with the remaining publications lacking information with regard to data availability.

CONCLUSIONS

The research on consistency prediction of CNS tumors is still at an early stage regarding the use of radiomics and different ML techniques. Best-practice procedures regarding radiomics and ML need to be followed more rigorously to facilitate the comparison between publications and, accordingly, the possible implementation into clinical practice in the future.

Is the radiomics-clinical combined model helpful in distinguishing between pancreatic cancer and mass-forming pancreatitis?

PURPOSE

To develop CT-based radiomics models for distinguishing between resectable PDAC and mass-forming pancreatitis (MFP) and to provide a non-invasive tool for cases of equivocal imaging findings with EUS-FNA needed.

METHODS

A total of 201 patients with resectable PDAC and 54 patients with MFP were included. Development cohort: patients without preoperative EUS-FNA (175 PDAC cases, 38 MFP cases); validation cohort: patients with EUS-FNA (26 PDAC cases, 16 MFP cases). Two radiomic signatures (LASSOscore, PCAscore) were developed based on the LASSO model and principal component analysis. LASSOCli and PCACli prediction models were established by combining clinical features with CT radiomic features. ROC analysis and decision curve analysis (DCA) were performed to evaluate the utility of the model versus EUS-FNA in the validation cohort.

RESULTS

In the validation cohort, the radiomic signatures (LASSOscore, PCAscore) were both effective in distinguishing between resectable PDAC and MFP (AUC_LASSO = 0.743, 95% CI: 0.590-0.896; AUC_PCA = 0.788, 95% CI: 0.639-0.938) and improved the diagnostic accuracy of the baseline onlyCli model (AUC_onlyCli = 0.760, 95% CI: 0.614-0.960) after combination with variables including age, CA19-9, and the double-duct sign (AUC_PCACli = 0.880, 95% CI: 0.776-0.983; AUC_LASSOCli = 0.825, 95% CI: 0.694-0.955). The PCACli model showed comparable performance to FNA (AUC_FNA = 0.810, 95% CI: 0.685-0.935). In DCA, the net benefit of the PCACli model was superior to that of EUS-FNA, avoiding biopsies in 70 per 1000 patients at a risk threshold of 35%.

CONCLUSIONS

The PCACli model showed comparable performance with EUS-FNA in discriminating resectable PDAC from MFP.

External validation of an MR-based radiomic model predictive of locoregional control in oropharyngeal cancer

OBJECTIVES

To externally validate a pre-treatment MR-based radiomics model predictive of locoregional control in oropharyngeal squamous cell carcinoma (OPSCC) and to assess the impact of differences between datasets on the predictive performance.

METHODS

Radiomic features, as defined in our previously published radiomics model, were extracted from the primary tumor volumes of 157 OPSCC patients in a different institute. The developed radiomics model was validated using this cohort. Additionally, parameters influencing performance, such as patient subgroups, MRI acquisition, and post-processing steps on prediction performance will be investigated. For this analysis, matched subgroups (based on human papillomavirus (HPV) status of the tumor, T-stage, and tumor subsite) and a subgroup with only patients with 4-mm slice thickness were studied. Also the influence of harmonization techniques (ComBat harmonization, quantile normalization) and the impact of feature stability across observers and centers were studied. Model performances were assessed by area under the curve (AUC), sensitivity, and specificity.

RESULTS

Performance of the published model (AUC/sensitivity/specificity: 0.74/0.75/0.60) drops when applied on the validation cohort (AUC/sensitivity/specificity: 0.64/0.68/0.60). The performance of the full validation cohort improves slightly when the model is validated using a patient group with comparable HPV status of the tumor (AUC/sensitivity/specificity: 0.68/0.74/0.60), using patients acquired with a slice thickness of 4 mm (AUC/sensitivity/specificity: 0.67/0.73/0.57), or when quantile harmonization was performed (AUC/sensitivity/specificity: 0.66/0.69/0.60).

CONCLUSION

The previously published model shows its generalizability and can be applied on data acquired from different vendors and protocols. Harmonization techniques and subgroup definition influence performance of predictive radiomics models.

KEY POINTS

• Radiomics, a noninvasive quantitative image analysis technique, can support the radiologist by enhancing diagnostic accuracy and/or treatment decision-making. • A previously published model shows its generalizability and could be applied on data acquired from different vendors and protocols.

The role of radiomics in tongue cancer: A new tool for prognosis prediction

BACKGROUND

Radiomics represents an emerging field of precision-medicine. Its application in head and neck is still at the beginning.

METHODS

Retrospective study about magnetic resonance imaging (MRI) based radiomics in oral tongue squamous cell carcinoma (OTSCC) surgically treated (2010-2019; 79 patients). All preoperative MRIs include different sequences (T1, T2, DWI, ADC). Tumor volume was manually segmented and exported to radiomic-software, to perform feature extraction. Statistically significant variables were included in multivariable analysis and related to survival endpoints. Predictive models were elaborated (clinical, radiomic, clinical-radiomic models) and compared using C-index.

RESULTS

In almost all clinical-radiomic models radiomic-score maintained statistical significance. In all cases C-index was higher in clinical-radiomic models than in clinical ones. ADC provided the best fit to the models (C-index 0.98, 0.86, 0.84 in loco-regional recurrence, cause-specific mortality, overall survival, respectively).

CONCLUSION

MRI-based radiomics in OTSCC represents a promising noninvasive method of precision medicine, improving prognosis prediction before surgery.

CT-based radiomics to distinguish progressive from stable neuroendocrine liver metastases treated with somatostatin analogues: an explorative study

BACKGROUND

Accurate response evaluation in patients with neuroendocrine liver metastases (NELM) remains a challenge. Radiomics has shown promising results regarding response assessment.

PURPOSE

To differentiate progressive (PD) from stable disease (SD) with radiomics in patients with NELM undergoing somatostatin analogue (SSA) treatment.

MATERIAL AND METHODS

A total of 46 patients with histologically confirmed gastroenteropancreatic neuroendocrine tumors (GEP-NET) with ≥1 NELM and ≥2 computed tomography (CT) scans were included. Response was assessed with Response Evaluation Criteria in Solid Tumors (RECIST1.1). Hepatic target lesions were manually delineated and analyzed with radiomics. Radiomics features were extracted from each NELM on both arterial-phase (AP) and portal-venous-phase (PVP) CT. Multiple instance learning with regularized logistic regression via LASSO penalization (with threefold cross-validation) was used to classify response. Three models were computed: (i) AP model; (ii) PVP model; and (iii) AP + PVP model for a lesion-based and patient-based outcome. Next, clinical features were added to each model.

RESULTS

In total, 19 (40%) patients had PD. Median follow-up was 13 months (range 1-50 months). Radiomics models could not accurately classify response (area under the curve 0.44-0.60). Adding clinical variables to the radiomics models did not significantly improve the performance of any model.

CONCLUSION

Radiomics features were not able to accurately classify response of NELM on surveillance CT scans during SSA treatment.

Systematic review of the radiomics quality score applications: an EuSoMII Radiomics Auditing Group Initiative

OBJECTIVE

The main aim of the present systematic review was a comprehensive overview of the Radiomics Quality Score (RQS)-based systematic reviews to highlight common issues and challenges of radiomics research application and evaluate the relationship between RQS and review features.

METHODS

The literature search was performed on multiple medical literature archives according to PRISMA guidelines for systematic reviews that reported radiomic quality assessment through the RQS. Reported scores were converted to a 0-100% scale. The Mann-Whitney and Kruskal-Wallis tests were used to compare RQS scores and review features.

RESULTS

The literature research yielded 345 articles, from which 44 systematic reviews were finally included in the analysis. Overall, the median of RQS was 21.00% (IQR = 11.50). No significant differences of RQS were observed in subgroup analyses according to targets (oncological/not oncological target, neuroradiology/body imaging focus and one imaging technique/more than one imaging technique, characterization/prognosis/detection/other).

CONCLUSIONS

Our review did not reveal a significant difference of quality of radiomic articles reported in systematic reviews, divided in different subgroups. Furthermore, low overall methodological quality of radiomics research was found independent of specific application domains. While the RQS can serve as a reference tool to improve future study designs, future research should also be aimed at improving its reliability and developing new tools to meet an ever-evolving research space.

KEY POINTS

• Radiomics is a promising high-throughput method that may generate novel imaging biomarkers to improve clinical decision-making process, but it is an inherently complex analysis and often lacks reproducibility and generalizability. • The Radiomics Quality Score serves a necessary role as the de facto reference tool for assessing radiomics studies. • External auditing of radiomics studies, in addition to the standard peer-review process, is valuable to highlight common limitations and provide insights to improve future study designs and practical applicability of the radiomics models.

Prediction of the Molecular Subtype of IDH Mutation Combined with MGMT Promoter Methylation in Gliomas via Radiomics Based on Preoperative MRI

BACKGROUND

The molecular subtype of IDH mut combined with MGMT meth in gliomas suggests a good prognosis and potential benefit from TMZ chemotherapy. The aim of this study was to establish a radiomics model to predict this molecular subtype.

METHOD

The preoperative MR images and genetic data of 498 patients with gliomas were retrospectively collected from our institution and the TCGA/TCIA dataset. A total of 1702 radiomics features were extracted from the tumour region of interest (ROI) of CE-T1 and T2-FLAIR MR images. Least absolute shrinkage and selection operator (LASSO) and logistic regression were used for feature selection and model building. Receiver operating characteristic (ROC) curves and calibration curves were used to evaluate the predictive performance of the model.

RESULTS

Regarding clinical variables, age and tumour grade were significantly different between the two molecular subtypes in the training, test and independent validation cohorts (p < 0.05). The areas under the curve (AUCs) of the radiomics model based on 16 selected features in the SMOTE training cohort, un-SMOTE training cohort, test set and independent TCGA/TCIA validation cohort were 0.936, 0.932, 0.916 and 0.866, respectively, and the corresponding F1-scores were 0.860, 0.797, 0.880 and 0.802. The AUC of the independent validation cohort increased to 0.930 for the combined model when integrating the clinical risk factors and radiomics signature.

CONCLUSIONS

radiomics based on preoperative MRI can effectively predict the molecular subtype of IDH mut combined with MGMT meth.

Determining the reliable feature change in longitudinal radiomics studies: A methodological approach using the reliable change index

PURPOSE

Determination of reliable change of radiomics feature over time is essential and vital in delta-radiomics, but has not yet been rigorously examined. This study attempts to propose a methodological approach using reliable change index (RCI), a statistical metric to determine the reliability of quantitative biomarker changes by accounting for the baseline measurement standard error, in delta-radiomics. The use of RCI was demonstrated with the MRI data acquired from a group of prostate cancer (PCa) patients treated by 1.5 T MRI-guided radiotherapy (MRgRT).

METHODS

Fifty consecutive PCa patients who underwent five-fractionated MRgRT were retrospectively included, and 1023 radiomics features were extracted from the clinical target volume (CTV) and planning target volume (PTV). The two MRI datasets acquired at the first fraction (MRI11 and MRI21) were used to calculate the baseline feature reliability against image acquisition using intraclass correlation coefficient (ICC). The RCI was constructed based on the baseline feature measurement standard deviation, ICC, and feature value differences at two time points between the fifth (MRI51) and the first fraction MRI (MRI11). The reliable change of features was determined in each patient only if the calculated RCI was over 1.96 or smaller than -1.96. The feature changes between MRI51 and MRI11 were correlated to two patient-reported quality-of-life clinical endpoints of urinary domain summary score (UDSS) and bowel domain summary score (BDSS) in 35 patients using the Spearman correlation test. Only the significant correlations between a feature that was reliably changed in ≥7 patients (20%) by RCI and an endpoint were considered as true significant correlations.

RESULTS

The 352 (34.4%) and 386 (37.7%) features among all 1023 features were determined by RCI to be reliably changed in more than five (10%) patients in the CTV and PTV, respectively. Nineteen features were found reliably changed in the CTV and 31 features in the PTV, respectively, in 10 (20%) or more patients. These features were not necessarily associated with significantly different longitudinal feature values (group p-value < 0.05). Most reliably changed features in more than 10 patients had excellent or good baseline test-retest reliability ICC, while none showed poor reliability. The RCI method ruled out the features to be reliably changed when substantial feature measurement bias was presented. After applying the RCI criterion, only four and five true significant correlations were confirmed with UDSS and BDSS in the CTV, respectively, with low true significance correlation rates of 10.8% (4/37) and 17.9% (5/28). No true significant correlations were found in the PTV.

CONCLUSIONS

The RCI method was proposed for delta-radiomics and demonstrated using PCa MRgRT data. The RCI has advantages over some other statistical metrics commonly used in the previous delta-radiomics studies, and is useful to reliably identify the longitudinal radiomics feature change on an individual basis. This proposed RCI method should be helpful for the development of essential feature selection methodology in delta-radiomics.

Musculoskeletal Ultrasound Image-Based Radiomics for the Diagnosis of Achilles Tendinopathy in Skiers

OBJECTIVES

Our study aimed to develop and validate an efficient ultrasound image-based radiomic model for determining the Achilles tendinopathy in skiers.

METHODS

A total of 88 feet of skiers clinically diagnosed with unilateral chronic Achilles tendinopathy and 51 healthy feet were included in our study. According to the time order of enrollment, the data were divided into a training set (n = 89) and a test set (n = 50). The regions of interest (ROIs) were segmented manually, and 833 radiomic features were extracted from red, green, blue color channels and grayscale of ROIs using Pyradiomics, respectively. Three feature selection and three machine learning modeling algorithms were implemented respectively, for determining the optimal radiomics pipeline. Finally, the area under the receiver operating characteristic curve (AUC), consistency analysis, and decision analysis were used to evaluate the diagnostic performance.

RESULTS

By comparing nine radiomics analysis strategies of three color channels and grayscale, the radiomic model under the green channel obtained the best diagnostic performance, using the Random Forest selection and Support Vector Machine modeling, which was selected as the final machine learning model. All the selected radiomic features were significantly associated with the Achilles tendinopathy (P < .05). The radiomic model had a training AUC of 0.98, a test AUC of 0.99, a sensitivity of 0.90, and a specificity of 1, which could bring sufficient clinical net benefits.

CONCLUSIONS

Ultrasound image-based radiomics achieved high diagnostic performance, which could be used as an intelligent auxiliary tool for the diagnosis of Achilles tendinopathy.

[18F]-FDG PET and MRI radiomic signatures to predict the risk and the location of tumor recurrence after re-irradiation in head and neck cancer

PURPOSE

To evaluate whether radiomics from [18F]-FDG PET and/or MRI before re-irradiation (reRT) of recurrent head and neck cancer (HNC) could predict the occurrence and the location "in-field" or "outside" of a second locoregional recurrence (LR).

METHODS

Among the 55 patients re-irradiated at curative intend for HNC from 2012 to 2019, 48 had an MRI and/or PET before the start of the reRT. Thirty-nine radiomic features (RF) were extracted from the re-irradiated GTV (rGTV) using LIFEx software. Student t tests and Spearman correlation coefficient were used to select the RF that best separate patients who recurred from those who did not, and "in-field" from "outside" recurrences. Principal component analysis involving these features only was used to create a prediction model. Leave-one-out cross-validation was performed to evaluate the models.

RESULTS

After a median follow-up of 17 months, 40/55 patients had developed a second LR, including 18 "in-field" and 22 "outside" recurrences. From pre-reRT MRI, a model based on three RF (GLSZM_SZHGLE, GLSZM_LGLZE, and skewness) predicted whether patients would recur with a balanced accuracy (BA) of 83.5%. Another model from pre-reRT MRI based on three other RF (GLSZM_ LZHGE, NGLDM_Busyness, and GLZLM_SZE) predicted whether patients would recur "in-field" or "outside" with a BA of 78.5%. From pre-reRT PET, a model based on four RF (Kurtosis, SUVbwmin, GLCM_Correlation, and GLCM_Contrast) predicted the LR location with a BA of 84.5%.

CONCLUSION

RF characterizing tumor heterogeneity extracted from pre-reRT PET and MRI predicted whether patients would recur, and whether they would recur "in-field" or "outside".