Radiomics in Brain Tumors

From pixels to prognosis: leveraging radiomics and machine learning to predict IDH1 genotype in gliomas

Gliomas are the most common primary tumors of the central nervous system, and advances in genetics and molecular medicine have significantly transformed their classification and treatment. This study aims to predict the IDH1 genotype in gliomas using radiomics and machine learning (ML) methods. Retrospective data from 108 glioma patients were analyzed, including MRI data supported by demographic details such as age, sex, and comorbidities. Tumor segmentation was manually performed using 3D Slicer software, and 112 radiomic features were extracted with the PyRadiomics library. Feature selection using the mRMR algorithm identified 17 significant radiomic features. Various ML algorithms, including KNN, Ensemble, DT, LR, Discriminant and SVM, were applied to predict the IDH1 genotype. The KNN and Ensemble models achieved the highest sensitivity (92-100%) and specificity (100%), emerging as the most successful models. Comparative analyses demonstrated that KNN achieved an accuracy of 92.59%, sensitivity of 92.38%, specificity of 100%, precision of 100%, and an F1-score of 95.02%. Similarly, the Ensemble model achieved an accuracy of 90.74%, sensitivity of 90.65%, specificity of 100%, precision of 100%, and an F1-score of 95.13%. To evaluate their effectiveness, KNN and Ensemble models were compared with commonly used machine learning approaches in glioma classification. LR, a conventional statistical approach, exhibited lower predictive performance with an accuracy of 79.63%, while SVM, a frequently utilized ML model for radiomics-based tumor classification, achieved an accuracy of 85.19%. Our findings are consistent with previous research indicating that radiomics-based ML models achieve high accuracy in IDH1 mutation prediction, with reported performances typically exceeding 80%. These findings suggest that KNN and Ensemble models are more effective in capturing the non-linear radiomic patterns associated with IDH1 status, compared to traditional ML approaches. Our findings indicate that radiomic analyses provide comprehensive genotypic classification by assessing the entire tumor and present a safer, faster, and more patient-friendly alternative to traditional biopsies. This study highlights the potential of radiomics and ML techniques, particularly KNN, Ensemble, and SVM, as powerful tools for predicting the molecular characteristics of gliomas and developing personalized treatment strategies.

Radiomic Consensus Clustering in Glioblastoma and Association with Gene Expression Profiles

BACKGROUND/OBJECTIVES

Glioblastoma (GBM) is the most common malignant primary central nervous system tumor with extremely poor prognosis and survival outcomes. Non-invasive methods like radiomic feature extraction, which assess sub-visual imaging features, provide a potentially powerful tool for distinguishing molecular profiles across groups of patients with GBM. Using consensus clustering of MRI-based radiomic features, this study aims to investigate differential gene expression profiles based on radiomic clusters.

METHODS

Patients from the TCGA and CPTAC datasets (n = 114) were included in this study. Radiomic features including T1, T1 with contrast, T2, and FLAIR MRI sequences were extracted using PyRadiomics. Selected radiomic features were then clustered using ConsensusClusterPlus (k-means base algorithm and Euclidean distance), which iteratively subsamples and clusters 80% of the data to identify stable clusters by calculating the frequency in which each patient is a member of a cluster across iterations. Gene expression data (available for n = 69 patients) was analyzed using differential gene expression (DEG) and gene set enrichment (GSEA) approaches, after batch correction using ComBat-seq.

RESULTS

Three distinct clusters were identified based on the relative consensus matrix and cumulative distribution plots (Cluster 1, n = 25; Cluster 2, n = 46; Cluster 3, n = 43). No significant differences in patient demographic characteristics, MGMT methylation status, tumor location, or overall survival were identified across clusters. Differentially expressed genes were identified in Cluster 1, which have been previously associated with GBM prognosis, recurrence, and treatment sensitivity. GSEA of Cluster 1 showed an enrichment of genes upregulated for immune-related and DNA metabolism pathways and genes downregulated in pathways associated with protein and histone deacetylation. Clusters 2 and 3 exhibited fewer DEGs which failed to reach significance after multiple testing corrections.

CONCLUSIONS

Consensus clustering of radiomic features revealed unique gene expression profiles in the GBM cohort which likely represent subtle differences in tumor biology and radiosensitivity that are not visually discernible, underscoring the potential of radiomics to serve as a non-invasive alternative for identifying GBM molecular heterogeneity. Further investigation is still required to validate these findings and their clinical implications.

A Radiomic Approach for Evaluating Intra-Subgroup Heterogeneity in SHH and Group 4 Pediatric Medulloblastoma: A Preliminary Multi-Institutional Study

Medulloblastoma (MB) is the most frequent malignant brain tumor in children with extensive heterogeneity that results in varied clinical outcomes. Recently, MB was categorized into four molecular subgroups, WNT, SHH, Group 3, and Group 4. While SHH and Group 4 are known for their intermediate prognosis, studies have reported wide disparities in patient outcomes within these subgroups. This study aims to create a radiomic prognostic signature, medulloblastoma radiomics risk (mRRisk), to identify the risk levels within the SHH and Group 4 subgroups, individually, for reliable risk stratification. Our hypothesis is that this signature can comprehensively capture tumor characteristics that enable the accurate identification of the risk level. In total, 70 MB studies (48 Group 4, and 22 SHH) were retrospectively curated from three institutions. For each subgroup, 232 hand-crafted features that capture the entropy, surface changes, and contour characteristics of the tumor were extracted. Features were concatenated and fed into regression models for risk stratification. Contrasted with Chang stratification that did not yield any significant differences within subgroups, significant differences were observed between two risk groups in Group 4 (p = 0.04, Concordance Index (CI) = 0.82) on the cystic core and non-enhancing tumor, and SHH (p = 0.03, CI = 0.74) on the enhancing tumor. Our results indicate that radiomics may serve as a prognostic tool for refining MB risk stratification, towards improved patient care.

Opportunities and Advances in Radiomics and Radiogenomics for Pediatric Medulloblastoma Tumors

Recent advances in artificial intelligence have greatly impacted the field of medical imaging and vastly improved the development of computational algorithms for data analysis. In the field of pediatric neuro-oncology, radiomics, the process of obtaining high-dimensional data from radiographic images, has been recently utilized in applications including survival prognostication, molecular classification, and tumor type classification. Similarly, radiogenomics, or the integration of radiomic and genomic data, has allowed for building comprehensive computational models to better understand disease etiology. While there exist excellent review articles on radiomics and radiogenomic pipelines and their applications in adult solid tumors, in this review article, we specifically review these computational approaches in the context of pediatric medulloblastoma tumors. Based on our systematic literature research via PubMed and Google Scholar, we provide a detailed summary of a total of 15 articles that have utilized radiomic and radiogenomic analysis for survival prognostication, tumor segmentation, and molecular subgroup classification in the context of pediatric medulloblastoma. Lastly, we shed light on the current challenges with the existing approaches as well as future directions and opportunities with using these computational radiomic and radiogenomic approaches for pediatric medulloblastoma tumors.

A Radiomics Model Based on Synthetic MRI Acquisition for Predicting Neoadjuvant Systemic Treatment Response in Triple-Negative Breast Cancer

Purpose To determine if a radiomics model based on quantitative maps acquired with synthetic MRI (SyMRI) is useful for predicting neoadjuvant systemic therapy (NAST) response in triple-negative breast cancer (TNBC). Materials and Methods In this prospective study, 181 women diagnosed with stage I-III TNBC were scanned with a SyMRI sequence at baseline and at midtreatment (after four cycles of NAST), producing T1, T2, and proton density (PD) maps. Histopathologic analysis at surgery was used to determine pathologic complete response (pCR) or non-pCR status. From three-dimensional tumor contours drawn on the three maps, 310 histogram and textural features were extracted, resulting in 930 features per scan. Radiomic features were compared between pCR and non-pCR groups by using Wilcoxon rank sum test. To build a multivariable predictive model, logistic regression with elastic net regularization and cross-validation was performed for texture feature selection using 119 participants (median age, 52 years [range, 26-77 years]). An independent testing cohort of 62 participants (median age, 48 years [range, 23-74 years]) was used to evaluate and compare the models by area under the receiver operating characteristic curve (AUC). Results Univariable analysis identified 15 T1, 10 T2, and 12 PD radiomic features at midtreatment that predicted pCR with an AUC greater than 0.70 in both the training and testing cohorts. Multivariable radiomics models of maps acquired at midtreatment demonstrated superior performance over those acquired at baseline, achieving AUCs as high as 0.78 and 0.72 in the training and testing cohorts, respectively. Conclusion SyMRI-based radiomic features acquired at midtreatment are potentially useful for identifying early NAST responders in TNBC. Keywords: MR Imaging, Breast, Outcomes Analysis ClinicalTrials.gov registration no. NCT02276443 Supplemental material is available for this article. © RSNA, 2023 See also the commentary by Houser and Rapelyea in this issue.

Differentiation of Lung Metastases Originated From Different Primary Tumors Using Radiomics Features Based on CT Imaging

RATIONALE AND OBJECTIVES

To explore the feasibility of differentiating three predominant metastatic tumor types using lung computed tomography (CT) radiomics features based on supervised machine learning.

MATERIALS AND METHODS

This retrospective analysis included 252 lung metastases (LM) (from 78 patients), which were divided into the training (n = 176) and test (n = 76) cohort randomly. The metastases originated from colorectal cancer (n = 97), breast cancer (n = 87), and renal carcinoma (n = 68). An additional 77 LM (from 35 patients) were used for external validation. All radiomics features were extracted from lung CT using an open-source software called 3D slicer. The least absolute shrinkage and selection operator (LASSO) method selected the optimal radiomics features to build the model. Random forest and support vector machine (SVM) were selected to build three-class and two-class models. The performance of the classification model was evaluated with the area under the receiver operating characteristic curve (AUC) by two strategies: one-versus-rest and one-versus-one.

RESULTS

Eight hundred and fifty-one quantitative radiomics features were extracted from lung CT. By LASSO, 23 optimal features were extracted in three-class, and 25, 29, and 35 features in two-class for differentiating every two of three LM (colorectal cancer vs. renal carcinoma, colorectal cancer vs. breast cancer, and breast cancer vs. renal carcinoma, respectively). The AUCs of the three-class model were 0.83 for colorectal cancer, 0.79 for breast cancer, and 0.91 for renal carcinoma in the test cohort. In the external validation cohort, the AUCs were 0.77, 0.83, and 0.81, respectively. Swarmplot shows the distribution of radiomics features among three different LM types. In the two-class model, high accuracy and AUC were obtained by SVM. The AUC of discriminating colorectal cancer LM from renal carcinoma LM was 0.84, and breast cancer LM from colorectal cancer LM and renal carcinoma LM were 0.80 and 0.94, respectively. The AUCs were 0.77, 0.78, and 0.84 in the external validation cohort.

CONCLUSION

Quantitative radiomics features based on Lung CT exhibited good discriminative performance in LM of primary colorectal cancer, breast cancer, and renal carcinoma.

Prediction of single pulmonary nodule growth by CT radiomics and clinical features - a one-year follow-up study

Background

With the development of imaging technology, an increasing number of pulmonary nodules have been found. Some pulmonary nodules may gradually grow and develop into lung cancer, while others may remain stable for many years. Accurately predicting the growth of pulmonary nodules in advance is of great clinical significance for early treatment. The purpose of this study was to establish a predictive model using radiomics and to study its value in predicting the growth of pulmonary nodules.

Materials and methods

According to the inclusion and exclusion criteria, 228 pulmonary nodules in 228 subjects were included in the study. During the one-year follow-up, 69 nodules grew larger, and 159 nodules remained stable. All the nodules were randomly divided into the training group and validation group in a proportion of 7:3. For the training data set, the t test, Chi-square test and Fisher exact test were used to analyze the sex, age and nodule location of the growth group and stable group. Two radiologists independently delineated the ROIs of the nodules to extract the radiomics characteristics using Pyradiomics. After dimension reduction by the LASSO algorithm, logistic regression analysis was performed on age and ten selected radiological features, and a prediction model was established and tested in the validation group. SVM, RF, MLP and AdaBoost models were also established, and the prediction effect was evaluated by ROC analysis.

Results

There was a significant difference in age between the growth group and the stable group (P < 0.05), but there was no significant difference in sex or nodule location (P > 0.05). The interclass correlation coefficients between the two observers were > 0.75. After dimension reduction by the LASSO algorithm, ten radiomic features were selected, including two shape-based features, one gray-level-cooccurence-matrix (GLCM), one first-order feature, one gray-level-run-length-matrix (GLRLM), three gray-level-dependence-matrix (GLDM) and two gray-level-size-zone-matrix (GLSZM). The logistic regression model combining age and radiomics features achieved an AUC of 0.87 and an accuracy of 0.82 in the training group and an AUC of 0.82 and an accuracy of 0.84 in the verification group for the prediction of nodule growth. For nonlinear models, in the training group, the AUCs of the SVM, RF, MLP and boost models were 0.95, 1.0, 1.0 and 1.0, respectively. In the validation group, the AUCs of the SVM, RF, MLP and boost models were 0.81, 0.77, 0.81, and 0.71, respectively.

Conclusions

In this study, we established several machine learning models that can successfully predict the growth of pulmonary nodules within one year. The logistic regression model combining age and imaging parameters has the best accuracy and generalization. This model is very helpful for the early treatment of pulmonary nodules and has important clinical significance.

Multimodal MRI-Based Radiomic Nomogram for the Early Differentiation of Recurrence and Pseudoprogression of High-Grade Glioma

Objective

To evaluate the diagnostic value of multimodal MRI radiomics based on T2-weighted fluid attenuated inversion recovery imaging (T2WI-FLAIR) combined with T1-weighted contrast enhanced imaging (T1WI-CE) in the early differentiation of high-grade glioma recurrence from pseudoprogression.

Methods

A total of one hundred eighteen patients with brain gliomas who were diagnosed from March 2014 to April 2020 were retrospectively analyzed. According to the clinical characteristics, the patients were randomly split into a training group (n = 83) and a test group (n = 35) at a 7 : 3 ratio. The region of interest (ROI) was delineated, and 2632 radiomic features were extracted. We used multiple logistic regression to establish a classification model, including the T1 model, T2 model, and T1 + T2 model, to differentiate recurrence from pseudoprogression. The diagnostic efficiency of the model was evaluated by calculating the area under the receiver operating characteristic curve (AUC) and accuracy (ACC) and by analyzing the calibration curve of the nomogram and decision curve.

Results

There were 75 cases of recurrence and 43 cases of pseudoprogression. The diagnostic efficacies of the multimodal MRI-based radiomic model were relatively high. The AUC values and ACC of the training group were 0.831 and 77.11%, respectively, and the AUC values and ACC of the test group were 0.829 and 88.57%, respectively. The calibration curve of the nomogram showed that the discrimination probability was consistent with the actual occurrence in the training group, and the discrimination probability was roughly the same as the actual occurrence in the test group. In the decision curve analysis, the T1 + T2 model showed greater overall net efficiency.

Conclusion

The multimodal MRI radiomic model has relatively high efficiency in the early differentiation of recurrence from pseudoprogression, and it could be helpful for clinicians in devising correct treatment plans so that patients can be treated in a timely and accurate manner.

AutoComBat: a generic method for harmonizing MRI-based radiomic features

The use of multicentric data is becoming essential for developing generalizable radiomic signatures. In particular, Magnetic Resonance Imaging (MRI) data used in brain oncology are often heterogeneous in terms of scanners and acquisitions, which significantly impact quantitative radiomic features. Various methods have been proposed to decrease dependency, including methods acting directly on MR images, i.e., based on the application of several preprocessing steps before feature extraction or the ComBat method, which harmonizes radiomic features themselves. The ComBat method used for radiomics may be misleading and presents some limitations, such as the need to know the labels associated with the "batch effect". In addition, a statistically representative sample is required and the applicability of a signature whose batch label is not present in the train set is not possible. This work aimed to compare a priori and a posteriori radiomic harmonization methods and propose a code adaptation to be machine learning compatible. Furthermore, we have developed AutoComBat, which aims to automatically determine the batch labels, using either MRI metadata or quality metrics as inputs of the proposed constrained clustering. A heterogeneous dataset consisting of high and low-grade gliomas coming from eight different centers was considered. The different methods were compared based on their ability to decrease relative standard deviation of radiomic features extracted from white matter and on their performance on a classification task using different machine learning models. ComBat and AutoComBat using image-derived quality metrics as inputs for batch assignment and preprocessing methods presented promising results on white matter harmonization, but with no clear consensus for all MR images. Preprocessing showed the best results on the T1w-gd images for the grading task. For T2w-flair, AutoComBat, using either metadata plus quality metrics or metadata alone as inputs, performs better than the conventional ComBat, highlighting its potential for data harmonization. Our results are MRI weighting, feature class and task dependent and require further investigations on other datasets.

A Radiomics Nomogram for Non-Invasive Prediction of Progression-Free Survival in Esophageal Squamous Cell Carcinoma

To construct a prognostic model for preoperative prediction on computed tomography (CT) images of esophageal squamous cell carcinoma (ESCC), we created radiomics signature with high throughput radiomics features extracted from CT images of 272 patients (204 in training and 68 in validation cohort). Multivariable logistic regression was applied to build the radiomics signature and the predictive nomogram model, which was composed of radiomics signature, traditional TNM stage, and clinical features. A total of 21 radiomics features were selected from 954 to build a radiomics signature which was significantly associated with progression-free survival (p &lt; 0.001). The area under the curve of performance was 0.878 (95% CI: 0.831–0.924) for the training cohort and 0.857 (95% CI: 0.767–0.947) for the validation cohort. The radscore of signatures' combination showed significant discrimination for survival status. Radiomics nomogram combined radscore with TNM staging and showed considerable improvement over TNM staging alone in the training cohort (C-index, 0.770 vs. 0.603; p &lt; 0.05), and it is the same with clinical data (C-index, 0.792 vs. 0.680; p &lt; 0.05), which were confirmed in the validation cohort. Decision curve analysis showed that the model would receive a benefit when the threshold probability was between 0 and 0.9. Collectively, multiparametric CT-based radiomics nomograms provided improved prognostic ability in ESCC.

AI-Based Detection, Classification and Prediction/Prognosis in Medical Imaging:: Towards Radiophenomics

Artificial intelligence (AI) techniques have significant potential to enable effective, robust, and automated image phenotyping including the identification of subtle patterns. AI-based detection searches the image space to find the regions of interest based on patterns and features. There is a spectrum of tumor histologies from benign to malignant that can be identified by AI-based classification approaches using image features. The extraction of minable information from images gives way to the field of "radiomics" and can be explored via explicit (handcrafted/engineered) and deep radiomics frameworks. Radiomics analysis has the potential to be used as a noninvasive technique for the accurate characterization of tumors to improve diagnosis and treatment monitoring. This work reviews AI-based techniques, with a special focus on oncological PET and PET/CT imaging, for different detection, classification, and prediction/prognosis tasks. We also discuss needed efforts to enable the translation of AI techniques to routine clinical workflows, and potential improvements and complementary techniques such as the use of natural language processing on electronic health records and neuro-symbolic AI techniques.

The Value of Predicting Human Epidermal Growth Factor Receptor 2 Status in Adenocarcinoma of the Esophagogastric Junction on CT-Based Radiomics Nomogram

Purpose

We developed and validated a CT-based radiomics nomogram to predict HER2 status in patients with adenocarcinoma of esophagogastric junction (AEG).

Method

A total of 101 patients with HER2-positive (n=46) and HER2-negative (n=55) esophagogastric junction adenocarcinoma (AEG) were retrospectively analyzed. They were then randomly divided into a training cohort (n=70) and a verification cohort (n=31). The radiomics features were obtained from the portal phase of the CT enhanced scan. We used the least absolute shrinkage and selection operator (LASSO) logistic regression method to select the best radiomics features in the training cohort, combined them linearly, and used the radiomics signature formula to calculate the radiomics score (Rad-score) of each AEG patient. A multivariable logistic regression method was applied to develop a prediction model that incorporated the radiomics signature and independent risk predictors. The prediction performance of the nomogram was evaluated using the training and validation cohorts.

Result

In the training (P<0.001) and verification groups (P<0.001), the radiomics signature combined with seven radiomics features was significantly correlated with HER2 status. The nomogram composed of CT-reported T stage and radiomics signature showed very good predictive performance for HER2 status. The area under the curve (AUC) of the training cohort was 0.946 (95% CI: 0.919–0.973), and that of the validation group was 0.903 (95% CI: 0.847–0.959). The calibration curve of the radiomics nomogram showed a good degree of calibration. Decision-curve analysis revealed that the radiomics nomogram was useful.

Conclusion

The nomogram CT-based radiomics signature combined with CT-reported T stage can better predict the HER2 status of AEG before surgery. It can be used as a non-invasive prediction tool for HER2 status and is expected to guide clinical treatment decisions in clinical practice, and it can assist in the formulation of individualized treatment plans.

A Comprehensive Nomogram Combining CT Imaging with Clinical Features for Prediction of Lymph Node Metastasis in Stage I-IIIB Non-small Cell Lung Cancer

OBJECTIVE

The status of lymph node metastasis (LNM) is highly correlated with the recurrence and survival outcomes of patients with lung cancer. Thus, a tool that predicts LNM could benefit patient treatment and prognosis. The present study established a new radiomic model by combining computed tomography (CT) radiomic features and clinical parameters to predict the LNM status in patients with non-small cell lung cancer (NSCLC).

METHODS

Demographic parameters and clinical laboratory values were analyzed in 217 patients with stage I-IIIB NSCLC; 107 of the patients received CT scanning and radiomic characteristics were used for LNM assessment (76 in the training cohort and 31 in the validation cohort). The minimum redundancy maximum relevance (mRMR) and the least absolute shrinkage and selection operator (LASSO) regression model were used to select the most predictive features on the basis of the 76 patients in the training set. The value of the area under the receiver operator characteristic (ROC) curve (AUC) was adopted to determine the correlation between LN status and the radiomics signature in training cohorts and then validated in the 31 patients of validation set. The radiomics nomogram was analyzed using univariate and multivariate logistic regression. Decision curve analysis (DCA) was performed to evaluate the clinical utility of this model.

RESULTS

This was a retrospective study. Five radiomic characteristics were significantly correlated with LNM in the two cohorts (P < 0.05). The radiomic nomogram that incorporated the above radiomic characteristics, the RDW, and the CT-based LN status had satisfactory discrimination and calibration in the training (AUC, 0.79; 95% CI 0.69-0.89) and validation cohorts (AUC, 0.70; 95% CI 0.50-0.89).The DCA showed that the developed nomogram had promising clinical utility.

CONCLUSIONS

The developed nomogram, combined with preoperative radiomics evidence, the RDW, and the CT-based LN status, has the potential to preoperatively predict LNM with high accuracy and can facilitate the prediction of LN status for NSCLC patients.

Evolving Role and Translation of Radiomics and Radiogenomics in Adult and Pediatric Neuro-Oncology

Exponential technologic advancements in imaging, high-performance computing, and artificial intelligence, in addition to increasing access to vast amounts of diverse data, have revolutionized the role of imaging in medicine. Radiomics is defined as a high-throughput feature-extraction method that unlocks microscale quantitative data hidden within standard-of-care medical imaging. Radiogenomics is defined as the linkage between imaging and genomics information. Multiple radiomics and radiogenomics studies performed on conventional and advanced neuro-oncology image modalities show that they have the potential to differentiate pseudoprogression from true progression, classify tumor subgroups, and predict recurrence, survival, and mutation status with high accuracy. In this article, we outline the technical steps involved in radiomics and radiogenomics analyses with the use of artificial intelligence methods and review current applications in adult and pediatric neuro-oncology.

Glioblastoma Surgery Imaging-Reporting and Data System: Validation and Performance of the Automated Segmentation Task

For patients with presumed glioblastoma, essential tumor characteristics are determined from preoperative MR images to optimize the treatment strategy. This procedure is time-consuming and subjective, if performed by crude eyeballing or manually. The standardized GSI-RADS aims to provide neurosurgeons with automatic tumor segmentations to extract tumor features rapidly and objectively. In this study, we improved automatic tumor segmentation and compared the agreement with manual raters, describe the technical details of the different components of GSI-RADS, and determined their speed. Two recent neural network architectures were considered for the segmentation task: nnU-Net and AGU-Net. Two preprocessing schemes were introduced to investigate the tradeoff between performance and processing speed. A summarized description of the tumor feature extraction and standardized reporting process is included. The trained architectures for automatic segmentation and the code for computing the standardized report are distributed as open-source and as open-access software. Validation studies were performed on a dataset of 1594 gadolinium-enhanced T1-weighted MRI volumes from 13 hospitals and 293 T1-weighted MRI volumes from the BraTS challenge. The glioblastoma tumor core segmentation reached a Dice score slightly below 90%, a patientwise F1-score close to 99%, and a 95th percentile Hausdorff distance slightly below 4.0 mm on average with either architecture and the heavy preprocessing scheme. A patient MRI volume can be segmented in less than one minute, and a standardized report can be generated in up to five minutes. The proposed GSI-RADS software showed robust performance on a large collection of MRI volumes from various hospitals and generated results within a reasonable runtime.

Radiomics and radiogenomics in gliomas: a contemporary update

The natural history and treatment landscape of primary brain tumours are complicated by the varied tumour behaviour of primary or secondary gliomas (high-grade transformation of low-grade lesions), as well as the dilemmas with identification of radiation necrosis, tumour progression, and pseudoprogression on MRI. Radiomics and radiogenomics promise to offer precise diagnosis, predict prognosis, and assess tumour response to modern chemotherapy/immunotherapy and radiation therapy. This is achieved by a triumvirate of morphological, textural, and functional signatures, derived from a high-throughput extraction of quantitative voxel-level MR image metrics. However, the lack of standardisation of acquisition parameters and inconsistent methodology between working groups have made validations unreliable, hence multi-centre studies involving heterogenous study populations are warranted. We elucidate novel radiomic and radiogenomic workflow concepts and state-of-the-art descriptors in sub-visual MR image processing, with relevant literature on applications of such machine learning techniques in glioma management.

Co-Clinical Imaging Resource Program (CIRP): Bridging the Translational Divide to Advance Precision Medicine

The National Institutes of Health’s (National Cancer Institute) precision medicine initiative emphasizes the biological and molecular bases for cancer prevention and treatment. Importantly, it addresses the need for consistency in preclinical and clinical research. To overcome the translational gap in cancer treatment and prevention, the cancer research community has been transitioning toward using animal models that more fatefully recapitulate human tumor biology. There is a growing need to develop best practices in translational research, including imaging research, to better inform therapeutic choices and decision-making. Therefore, the National Cancer Institute has recently launched the Co-Clinical Imaging Research Resource Program (CIRP). Its overarching mission is to advance the practice of precision medicine by establishing consensus-based best practices for co-clinical imaging research by developing optimized state-of-the-art translational quantitative imaging methodologies to enable disease detection, risk stratification, and assessment/prediction of response to therapy. In this communication, we discuss our involvement in the CIRP, detailing key considerations including animal model selection, co-clinical study design, need for standardization of co-clinical instruments, and harmonization of preclinical and clinical quantitative imaging pipelines. An underlying emphasis in the program is to develop best practices toward reproducible, repeatable, and precise quantitative imaging biomarkers for use in translational cancer imaging and therapy. We will conclude with our thoughts on informatics needs to enable collaborative and open science research to advance precision medicine.

Predicting local failure of brain metastases after stereotactic radiosurgery with radiomics on planning MR images and dose maps

PURPOSE

Stereotactic radiosurgery (SRS) has become an important modality in the treatment of brain metastases. The purpose of this study is to investigate the potential of radiomic features from planning magnetic resonance (MR) images and dose maps to predict local failure after SRS for brain metastases.

MATERIALS/METHODS

Twenty-eight patients who received Gamma Knife (GK) radiosurgery for brain metastases were retrospectively reviewed in this IRB-approved study. 179 irradiated tumors included 42 that locally failed within one-year follow-up. Using SRS tumor volumes, radiomic features were calculated on T1-weighted contrast-enhanced MR images acquired for treatment planning and planned dose maps. 125 radiomic features regarding tumor shape, dose distribution, MR intensities and textures were extracted for each tumor. Logistic regression with automatic feature selection was built to predict tumor progression from local control after SRS. Feature selection and model evaluation using receiver operating characteristic (ROC) curves were performed in a nested cross validation (CV) scheme. The associations between selected radiomic features and treatment outcomes were statistically assessed by univariate analysis.

RESULTS

The logistic model with feature selection achieved ROC AUC of 0.82 ± 0.09 on 5-fold CV, providing 83% sensitivity and 70% specificity for predicting local failure. A total of 10 radiomic features including 1 shape feature, 6 MR images and 3 dose distribution features were selected. These features were significantly associated with treatment outcomes (p < 0.05). The model was validated on independent holdout data with an AUC of 0.78.

CONCLUSIONS

Radiomic features from planning MR images and dose maps provided prognostic information in SRS for brain metastases. A model built on the radiomic features shows promise for early prediction of tumor local failure after treatment, potentially aiding in personalized care for brain metastases.

Development of a Machine Learning Classifier Based on Radiomic Features Extracted From Post-Contrast 3D T1-Weighted MR Images to Distinguish Glioblastoma From Solitary Brain Metastasis

Objectives

To differentiate Glioblastomas (GBM) and Brain Metastases (BM) using a radiomic features-based Machine Learning (ML) classifier trained from post-contrast three-dimensional T1-weighted (post-contrast 3DT1) MR imaging, and compare its performance in medical diagnosis versus human experts, on a testing cohort.

Methods

We enrolled 143 patients (71 GBM and 72 BM) in a retrospective bicentric study from January 2010 to May 2019 to train the classifier. Post-contrast 3DT1 MR images were performed on a 3-Tesla MR unit and 100 radiomic features were extracted. Selection and optimization of the Machine Learning (ML) classifier was performed using a nested cross-validation. Sensitivity, specificity, balanced accuracy, and area under the receiver operating characteristic curve (AUC) were calculated as performance metrics. The model final performance was cross-validated, then evaluated on a test set of 37 patients, and compared to human blind reading using a McNemar’s test.

Results

The ML classifier had a mean [95% confidence interval] sensitivity of 85% [77; 94], a specificity of 87% [78; 97], a balanced accuracy of 86% [80; 92], and an AUC of 92% [87; 97] with cross-validation. Sensitivity, specificity, balanced accuracy and AUC were equal to 75, 86, 80 and 85% on the test set. Sphericity 3D radiomic index highlighted the highest coefficient in the logistic regression model. There were no statistical significant differences observed between the performance of the classifier and the experts’ blinded examination.

Conclusions

The proposed diagnostic support system based on radiomic features extracted from post-contrast 3DT1 MR images helps in differentiating solitary BM from GBM with high diagnosis performance and generalizability.

Radiomics-based neural network predicts recurrence patterns in glioblastoma using dynamic susceptibility contrast-enhanced MRI

Glioblastoma remains the most devastating brain tumor despite optimal treatment, because of the high rate of recurrence. Distant recurrence has distinct genomic alterations compared to local recurrence, which requires different treatment planning both in clinical practice and trials. To date, perfusion-weighted MRI has revealed that perfusional characteristics of tumor are associated with prognosis. However, not much research has focused on recurrence patterns in glioblastoma: namely, local and distant recurrence. Here, we propose two different neural network models to predict the recurrence patterns in glioblastoma that utilizes high-dimensional radiomic profiles based on perfusion MRI: area under the curve (AUC) (95% confidence interval), 0.969 (0.903-1.000) for local recurrence; 0.864 (0.726-0.976) for distant recurrence for each patient in the validation set. This creates an opportunity to provide personalized medicine in contrast to studies investigating only group differences. Moreover, interpretable deep learning identified that salient radiomic features for each recurrence pattern are related to perfusional intratumoral heterogeneity. We also demonstrated that the combined salient radiomic features, or "radiomic risk score", increased risk of recurrence/progression (hazard ratio, 1.61; p = 0.03) in multivariate Cox regression on progression-free survival.

Machine Learning for the Prediction of Molecular Markers in Glioma on Magnetic Resonance Imaging: A Systematic Review and Meta-Analysis

BACKGROUND

Molecular characterization of glioma has implications for prognosis, treatment planning, and prediction of treatment response. Current histopathology is limited by intratumoral heterogeneity and variability in detection methods. Advances in computational techniques have led to interest in mining quantitative imaging features to noninvasively detect genetic mutations.

OBJECTIVE

To evaluate the diagnostic accuracy of machine learning (ML) models in molecular subtyping gliomas on preoperative magnetic resonance imaging (MRI).

METHODS

A systematic search was performed following PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) guidelines to identify studies up to April 1, 2020. Methodological quality of studies was assessed using the Quality Assessment for Diagnostic Accuracy Studies (QUADAS)-2. Diagnostic performance estimates were obtained using a bivariate model and heterogeneity was explored using metaregression.

RESULTS

Forty-four original articles were included. The pooled sensitivity and specificity for predicting isocitrate dehydrogenase (IDH) mutation in training datasets were 0.88 (95% CI 0.83-0.91) and 0.86 (95% CI 0.79-0.91), respectively, and 0.83 to 0.85 in validation sets. Use of data augmentation and MRI sequence type were weakly associated with heterogeneity. Both O6-methylguanine-DNA methyltransferase (MGMT) gene promoter methylation and 1p/19q codeletion could be predicted with a pooled sensitivity and specificity between 0.76 and 0.83 in training datasets.

CONCLUSION

ML application to preoperative MRI demonstrated promising results for predicting IDH mutation, MGMT methylation, and 1p/19q codeletion in glioma. Optimized ML models could lead to a noninvasive, objective tool that captures molecular information important for clinical decision making. Future studies should use multicenter data, external validation and investigate clinical feasibility of ML models.

Creative and Innovative Methods and Techniques for the Challenges in the Management of Adult Craniopharyngioma

Craniopharyngioma remains a major challenge in daily clinical practice. The pathobiology of the tumor is still elusive, and there are no consensus or treatment guidelines on the optimal management strategy for this relatively rare tumor. However, recent technical and scientific advances, including genomic and radiomic profiling, innovation in surgical approaches, more precise radiotherapy protocols, targeted therapy, and restoration of lost functions all have the potential to significantly improve the outcome of patients with craniopharyngioma in the near future. Although many of these innovative tools in the new armamentarium of the clinician are still in their infancy, they could reduce craniopharyngioma-related morbidity and mortality and improve the patients' quality of life. In this article, we discuss these creative and innovative approaches that may offer solutions to the obstacles faced in treating craniopharyngioma and future possibilities in improving the care of these patients.

Glioblastoma and primary central nervous system lymphoma: differentiation using MRI derived first-order texture analysis - a machine learning study

OBJECTIVES

To evaluate the diagnostic performance of multiple machine learning classifier models derived from first-order histogram texture parameters extracted from T1-weighted contrast-enhanced images in differentiating glioblastoma and primary central nervous system lymphoma.

METHODS

Retrospective study with 97 glioblastoma and 46 primary central nervous system lymphoma patients. Thirty-six different combinations of classifier models and feature selection techniques were evaluated. Five-fold nested cross-validation was performed. Model performance was assessed for whole tumour and largest single slice using receiver operating characteristic curve.

RESULTS

The cross-validated model performance was relatively similar for the top performing models for both whole tumour and largest single slice (area under the curve 0.909-0.924). However, there was a considerable difference between the worst performing model (logistic regression with full feature set, area under the curve 0.737) and the highest performing model for whole tumour (least absolute shrinkage and selection operator model with correlation filter, area under the curve 0.924). For single slice, the multilayer perceptron model with correlation filter had the highest performance (area under the curve 0.914). No significant difference was seen between the diagnostic performance of the top performing model for both whole tumour and largest single slice.

CONCLUSIONS

T1 contrast-enhanced derived first-order texture analysis can differentiate between glioblastoma and primary central nervous system lymphoma with good diagnostic performance. The machine learning performance can vary significantly depending on the model and feature selection methods. Largest single slice and whole tumour analysis show comparable diagnostic performance.

The value of MR-based radiomics in identifying residual disease in patients with carcinoma in situ after cervical conization

Carcinoma in situ (CIS) of the uterine cervix is a precursor to cervical carcinoma. However, hysterectomy can be avoided in patients who can be treated by cone biopsy. Previous studies have shown that imaging-based approaches allow for the noninvasive visualization of cervical cancer, and radiomics has high accuracy in classifying cancer and predicting treatment outcome for different cancer types. To develop a magnetic resonance (MR)-based radiomics model for identifying residual disease in patients with CIS after cervical conization. Patients who had CIS after conization and finally underwent hysterectomy were collected to comprise a database to establish an imaging model for predicting the residual status after conization. Then, patients who opted for uterine preservation were classified as high-risk or low-risk patients according to the model. The disease-free survival was compared between the different risk groups using the Kaplan–Meier curve. The model built with the Boruta features outperformed the random forest model. Further validation with patients with uterine preservation showed that the patients classified as high risk were more likely to have tumor recurrence/residual disease in the follow-up period. In conclusion, radiomics can be used to identify residual disease in patients with CIS after cervical conization and could have the potential to predict recurrence in patients who opt for uterine preservation.

Novel, non-invasive imaging approach to identify patients with advanced non-small cell lung cancer at risk of hyperprogressive disease with immune checkpoint blockade

Purpose

Hyperprogression is an atypical response pattern to immune checkpoint inhibition that has been described within non-small cell lung cancer (NSCLC). The paradoxical acceleration of tumor growth after immunotherapy has been associated with significantly shortened survival, and currently, there are no clinically validated biomarkers to identify patients at risk of hyperprogression.

Experimental design

A total of 109 patients with advanced NSCLC who underwent monotherapy with Programmed cell death protein-1 (PD1)/Programmed death-ligand-1 (PD-L1) inhibitors were included in the study. Using RECIST measurements, we divided the patients into responders (n=50) (complete/partial response or stable disease) and non-responders (n=59) (progressive disease). Tumor growth kinetics were used to further identify hyperprogressors (HPs, n=19) among non-responders. Patients were randomized into a training set (D₁=30) and a test set (D₂=79) with the essential caveat that HPs were evenly distributed among the two sets. A total of 198 radiomic textural patterns from within and around the target nodules and features relating to tortuosity of the nodule associated vasculature were extracted from the pretreatment CT scans.

Results

The random forest classifier using the top features associated with hyperprogression was able to distinguish between HP and other radiographical response patterns with an area under receiver operating curve of 0.85±0.06 in the training set (D₁=30) and 0.96 in the validation set (D₂=79). These features included one peritumoral texture feature from 5 to 10 mm outside the tumor and two nodule vessel-related tortuosity features. Kaplan-Meier survival curves showed a clear stratification between classifier predicted HPs versus non-HPs for overall survival (D₂: HR=2.66, 95% CI 1.27 to 5.55; p=0.009).

Conclusions

Our study suggests that image-based radiomics markers extracted from baseline CTs of advanced NSCLC treated with PD-1/PD-L1 inhibitors may help identify patients at risk of hyperprogressions.

A Nomogram Based on Radiomics with Mammography Texture Analysis for the Prognostic Prediction in Patients with Triple-Negative Breast Cancer

Objectives

To develop and validate a radiomics-based nomogram with texture features from mammography for the prognostic prediction in patients with early-stage triple-negative breast cancer (TNBC).

Methods

The study included 200 consecutive patients with TNBC (training cohort: n = 133, validation cohort: n = 67). A total of 136 mammography-derived textural features were extracted, and LASSO (least absolute shrinkage and selection operator) was applied to select features for building the radiomics score (Rad-score). After univariate and multivariate logistic regression, a radiomics-based nomogram was constructed with independent prognostic factors. The discrimination and calibration power were assessed, and further the clinical applicability of the nomograms was evaluated.

Results

Among the 136 mammography-derived textural features, fourteen were used to build the Rad-score after LASSO regression. A radiomics nomogram that incorporates Rad-score and pN stage was constructed. This nomogram achieved a C-index of 0.873 (95% CI: 0.758-0.989) for predicting iDFS (invasive disease-free survival), which outperformed the clinical model. Moreover, it is feasible to stratify patients into high-risk and low-risk groups based on the optimal cut-off point of Rad-score. The validations of the nomogram confirmed favorable discrimination and considerable predictive efficiency.

Conclusions

The radiomics nomogram that incorporates Rad-score and pN stage exhibited favorable performance in the prediction of iDFS in patients with early-stage TNBCs.

Repeatability of radiomic features in magnetic resonance imaging of glioblastoma: Test-retest and image registration analyses

PURPOSE

To assess the repeatability of radiomic features in magnetic resonance (MR) imaging of glioblastoma (GBM) tumors with respect to test-retest, different image registration approaches and inhomogeneity bias field correction.

METHODS

We analyzed MR images of 17 GBM patients including T1- and T2-weighted images (performed within the same imaging unit on two consecutive days). For image segmentation, we used a comprehensive segmentation approach including entire tumor, active area of tumor, necrotic regions in T1-weighted images, and edema regions in T2-weighted images (test studies only; registration to retest studies is discussed next). Analysis included N3, N4 as well as no bias correction performed on raw MR images. We evaluated 20 image registration approaches, generated by cross-combination of four transformation and five cost function methods. In total, 714 images (17 patients × 2 images × ((4 transformations × 5 cost functions) + 1 test image) and 2856 segmentations (714 images × 4 segmentations) were prepared for feature extraction. Various radiomic features were extracted, including the use of preprocessing filters, specifically wavelet (WAV) and Laplacian of Gaussian (LOG), as well as discretization into fixed bin width and fixed bin count (16, 32, 64, 128, and 256), Exponential, Gradient, Logarithm, Square and Square Root scales. Intraclass correlation coefficients (ICC) were calculated to assess the repeatability of MRI radiomic features (high repeatability defined as ICC ≥ 95%).

RESULTS

In our ICC results, we observed high repeatability (ICC ≥ 95%) with respect to image preprocessing, different image registration algorithms, and test-retest analysis, for example: RLNU and GLNU from GLRLM, GLNU and DNU from GLDM, Coarseness and Busyness from NGTDM, GLNU and ZP from GLSZM, and Energy and RMS from first order. Highest fraction (percent) of repeatable features was observed, among registration techniques, for the method Full Affine transformation with 12 degrees of freedom using Mutual Information cost function (mean 32.4%), and among image processing methods, for the method Laplacian of Gaussian (LOG) with Sigma (2.5-4.5 mm) (mean 78.9%). The trends were relatively consistent for N4, N3, or no bias correction.

CONCLUSION

Our results showed varying performances in repeatability of MR radiomic features for GBM tumors due to test-retest and image registration. The findings have implications for appropriate usage in diagnostic and predictive models.

Pituitary Tumors in the Computational Era, Exploring Novel Approaches to Diagnosis, and Outcome Prediction with Machine Learning

BACKGROUND

Machine learning has emerged as a viable asset in the setting of pituitary surgery. In the past decade, the number of machine learning models developed to aid in the diagnosis of pituitary lesions and predict intraoperative and postoperative complications following transsphenoidal surgery has increased exponentially. As computational processing power continues to increase, big data sets continue to expand, and learning algorithms continue to surpass gold standard predictive tools, machine learning will serve to become an important component in improving patient care and outcomes.

METHODS

Relevant studies were identified based on a literature search in PubMed and MEDLINE databases, as well as from other sources including reference lists of published articles.

RESULTS

Radiomics and artificial neural networks comprise the majority of machine learning-based applications in pituitary surgery. Radiomics serves to quantify specific imaging features, which can then be used to noninvasively identify tumor characteristics and make definitive diagnoses, circumventing presurgical biopsy altogether. Neural networks can be adapted to predict intraoperative changes in visual evoked potentials or cerebral spinal fluid leak. In addition, these algorithms may be combined with others to predict tumor aggressiveness, gross total resection, recurrence and remission, and even total cost burden.

CONCLUSIONS

The field of machine learning is broad, with radiomics and artificial neural networks comprising 2 commonly used supervised learning methods in pituitary surgery. Given the large heterogeneity of pituitary and sellar lesions, the promise of machine learning lies in its ability to identify relationships and patterns that are otherwise hidden from standard statistical methods. While machine learning has great potential as a clinical adjunct during the surgical preplanning process and in predicting complications and outcomes, challenges moving forward include standardization and validation of these paradigms.

Glioblastomas and brain metastases differentiation following an MRI texture analysis-based radiomics approach

PURPOSE

To evaluate the potential of 2D texture features extracted from magnetic resonance (MR) images for differentiating brain metastasis (BM) and glioblastomas (GBM) following a radiomics approach.

METHODS

This retrospective study included 50 patients with BM and 50 with GBM who underwent T1-weighted MRI between December 2010 and January 2017. Eighty-eight rotation-invariant texture features were computed for each segmented lesion using six texture analysis methods. These features were also extracted from the four images obtained after applying the discrete wavelet transform (88 features × 4 images). Three feature selection methods and five predictive models were evaluated. A 5-fold cross-validation scheme was used to randomly split the study group into training (80 patients) and testing (20 patients), repeating the process ten times. Classification was evaluated computing the average area under the receiver operating characteristic curve. Sensibility, specificity and accuracy were also computed. The whole process was tested quantizing the images with different gray-level values to evaluate their influence in the final results.

RESULTS

Highest classification accuracy was obtained using the original images quantized with 128 gray-levels and a feature selection method based on the p-value. The best overall performance was achieved using a support vector machine model with a subset of 32 features (AUC = 0.896 ± 0.067, sensitivity of 82% and specificity of 80%). Naïve Bayes and k-nearest neighbors models showed also valuable results (AUC ≈ 0.8) with a lower number of features (<13), thus suggesting that these models may be more generalizable when using external validations.

CONCLUSION

The proposed radiomics MRI approach is able to discriminate between GBM and BM with high accuracy employing a set of 2D texture features, thus helping in the diagnosis of brain lesions in a fast and non-invasive way.

Contrast-enhanced T1-weighted image radiomics of brain metastases may predict EGFR mutation status in primary lung cancer

Identification of EGFR mutations is critical to the treatment of primary lung cancer and brain metastases (BMs). Here, we explored whether radiomic features of contrast-enhanced T1-weighted images (T1WIs) of BMs predict EGFR mutation status in primary lung cancer cases. In total, 1209 features were extracted from the contrast-enhanced T1WIs of 61 patients with 210 measurable BMs. Feature selection and classification were optimized using several machine learning algorithms. Ten-fold cross-validation was applied to the T1WI BM dataset (189 BMs for training and 21 BMs for the test set). Area under receiver operating characteristic curves (AUC), accuracy, sensitivity, and specificity were calculated. Subgroup analyses were also performed according to metastasis size. For all measurable BMs, random forest (RF) classification with RF selection demonstrated the highest diagnostic performance for identifying EGFR mutation (AUC: 86.81). Support vector machine and AdaBoost were comparable to RF classification. Subgroup analyses revealed that small BMs had the highest AUC (89.09). The diagnostic performance for large BMs was lower than that for small BMs (the highest AUC: 78.22). Contrast-enhanced T1-weighted image radiomics of brain metastases predicted the EGFR mutation status of lung cancer BMs with good diagnostic performance. However, further study is necessary to apply this algorithm more widely and to larger BMs.

A simple model for glioma grading based on texture analysis applied to conventional brain MRI

Accuracy of glioma grading is fundamental for the diagnosis, treatment planning and prognosis of patients. The purpose of this work was to develop a low-cost and easy-to-implement classification model which distinguishes low-grade gliomas (LGGs) from high-grade gliomas (HGGs), through texture analysis applied to conventional brain MRI. Different combinations of MRI contrasts (T1Gd and T2) and one segmented glioma region (necrotic and non-enhancing tumor core, NCR/NET) were studied. Texture features obtained from the gray level size zone matrix (GLSZM) were calculated. An under-sampling method was proposed to divide the data into different training subsets and subsequently extract complementary information for the creation of distinct classification models. The sensitivity, specificity and accuracy of the models were calculated, and the best model explicitly reported. The best model included only three texture features and reached a sensitivity, specificity and accuracy of 94.12%, 88.24% and 91.18%, respectively. According to the features of the model, when the NCR/NET region was studied, HGGs had a more heterogeneous texture than LGGs in the T1Gd images, and LGGs had a more heterogeneous texture than HGGs in the T2 images. These novel results partially contrast with results from the literature. The best model proved to be useful for the classification of gliomas. Complementary results showed that the heterogeneity of gliomas depended on the MRI contrast studied. The chosen model stands out as a simple, low-cost, easy-to-implement, reproducible and highly accurate glioma classifier. Importantly, it should be accessible to populations with reduced economic and scientific resources.

Preoperative diagnosis of malignant pulmonary nodules in lung cancer screening with a radiomics nomogram

Background

Lung cancer is the most commonly diagnosed cancer worldwide. Its survival rate can be significantly improved by early screening. Biomarkers based on radiomics features have been found to provide important physiological information on tumors and considered as having the potential to be used in the early screening of lung cancer. In this study, we aim to establish a radiomics model and develop a tool to improve the discrimination between benign and malignant pulmonary nodules.

Methods

A retrospective study was conducted on 875 patients with benign or malignant pulmonary nodules who underwent computed tomography (CT) examinations between June 2013 and June 2018. We assigned 612 patients to a training cohort and 263 patients to a validation cohort. Radiomics features were extracted from the CT images of each patient. Least absolute shrinkage and selection operator (LASSO) was used for radiomics feature selection and radiomics score calculation. Multivariate logistic regression analysis was used to develop a classification model and radiomics nomogram. Radiomics score and clinical variables were used to distinguish benign and malignant pulmonary nodules in logistic model. The performance of the radiomics nomogram was evaluated by the area under the curve (AUC), calibration curve and Hosmer‐Lemeshow test in both the training and validation cohorts.

Results

A radiomics score was built and consisted of 20 features selected by LASSO from 1288 radiomics features in the training cohort. The multivariate logistic model and radiomics nomogram were constructed using the radiomics score and patients’ age. Good discrimination of benign and malignant pulmonary nodules was obtained from the training cohort (AUC, 0.836; 95% confidence interval [CI]: 0.793‐0.879) and validation cohort (AUC, 0.809; 95% CI: 0.745‐0.872). The Hosmer‐Lemeshow test also showed good performance for the logistic regression model in the training cohort (P = 0.765) and validation cohort (P = 0.064). Good alignment with the calibration curve indicated the good performance of the nomogram.

Conclusions

The established radiomics nomogram is a noninvasive preoperative prediction tool for malignant pulmonary nodule diagnosis. Validation revealed that this nomogram exhibited excellent discrimination and calibration capacities, suggesting its clinical utility in the early screening of lung cancer.

An ensemble learning approach for brain cancer detection exploiting radiomic features

BACKGROUND AND OBJECTIVE

The brain cancer is one of the most aggressive tumour: the 70% of the patients diagnosed with this malignant cancer will not survive. Early detection of brain tumours can be fundamental to increase survival rates. The brain cancers are classified into four different grades (i.e., I, II, III and IV) according to how normal or abnormal the brain cells look. The following work aims to recognize the different brain cancer grades by analysing brain magnetic resonance images.

METHODS

A method to identify the components of an ensemble learner is proposed. The ensemble learner is focused on the discrimination between different brain cancer grades using non invasive radiomic features. The considered radiomic features are belonging to five different groups: First Order, Shape, Gray Level Co-occurrence Matrix, Gray Level Run Length Matrix and Gray Level Size Zone Matrix. We evaluate the features effectiveness through hypothesis testing and through decision boundaries, performance analysis and calibration plots thus we select the best candidate classifiers for the ensemble learner.

RESULTS

We evaluate the proposed method with 111,205 brain magnetic resonances belonging to two freely available data-sets for research purposes. The results are encouraging: we obtain an accuracy of 99% for the benign grade I and the II, III and IV malignant brain cancer detection.

CONCLUSION

The experimental results confirm that the ensemble learner designed with the proposed method outperforms the current state-of-the-art approaches in brain cancer grade detection starting from magnetic resonance images.

Amide Proton Transfer Imaging in Predicting Isocitrate Dehydrogenase 1 Mutation Status of Grade II/III Gliomas Based on Support Vector Machine

Background

To compare the efficacies of univariate and radiomics analyses of amide proton transfer weighted (APT_W) imaging in predicting isocitrate dehydrogenase 1 (IDH1) mutation of grade II/III gliomas.

Methods

Fifty-nine grade II/III glioma patients with known IDH1 mutation status were prospectively included (IDH1 wild type, 16; IDH1 mutation, 43). A total of 1044 quantitative radiomics features were extracted from APT_W images. The efficacies of univariate and radiomics analyses in predicting IDH1 mutation were compared. Feature values were compared between two groups with independent t-test and receiver operating characteristic (ROC) analysis was applied to evaluate the predicting efficacy of each feature. Cases were randomly assigned to either the training (n = 49) or test cohort (n = 10) for the radiomics analysis. Support vector machine with recursive feature elimination (SVM-RFE) was adopted to select the optimal feature subset. The adverse impact of the imbalance dataset in the training cohort was solved by synthetic minority oversampling technique (SMOTE). Subsequently, the performance of SVM model was assessed on both training and test cohort.

Results

As for univariate analysis, 18 features were significantly different between IDH1 wild-type and mutant groups (P < 0.05). Among these parameters, High Gray Level Run Emphasis All Direction offset 8 SD achieved the biggest area under the curve (AUC) (0.769) with the accuracy of 0.799. As for radiomics analysis, SVM model was established using 19 features selected with SVM-RFE. The AUC and accuracy for IDH1 mutation on training set were 0.892 and 0.952, while on the testing set were 0.7 and 0.84, respectively.

Conclusion

Radiomics strategy based on APT image features is potentially useful for preoperative estimating IDH1 mutation status.

Manual and semiautomatic segmentation of bone sarcomas on MRI have high similarity

The aims of this study were to evaluate the intra- and interobserver reproducibility of manual segmentation of bone sarcomas in magnetic resonance imaging (MRI) studies and to compare manual and semiautomatic segmentation methods. This retrospective study included twelve osteosarcoma and eight Ewing sarcoma MRI studies performed prior to any therapeutic intervention. All cases were histopathologically confirmed. Three radiologists used 3D-Slicer software to perform manual segmentation of bone sarcomas in a blinded and independent manner. One radiologist segmented manually and also performed semiautomatic segmentation with the GrowCut tool. Segmentation exercises were timed for comparison. The dice similarity coefficient (DSC) and Hausdorff distance (HD) were used to evaluate similarity between the segmentation results and further statistical analyses were performed to compare DSC, HD, and volumetric results. Manual segmentation was reproducible with intraobserver DSC varying from 0.83 to 0.97 and HD from 3.37 to 28.73 mm. Interobserver DSC of manual segmentation showed variation from 0.73 to 0.97 and HD from 3.93 to 33.40 mm. Semiautomatic segmentation compared to manual segmentation resulted in DSCs of 0.71−0.96 and HDs of 5.38−31.54 mm. Semiautomatic segmentation required significantly less time compared to manual segmentation (P value ≤0.05). Among all situations compared, tumor volumetry did not show significant statistical differences (P value >0.05). We found excellent intra- and interobserver agreement for manual segmentation of osteosarcoma and Ewing sarcoma. There was high similarity between manual and semiautomatic segmentation, with a significant reduction of segmentation time using the semiautomatic method.

Radiogenomic analysis of vascular endothelial growth factor in patients with diffuse gliomas

OBJECTIVE

To predict vascular endothelial growth factor (VEGF) expression in patients with diffuse gliomas using radiomic analysis.

MATERIALS AND METHODS

Preoperative magnetic resonance images were retrospectively obtained from 239 patients with diffuse gliomas (World Health Organization grades II-IV). The patients were randomly assigned to a training group (n = 160) or a validation group (n = 79) at a 2:1 ratio. For each patient, a total of 431 radiomic features were extracted. The minimum redundancy maximum relevance (mRMR) algorithm was used for feature selection. A machine-learning model for predicting VEGF status was then developed using the selected features and a support vector machine classifier. The predictive performance of the model was evaluated in both groups using receiver operating characteristic curve analysis, and correlations between selected features were assessed.

RESULTS

Nine radiomic features were selected to generate a VEGF-associated radiomic signature of diffuse gliomas based on the mRMR algorithm. This radiomic signature consisted of two first-order statistics or related wavelet features (Entropy and Minimum) and seven textural features or related wavelet features (including Cluster Tendency and Long Run Low Gray Level Emphasis). The predictive efficiencies measured by the area under the curve were 74.1% in the training group and 70.2% in the validation group. The overall correlations between the 9 radiomic features were low in both groups.

CONCLUSIONS

Radiomic analysis facilitated efficient prediction of VEGF status in diffuse gliomas, suggesting that using tumor-derived radiomic features for predicting genomic information is feasible.

Pediatric Brain Tumor Genetics: What Radiologists Need to Know

Brain tumors are the most common solid tumors in the pediatric population. Pediatric neuro-oncology has changed tremendously during the past decade owing to ongoing genomic advances. The diagnosis, prognosis, and treatment of pediatric brain tumors are now highly reliant on the genetic profile and histopathologic features of the tumor rather than the histopathologic features alone, which previously were the reference standard. The clinical information expected to be gleaned from radiologic interpretations also has evolved. Imaging is now expected to not only lead to a relevant short differential diagnosis but in certain instances also aid in predicting the specific tumor and subtype and possibly the prognosis. These processes fall under the umbrella of radiogenomics. Therefore, to continue to actively participate in patient care and/or radiogenomic research, it is important that radiologists have a basic understanding of the molecular mechanisms of common pediatric central nervous system tumors. The genetic features of pediatric low-grade gliomas, high-grade gliomas, medulloblastomas, and ependymomas are reviewed; differences between pediatric and adult gliomas are highlighted; and the critical oncogenic pathways of each tumor group are described. The role of the mitogen-activated protein kinase pathway in pediatric low-grade gliomas and of histone mutations as epigenetic regulators in pediatric high-grade gliomas is emphasized. In addition, the oncogenic drivers responsible for medulloblastoma, the classification of ependymomas, and the associated imaging correlations and clinical implications are discussed. ^©RSNA, 2018.

Pretreatment texture analysis of routine MR images and shape analysis of the diffusion tensor for prediction of volumetric response after radiosurgery for meningioma

OBJECTIVEThe goal of this study was to identify parameters from routine T1- and T2-weighted MR sequences and diffusion tensor imaging (DTI) that best predict the volumetric changes in a meningioma after treatment with Gamma Knife radiosurgery (GKRS).METHODSIn 32 patients with meningioma, routine MRI and DTI data were measured before GKRS. A total of 78 parameters derived from first-level texture analysis of the pretreatment MR images, including calculation of the mean, SD, 2.5th and 97.5th percentiles, and kurtosis and skewness of data in histograms on a voxel-wise basis, were correlated with lesion volume change after a mean follow-up period of 3 years (range 19.5-63.3 months).RESULTSSeveral DTI-derived parameters correlated significantly with a meningioma volume change. The parameter that best predicted the results of GKRS was the 2.5th percentile value of the smallest eigenvalue (L3) of the diffusion tensor (correlation coefficient 0.739, p ≤ 0.001), whereas among the non-DTI parameters, only the SD of T2-weighted images correlated significantly with a tumor volume change (correlation coefficient 0.505, p ≤ 0.05, after correction for family-wise errors using false-detection-rate correction).CONCLUSIONSDTI-derived data had a higher correlation to shrinkage of meningioma volume after GKRS than data from T1- and T2-weighted image sequences. However, if only routine MR images are available, the SD of T2-weighted images can be used to predict control or possible progression of a meningioma after GKRS.

Multicenter study demonstrates radiomic features derived from magnetic resonance perfusion images identify pseudoprogression in glioblastoma

Pseudoprogression (PsP) is a diagnostic clinical dilemma in cancer. In this study, we retrospectively analyse glioblastoma patients, and using their dynamic susceptibility contrast and dynamic contrast-enhanced perfusion MRI images we build a classifier using radiomic features obtained from both Ktrans and rCBV maps coupled with support vector machines. We achieve an accuracy of 90.82% (area under the curve (AUC) = 89.10%, sensitivity = 91.36%, 67 specificity = 88.24%, p = 0.017) in differentiating between pseudoprogression (PsP) and progressive disease (PD). The diagnostic performances of the models built using radiomic features from Ktrans and rCBV separately were equally high (Ktrans: AUC = 94%, 69 p = 0.012; rCBV: AUC = 89.8%, p = 0.004). Thus, this MR perfusion-based radiomic model demonstrates high accuracy, sensitivity and specificity in discriminating PsP from PD, thus provides a reliable alternative for noninvasive identification of PsP versus PD at the time of clinical/radiologic question. This study also illustrates the successful application of radiomic analysis as an advanced processing step on different MR perfusion maps.

MRI scans of glioblastoma patients can be misleading and some patients appear to show features of progressive disease although they respond to treatment. Here, the authors use MRI images of progressive disease or pseudoprogression and build a classifier using machine learning to distinguish the two.

Machine Learning-Based Radiomics Predicts Radiotherapeutic Response in Patients With Acromegaly

Background: Prediction of radiotherapeutic response before radiotherapy could help determine individual treatment strategies for patients with acromegaly. Objective: To develop and validate a machine-learning-based multiparametric MRI radiomics model to non-invasively predict radiotherapeutic response in patients with acromegaly. Methods: This retrospective study included 57 acromegaly patients who underwent postoperative radiotherapy between January 2008 and January 2016. Manual lesion segmentation and radiomics analysis were performed on each pituitary adenoma, and 1561 radiomics features were extracted from each sequence. A radiomics signature was built with a support vector machine using leave-one-out cross-validation for feature selection. Multivariable logistic regression analysis was used to select appropriate clinicopathological features to construct a clinical model, which was then combined with the radiomics signature to construct a radiomics model. The performance of this radiomic model was assessed using receiver operating characteristics (ROC) analysis and its calibration, discriminating ability, clinical usefulness. Results: At 3-years after radiotherapy, 25 patients had achieved remission and 32 patients had not. The clinical model incorporating seven clinical features had an area under the ROC (AUC) of 0.86 for predicting radiotherapeutic response, and performed better than any single clinical feature. The radiomics signature constructed with six radiomics features had a significantly higher AUC of 0.92. The radiomics model showed good discrimination abilities and calibration, with an AUC of 0.96. Decision curve analysis confirmed the clinical utility of the radiomics model. Conclusion: Using pre-radiotherapy clinical and MRI data, we developed a radiomics model with favorable performance for individualized non-invasive prediction of radiotherapeutic response, which may help in identifying acromegaly patients who are likely to benefit from radiotherapy.

Prediction of successful shock wave lithotripsy with CT: a phantom study using texture analysis

OBJECTIVE

To apply texture analysis (TA) in computed tomography (CT) of urinary stones and to correlate TA findings with the number of required shockwaves for successful shock wave lithotripsy (SWL).

MATERIALS AND METHODS

CT was performed on thirty-four urinary stones in an in vitro setting. Urinary stones underwent SWL and the number of required shockwaves for disintegration was recorded. TA was performed after post-processing for pixel spacing and image normalization. Feature selection and dimension reduction were performed according to inter- and intrareader reproducibility and by evaluating the predictive ability of the number of shock waves with the degree of redundancy between TA features. Three regression models were tested: (1) linear regression with elimination of colinear attributes (2), sequential minimal optimization regression (SMOreg) employing machine learning, and (3) simple linear regression model of a single TA feature with lowest squared error.

RESULTS

Highest correlations with the absolute number of required SWL shockwaves were found for the linear regression model (r = 0.55, p = 0.005) using two weighted TA features: Histogram 10th Percentile, and Gray-Level Co-Occurrence Matrix (GLCM) S(3, 3) SumAverg. Using the median number of required shockwaves (n = 72) as a threshold, receiver-operating characteristic analysis showed largest area-under-the-curve values for the SMOreg model (AUC = 0.84, r = 0.51, p < 0.001) using four weighted TA features: Histogram 10^th Percentile, and GLCM S(1, 1) InvDfMom, S(3, 3) SumAverg, and S(4, -4) SumVarnc.

CONCLUSION

Our in vitro study illustrates the proof-of-principle of TA of urinary stone CT images for predicting the success of stone disintegration with SWL.

Novel Quantitative Imaging for Predicting Response to Therapy: Techniques and Clinical Applications

The current standard of Response Evaluation Criteria in Solid Tumors (RECIST)-based tumor response evaluation is limited in its ability to accurately monitor treatment response. Radiomics, an approach involving computerized extraction of several quantitative imaging features, has shown promise in predicting as well as monitoring response to therapy. In this article, we provide a brief overview of radiomic approaches and the various analytical methods and techniques, specifically in the context of predicting and monitoring treatment response for non-small cell lung cancer (NSCLC). We briefly summarize some of the various types of radiomic features, including tumor shape and textural patterns, both within the tumor and within the adjacent tumor microenvironment. Additionally, we also discuss work in delta-radiomics or change in radiomic features (e.g., texture within the nodule) across longitudinally interspersed images in time for monitoring changes in therapy. We discuss the utility of these approaches for NSCLC, specifically the role of radiomics as a prognostic marker for treatment effectiveness and early therapy response, including chemoradiation, immunotherapy, and trimodality therapy.