Radiomics-based Machine Learning Model Research Articles

MRI T2w Radiomics-Based Machine Learning Models in Imaging Simulated Biopsy Add Diagnostic Value to PI-RADS in Predicting Prostate Cancer: A Retrospective Diagnostic Study.

Currently, prostate cancer (PCa) prebiopsy medical image diagnosis mainly relies on mpMRI and PI-RADS scores. However, PI-RADS has its limitations, such as inter- and intra-radiologist variability and the potential for imperceptible features. The primary objective of this study is to evaluate the effectiveness of a machine learning model based on radiomics analysis of MRI T2-weighted (T2w) images for predicting PCa in prebiopsy cases. A retrospective analysis was conducted using 820 lesions (363 cases, 457 controls) from The Cancer Imaging Archive (TCIA) Database for model development and validation. An additional 83 lesions (30 cases, 53 controls) from Hong Kong Queen Mary Hospital were used for independent external validation. The MRI T2w images were preprocessed, and radiomic features were extracted. Feature selection was performed using Cross Validation Least Angle Regression (CV-LARS). Using three different machine learning algorithms, a total of 18 prediction models and 3 shape control models were developed. The performance of the models, including the area under the curve (AUC) and diagnostic values such as sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV), were compared to the PI-RADS scoring system for both internal and external validation. All the models showed significant differences compared to the shape control model (all p < 0.001, except SVM model PI-RADS+2 Features p = 0.004, SVM model PI-RADS+3 Features p = 0.002). In internal validation, the best model, based on the LR algorithm, incorporated 3 radiomic features (AUC = 0.838, sensitivity = 76.85%, specificity = 77.36%). In external validation, the LR (3 features) model outperformed PI-RADS in predictive value with AUC 0.870 vs. 0.658, sensitivity 56.67% vs. 46.67%, specificity 92.45% vs. 84.91%, PPV 80.95% vs. 63.64%, and NPV 79.03% vs. 73.77%. The machine learning model based on radiomics analysis of MRI T2w images, along with simulated biopsy, provides additional diagnostic value to the PI-RADS scoring system in predicting PCa.

Commentary on “develop a radiomics-based machine learning model to predict the stone-free rate post-percutaneous nephrolithotomy”

Commentary on “develop a radiomics-based machine learning model to predict the stone-free rate post-percutaneous nephrolithotomy”

Identification of patients with internet gaming disorder via a radiomics-based machine learning model of subcortical structures in high-resolution T1-weighted MRI

It is of vital importance to establish an objective and reliable model to facilitate the early diagnosis and intervention of internet gaming disorder (IGD). A total of 133 patients with IGD and 110 healthy controls (HCs) were included. We extracted radiomic features of subcortical structures in high-resolution T1-weighted MRI. Different combinations of four feature selection methods (analysis of variance, Kruskal–Wallis, recursive feature elimination and relief) and ten classification algorithms were used to identify the most robust combined models for distinguishing IGD patients from HCs. Furthermore, a nomogram incorporating radiomic signatures and independent clinical factors was developed. Calibration curve and decision curve analyses were used to evaluate the nomogram. The combination of analysis of variance selector and logistic regression classifier identified that the radiomic model constructed with 20 features from the right caudate nucleus and amygdala showed better IGD screening performance. The radiomic model produced good areas under the curves (AUCs) in the training, validation and test cohorts (AUCs of 0.961, 0.903 and 0.895, respectively). In addition, sex, internet addiction test scores and radiomic scores were included in the nomogram as independent risk factors for IGD. Analysis of the correction curve and decision curve showed that the clinical-radiomic model has good reliability (C-index: 0.987). The nomogram incorporating radiomic features of subcortical structures and clinical characteristics achieved satisfactory classification performance and could serve as an effective tool for distinguishing IGD patients from HCs.

Preoperatively predicting survival outcome for clinical stage IA pure-solid non–small cell lung cancer by radiomics-based machine learning

ObjectiveClinical stage IA non–small cell lung cancer (NSCLC) showing a pure-solid appearance on computed tomography is associated with a worse prognosis. This study aimed to develop and validate machine-learning models using preoperative clinical and radiomic features to predict overall survival (OS) in clinical stage IA pure-solid NSCLC. MethodsPatients who underwent lung resection for NSCLC between January 2012 and December 2020 were reviewed. The radiomic features were extracted from the intratumoral and peritumoral regions on computed tomography. The machine-learning models were developed using random survival forest and eXtreme Gradient Boosting (XGBoost) algorithms, whereas the Cox regression model was set as a benchmark. Model performance was assessed using the integrated time-dependent area under the curve (iAUC) and validated by 5-fold cross-validation. ResultsIn total, 642 patients with clinical stage IA pure-solid NSCLC were included. Among 3748 radiomic and 34 preoperative clinical features, 42 features were selected. Both machine-learning models outperformed the Cox regression model (iAUC, 0.753; 95% confidence interval [CI], 0.629-0.829). The XGBoost model showed a better performance (iAUC, 0.832; 95% CI, 0.779-0.880) than the random survival forest model (iAUC, 0.795; 95% CI, 0.734-0.856). The XGBoost model showed an excellent survival stratification performance with a significant OS difference among the low-risk (5-year OS, 100.0%), moderate low-risk (5-year OS, 88.5%), moderate high-risk (5-year OS, 75.6%), and high-risk (5-year OS, 41.7%) groups (P < .0001). ConclusionsA radiomics-based machine-learning model can preoperatively and accurately predict OS and improve survival stratification in clinical stage IA pure-solid NSCLC.

Develop a radiomics-based machine learning model to predict the stone-free rate post-percutaneous nephrolithotomy.

Radiomics and machine learning have been extensively utilized in the realm of urinary stones, particularly in forecasting stone treatment outcomes. The objective of this study was to integrate clinical variables and radiomic features to develop a machine learning model for predicting the stone-free rate (SFR) following percutaneous nephrolithotomy (PCNL).A total of 212 eligible patients who underwent PCNL surgery at the Second Affiliated Hospital of Nanchang University were included in a retrospective analysis. Preoperative clinical variables and non-contrast-enhanced CT images of all patients were collected, and radiomic features were extracted after delineating the stone ROI. Univariate analysis was conducted to identify clinical variables strongly correlated with the stone-free rate after PCNL, and the least absolute shrinkage and selection operator algorithm (lasso regression) was utilized to screen radiomic features. Four supervised machine learning algorithms, including Logistic Regression, Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Gradient Boosting Decision Tree (GBDT), were employed. The clinical variables with strong correlation and screened radiomic features were integrated into the four machine learning algorithms to construct a prediction model, and the receiver operating curve was plotted. The area under the receiver operating curve (AUC), the accuracy rate, the specificity, etc., were used to evaluate the predictive performance of the four models.After analyzing postoperative statistics, the stone-free rate following the procedure was found to be 70.3% (n = 149). Among the various clinical variables examined, factors, such as stone number, stone diameter, stone CT value, stone location, and history of stone surgery, were identified as statistically significant in relation to the stone-free rate after PCNL. A total of 121 radiomic features were extracted, and through lasso regression, 7 features most closely associated with the stone-free rate post-PCNL were identified. The predictive accuracy of different models (Logistic Regression, RF, XGBoost, and GBDT) for determining the stone-free rate after PCNL was evaluated, yielding accuracies of 78.1%, 76.6%, 75.0%, and 73.4%, respectively. The corresponding area under the curve AUC (95%CI) were 0.85 (0.83-0.89), 0.81 (0.76-0.85), 0.82 (0.78-0.85), and 0.77 (0.73-0.81), positioning these models among the top performers in logistic regression prediction. In terms of predictive importance scores, the key factors identified by the logistic regression model were number of stone, zone percentage, stone diameter, and surface area. Similarly, the RF model highlighted number of stone, stone CT value, stone diameter, and surface area as the top predictors.Among the four machine learning models, the logistic regression model demonstrated the highest accuracy and discrimination ability in predicting the stone-free rate following PCNL. In comparison to XGBoost and GBDT, RF also exhibited superior accuracy and a certain level of discrimination ability. However, based on the performance of all four models, logistic regression is more likely to aid in clinical decision-making by assisting clinicians in diagnosing PCNL in patients. This enables us to effectively predict the presence of residual stones post-surgery and ultimately select patients who are suitable candidates for PCNL.

Machine learning techniques based on 18F-FDG PET radiomics features of temporal regions for the classification of temporal lobe epilepsy patients from healthy controls.

This study aimed to investigate the clinical application of 18F-FDG PET radiomics features for temporal lobe epilepsy and to create PET radiomics-based machine learning models for differentiating temporal lobe epilepsy (TLE) patients from healthy controls. A total of 347 subjects who underwent 18F-FDG PET scans from March 2014 to January 2020 (234 TLE patients: 25.50 ± 8.89 years, 141 male patients and 93 female patients; and 113 controls: 27.59 ± 6.94 years, 48 male individuals and 65 female individuals) were allocated to the training (n = 248) and test (n = 99) sets. All 3D PET images were registered to the Montreal Neurological Institute template. PyRadiomics was used to extract radiomics features from the temporal regions segmented according to the Automated Anatomical Labeling (AAL) atlas. The least absolute shrinkage and selection operator (LASSO) and Boruta algorithms were applied to select the radiomics features significantly associated with TLE. Eleven machine-learning algorithms were used to establish models and to select the best model in the training set. The final radiomics features (n = 7) used for model training were selected through the combinations of the LASSO and the Boruta algorithms with cross-validation. All data were randomly divided into a training set (n = 248) and a testing set (n = 99). Among 11 machine-learning algorithms, the logistic regression (AUC 0.984, F1-Score 0.959) model performed the best in the training set. Then, we deployed the corresponding online website version (https://wane199.shinyapps.io/TLE_Classification/), showing the details of the LR model for convenience. The AUCs of the tuned logistic regression model in the training and test sets were 0.981 and 0.957, respectively. Furthermore, the calibration curves demonstrated satisfactory alignment (visually assessed) for identifying the TLE patients. The radiomics model from temporal regions can be a potential method for distinguishing TLE. Machine learning-based diagnosis of TLE from preoperative FDG PET images could serve as a useful preoperative diagnostic tool.

Radiomics-Based Machine Learning Model for Diagnosis of Acute Pancreatitis Using Computed Tomography.

In the early diagnostic workup of acute pancreatitis (AP), the role of contrast-enhanced CT is to establish the diagnosis in uncertain cases, assess severity, and detect potential complications like necrosis, fluid collections, bleeding or portal vein thrombosis. The value of texture analysis/radiomics of medical images has rapidly increased during the past decade, and the main focus has been on oncological imaging and tumor classification. Previous studies assessed the value of radiomics for differentiating between malignancies and inflammatory diseases of the pancreas as well as for prediction of AP severity. The aim of our study was to evaluate an automatic machine learning model for AP detection using radiomics analysis. Patients with abdominal pain and contrast-enhanced CT of the abdomen in an emergency setting were retrospectively included in this single-center study. The pancreas was automatically segmented using TotalSegmentator and radiomics features were extracted using PyRadiomics. We performed unsupervised hierarchical clustering and applied the random-forest based Boruta model to select the most important radiomics features. Important features and lipase levels were included in a logistic regression model with AP as the dependent variable. The model was established in a training cohort using fivefold cross-validation and applied to the test cohort (80/20 split). From a total of 1012 patients, 137 patients with AP and 138 patients without AP were included in the final study cohort. Feature selection confirmed 28 important features (mainly shape and first-order features) for the differentiation between AP and controls. The logistic regression model showed excellent diagnostic accuracy of radiomics features for the detection of AP, with an area under the curve (AUC) of 0.932. Using lipase levels only, an AUC of 0.946 was observed. Using both radiomics features and lipase levels, we showed an excellent AUC of 0.933 for the detection of AP. Automated segmentation of the pancreas and consecutive radiomics analysis almost achieved the high diagnostic accuracy of lipase levels, a well-established predictor of AP, and might be considered an additional diagnostic tool in unclear cases. This study provides scientific evidence that automated image analysis of the pancreas achieves comparable diagnostic accuracy to lipase levels and might therefore be used in the future in the rapidly growing era of AI-based image analysis.

The value of MRI-based radiomics for evaluating early parotid gland injury in primary Sjögren's syndrome.

This study aimed to evaluate the value of machine learning models (ML) based on MRI radiomics in diagnosing early parotid gland injury in primary Sjögren's syndrome (pSS). A total of 164 patients (114 in the training cohort and 50 in the testing cohort) with pSS (n=82) or healthy controls (HC) (n=82) were enrolled. Itksnap software was used to perform two-dimensional segmentation of the bilateral parotid glands on T1-weighted (T1WI) and fat-suppressed T2-weighted imaging (fs-T2WI) images. A total of 1548 texture features of the parotid glands were extracted using radiomics software. A radiomics score (Radscore) was constructed and calculated. A t-test was used to compare the Radscore between the two groups. Finally, five machine learning models were trained and tested to identify early pSS parotid injury, and the performance of the machine learning models was evaluated by calculating the acceptance operating curve (ROC) and other parameters. The Radscores between the pSS and HC groups showed significant statistical differences (p<0.001). Among the five machine learning models, the Extra Trees Classifier (ETC) model performed high predictive efficacy in identifying early pSS parotid injury, with an AUC of 0.87 in the testing set. MRI radiomics-based machine learning models can effectively diagnose early parotid gland injury in primary Sjögren's syndrome.

Comparison of MRI Sequences to Predict IDH Mutation Status in Gliomas Using Radiomics-Based Machine Learning.

Regarding the 2021 World Health Organization (WHO) classification of central nervous system (CNS) tumors, the isocitrate dehydrogenase (IDH) mutation status is one of the most important factors for CNS tumor classification. The aim of our study is to analyze which of the commonly used magnetic resonance imaging (MRI) sequences is best suited to obtain this information non-invasively using radiomics-based machine learning models. We developed machine learning models based on different MRI sequences and determined which of the MRI sequences analyzed yields the highest discriminatory power in predicting the IDH mutation status. In our retrospective IRB-approved study, we used the MRI images of 106 patients with histologically confirmed gliomas. The MRI images were acquired using the T1 sequence with and without administration of a contrast agent, the T2 sequence, and the Fluid-Attenuated Inversion Recovery (FLAIR) sequence. To objectively compare performance in predicting the IDH mutation status as a function of the MRI sequence used, we included only patients in our study cohort for whom MRI images of all four sequences were available. Seventy-one of the patients had an IDH mutation, and the remaining 35 patients did not have an IDH mutation (IDH wild-type). For each of the four MRI sequences used, 107 radiomic features were extracted from the corresponding MRI images by hand-delineated regions of interest. Data partitioning into training data and independent test data was repeated 100 times to avoid random effects associated with the data partitioning. Feature preselection and subsequent model development were performed using Random Forest, Lasso regression, LDA, and Naïve Bayes. The performance of all models was determined with independent test data. Among the different approaches we examined, the T1-weighted contrast-enhanced sequence was found to be the most suitable for predicting IDH mutations status using radiomics-based machine learning models. Using contrast-enhanced T1-weighted MRI images, our seven-feature model developed with Lasso regression achieved a mean area under the curve (AUC) of 0.846, a mean accuracy of 0.792, a mean sensitivity of 0.847, and a mean specificity of 0.681. The administration of contrast agents resulted in a significant increase in the achieved discriminatory power. Our analyses show that for the prediction of the IDH mutation status using radiomics-based machine learning models, among the MRI images acquired with the commonly used MRI sequences, the contrast-enhanced T1-weighted images are the most suitable.

Prediction of initial objective response to drug-eluting beads transcatheter arterial chemoembolization for hepatocellular carcinoma using CT radiomics-based machine learning model.

Objective: A prognostic model utilizing CT radiomics, radiological, and clinical features was developed and validated in this study to predict an objective response to initial transcatheter arterial chemoembolization with drug-eluting beads (DEB-TACE) for hepatocellular carcinoma (HCC). Methods: Between January 2017 and December 2022, the baseline clinical characteristics and preoperative and postoperative follow-up imaging data of 108 HCC patients who underwent the first time treatment of DEB-TACE were analyzed retrospectively. The training group (n = 86) and the validation group (n = 22) were randomly assigned in an 8:2 ratio. By logistic regression in machine learning, radiomics, and clinical-radiological models were constructed separately. Finally, the integrated model construction involved the integration of both radiomics and clinical-radiological signatures. The study compared the integrated model with radiomics and clinical-radiological models using calibration curves, receiver operating characteristic (ROC) curves, and decision curve analysis (DCA). Results: The objective response rate observed in a group of 108 HCC patients who received initial DEB-TACE treatment was found to be 51.9%. Among the three models, the integrated model exhibited superior predictive accuracy in both the training and validation groups. The training group resulted in an area under the curve (AUC) of 0.860, along with sensitivity and specificity values of 0.650 and 0.913, respectively. Based on the findings from the validation group, the AUC was estimated to be 0.927. Additionally, it was found that values of sensitivity and specificity were 0.875 and 0.833, respectively. In the validation group, the AUC of the integrated model showed a significant improvement when contrasted to the clinical-radiological model (p = 0.042). Nevertheless, no significant distinction was observed in the AUC when comparing the integrated model with the radiomics model (p = 0.734). The DCA suggested that the integrated model demonstrates advantageous clinical utility. Conclusion: The integrated model, which combines the CT radiomics signature and the clinical-radiological signature, exhibited higher predictive efficacy than either the radiomics or clinical-radiological models alone. This suggests that during the prediction of the objective responsiveness of HCC patients to the first DEB-TACE treatment, the integrated model yields superior outcomes.

Machine learning-based radiomics strategy for prediction of acquired EGFR T790M mutation following treatment with EGFR-TKI in NSCLC

The epidermal growth factor receptor (EGFR) Thr790 Met (T790M) mutation is responsible for approximately half of the acquired resistance to EGFR-tyrosine kinase inhibitor (TKI) in non-small-cell lung cancer (NSCLC) patients. Identifying patients at diagnosis who are likely to develop this mutation after first- or second-generation EGFR-TKI treatment is crucial for better treatment outcomes. This study aims to develop and validate a radiomics-based machine learning (ML) approach to predict the T790M mutation in NSCLC patients at diagnosis. We collected retrospective data from 210 positive EGFR mutation NSCLC patients, extracting 1316 radiomics features from CT images. Using the LASSO algorithm, we selected 10 radiomics features and 2 clinical features most relevant to the mutations. We built models with 7 ML approaches and assessed their performance through the receiver operating characteristic (ROC) curve. The radiomics model and combined model, which integrated radiomics features and relevant clinical factors, achieved an area under the curve (AUC) of 0.80 (95% confidence interval [CI] 0.79–0.81) and 0.86 (0.87–0.88), respectively, in predicting the T790M mutation. Our study presents a convenient and noninvasive radiomics-based ML model for predicting this mutation at the time of diagnosis, aiding in targeted treatment planning for NSCLC patients with EGFR mutations.

Identifying genetic mutation status in patients with colorectal liver metastases using radiomics based machine learning models

Identifying genetic mutation status in patients with colorectal liver metastases using radiomics based machine learning models

MRI radiomics based machine learning model of the periaqueductal gray matter in migraine patients.

<p>The aim of the study was to investigate the question: Can MRI radiomics analysis of the periaqueductal gray region elucidate the pathophysiological mechanisms underlying various migraine subtypes, and can a machine learning model using these radiomics features accurately differentiate between migraine patients and healthy individuals, as well as between migraine subtypes, including atypical cases with overlapping symptoms?</p>. <p>The study analyzed initial MRI images of individuals taken after their first migraine diagnosis, and additional MRI scans were acquired from healthy subjects. Radiomics modeling was applied to analyze all the MRI images in the periaqueductal gray region. The dataset was randomized, and oversampling was used if there was class imbalance between groups. The optimal algorithm-based feature selection method was employed to select the most important 5-10 features to differentiate between the two groups. The classification performance of AI algorithms was evaluated using receiver operating characteristic analysis to calculate the area under the curve, classification accuracy, sensitivity, and specificity values. Participants were required to have a confirmed diagnosis of either episodic migraine, probable migraine, or chronic migraine. Patients with aura, those who used migraine-preventive medication within the past six months, or had chronic illnesses, psychiatric disorders, cerebrovascular conditions, neoplastic diseases, or other headache types were excluded from the study. Additionally, 102 healthy subjects who met the inclusion and exclusion criteria were included.&nbsp;</p>. <p>The algorithm-based information gain method for feature reduction had the best performance among all methods, with the first-order, gray-level size zone matrix, and gray-level co-occurrence matrix classes being the dominant feature classes. The machine learning model correctly classified 82.4% of migraine patients from healthy subjects. Within the migraine group, 74.1% of the episodic migraine-probable migraine patients and 90.5% of the chronic migraine patients were accurately classified. No significant difference was found between probable migraine and episodic migraine patients in terms of the periaqueductal gray region radiomics features. The kNN algorithm showed the best performance for classifying episodic migraine-probable migraine subtypes, while the Random Forest algorithm demonstrated the best performance for classifying the migraine group and chronic migraine subtype.</p>. <p>A radiomics-based machine learning model, utilizing standard MR images obtained during the diagnosis and follow-up of migraine patients, shows promise not only in aiding migraine diagnosis and classification for clinical approach, but also in understanding the neurological mechanisms underlying migraines.&nbsp;</p>.

Machine learning model based on radiomics features for AO/OTA classification of pelvic fractures on pelvic radiographs.

Depending on the degree of fracture, pelvic fracture can be accompanied by vascular damage, and in severe cases, it may progress to hemorrhagic shock. Pelvic radiography can quickly diagnose pelvic fractures, and the Association for Osteosynthesis Foundation and Orthopedic Trauma Association (AO/OTA) classification system is useful for evaluating pelvic fracture instability. This study aimed to develop a radiomics-based machine-learning algorithm to quickly diagnose fractures on pelvic X-ray and classify their instability. data used were pelvic anteroposterior radiographs of 990 adults over 18 years of age diagnosed with pelvic fractures, and 200 normal subjects. A total of 93 features were extracted based on radiomics:18 first-order, 24 GLCM, 16 GLRLM, 16 GLSZM, 5 NGTDM, and 14 GLDM features. To improve the performance of machine learning, the feature selection methods RFE, SFS, LASSO, and Ridge were used, and the machine learning models used LR, SVM, RF, XGB, MLP, KNN, and LGBM. Performance measurement was evaluated by area under the curve (AUC) by analyzing the receiver operating characteristic curve. The machine learning model was trained based on the selected features using four feature-selection methods. When the RFE feature selection method was used, the average AUC was higher than that of the other methods. Among them, the combination with the machine learning model SVM showed the best performance, with an average AUC of 0.75±0.06. By obtaining a feature-importance graph for the combination of RFE and SVM, it is possible to identify features with high importance. The AO/OTA classification of normal pelvic rings and pelvic fractures on pelvic AP radiographs using a radiomics-based machine learning model showed the highest AUC when using the SVM classification combination. Further research on the radiomic features of each part of the pelvic bone constituting the pelvic ring is needed.

Predicting T Cell-Inflamed Gene Expression Profile in Hepatocellular Carcinoma Based on Dynamic Contrast-Enhanced Ultrasound Radiomics.

The T cell-inflamed gene expression profile (GEP) quantifies 18 genes' expression indicative of a T-cell immune tumor microenvironment, playing a crucial role in the immunotherapy of hepatocellular carcinoma (HCC). Our study aims to develop a radiomics-based machine learning model using contrast-enhanced ultrasound (CEUS) for predicting T cell-inflamed GEP in HCC. The primary cohort of HCC patients with preoperative CEUS and RNA sequencing data of tumor tissues at the single center was used to construct the model. A total of 5936 radiomics features were extracted from the regions of interest in representative images of each phase, and the least absolute shrinkage and selection operator and logistic regression were used to construct four models including three phase-specific models and an integrated model. The area under the curve (AUC) was calculated to evaluate the performance of the model. The independent cohort of HCC patients with preoperative CEUS and Immunoscore based on immunohistochemistry and digital pathology was used to validate the correlation between model prediction value and T-cell infiltration. There were 268 patients enrolled in the primary cohort and 46 patients enrolled in the independent cohort. Compared with the other three models, the AP model constructed by 36 arterial phase (AP) features showed good performance with a mean AUC of 0.905 in the 5-fold cross-validation and was easier to apply in the clinical setting. The decision curve and calibration curve confirmed the clinical utility of the model. In the independent cohort, patients with high Immunoscores showed significantly higher GEP prediction values than those with low Immunoscores (t=-2.359, p=0.029). The CEUS-based model is a reliable predictive tool for T cell-inflamed GEP in HCC, and might facilitate individualized immunotherapy decision-making.

Identifying Genetic Mutation Status in Patients with Colorectal Cancer Liver Metastases Using Radiomics-Based Machine-Learning Models

For patients with colorectal cancer liver metastases (CRLM), the genetic mutation status is important in treatment selection and prognostication for survival outcomes. This study aims to investigate the relationship between radiomics imaging features and the genetic mutation status (KRAS mutation versus no mutation) in a large multicenter dataset of patients with CRLM and validate these findings in an external dataset. Patients with initially unresectable CRLM treated with systemic therapy of the randomized controlled CAIRO5 trial (NCT02162563) were included. All CRLM were semi-automatically segmented in pre-treatment CT scans and radiomics features were calculated from these segmentations. Additionally, data from the Netherlands Cancer Institute (NKI) were used for external validation. A total of 255 patients from the CAIRO5 trial were included. Random Forest, Gradient Boosting, Gradient Boosting + LightGBM, and Ensemble machine-learning classifiers showed AUC scores of 0.77 (95%CI 0.62–0.92), 0.77 (95%CI 0.64–0.90), 0.72 (95%CI 0.57–0.87), and 0.86 (95%CI 0.76–0.95) in the internal test set. Validation of the models on the external dataset with 129 patients resulted in AUC scores of 0.47–0.56. Machine-learning models incorporating CT imaging features could identify the genetic mutation status in patients with CRLM with a good accuracy in the internal test set. However, in the external validation set, the models performed poorly. External validation of machine-learning models is crucial for the assessment of clinical applicability and should be mandatory in all future studies in the field of radiomics.

Predicting stereotactic radiosurgery outcomes with multi-observer qualitative appearance labelling versus MRI radiomics

Qualitative observer-based and quantitative radiomics-based analyses of T1w contrast-enhanced magnetic resonance imaging (T1w-CE MRI) have both been shown to predict the outcomes of brain metastasis (BM) stereotactic radiosurgery (SRS). Comparison of these methods and interpretation of radiomics-based machine learning (ML) models remains limited. To address this need, we collected a dataset of n = 123 BMs from 99 patients including 12 clinical features, 107 pre-treatment T1w-CE MRI radiomic features, and BM post-SRS progression scores. A previously published outcome model using SRS dose prescription and five-way BM qualitative appearance scoring was evaluated. We found high qualitative scoring interobserver variability across five observers that negatively impacted the model’s risk stratification. Radiomics-based ML models trained to replicate the qualitative scoring did so with high accuracy (bootstrap-corrected AUC = 0.84–0.94), but risk stratification using these replicated qualitative scores remained poor. Radiomics-based ML models trained to directly predict post-SRS progression offered enhanced risk stratification (Kaplan–Meier rank-sum p = 0.0003) compared to using qualitative appearance. The qualitative appearance scoring enabled interpretation of the progression radiomics-based ML model, with necrotic BMs and a subset of heterogeneous BMs predicted as being at high-risk of post-SRS progression, in agreement with current radiobiological understanding. Our study’s results show that while radiomics-based SRS outcome models out-perform qualitative appearance analysis, qualitative appearance still provides critical insight into ML model operation.

Optimization and validation of 18F-DCFPyL PET radiomics-based machine learning models in intermediate- to high-risk primary prostate cancer.

Radiomics extracted from prostate-specific membrane antigen (PSMA)-PET modeled with machine learning (ML) may be used for prediction of disease risk. However, validation of previously proposed approaches is lacking. We aimed to optimize and validate ML models based on 18F-DCFPyL-PET radiomics for the prediction of lymph-node involvement (LNI), extracapsular extension (ECE), and postoperative Gleason score (GS) in primary prostate cancer (PCa) patients. Patients with intermediate- to high-risk PCa who underwent 18F-DCFPyL-PET/CT before radical prostatectomy with pelvic lymph-node dissection were evaluated. The training dataset included 72 patients, the internal validation dataset 24 patients, and the external validation dataset 27 patients. PSMA-avid intra-prostatic lesions were delineated semi-automatically on PET and 480 radiomics features were extracted. Conventional PET-metrics were derived for comparative analysis. Segmentation, preprocessing, and ML methods were optimized in repeated 5-fold cross-validation (CV) on the training dataset. The trained models were tested on the combined validation dataset. Combat harmonization was applied to external radiomics data. Model performance was assessed using the receiver-operating-characteristics curve (AUC). The CV-AUCs in the training dataset were 0.88, 0.79 and 0.84 for LNI, ECE, and GS, respectively. In the combined validation dataset, the ML models could significantly predict GS with an AUC of 0.78 (p<0.05). However, validation AUCs for LNI and ECE prediction were not significant (0.57 and 0.63, respectively). Conventional PET metrics-based models had comparable AUCs for LNI (0.59, p>0.05) and ECE (0.66, p>0.05), but a lower AUC for GS (0.73, p<0.05). In general, Combat harmonization improved external validation AUCs (-0.03 to +0.18). In internal and external validation, 18F-DCFPyL-PET radiomics-based ML models predicted high postoperative GS but not LNI or ECE in intermediate- to high-risk PCa. Therefore, the clinical benefit seems to be limited. These results underline the need for external and/or multicenter validation of PET radiomics-based ML model analyses to assess their generalizability.

Radiomics Analyses to Predict Histopathology in Patients with Metastatic Testicular Germ Cell Tumors before Post-Chemotherapy Retroperitoneal Lymph Node Dissection.

The identification of histopathology in metastatic non-seminomatous testicular germ cell tumors (TGCT) before post-chemotherapy retroperitoneal lymph node dissection (PC-RPLND) holds significant potential to reduce treatment-related morbidity in young patients, addressing an important survivorship concern. To explore this possibility, we conducted a study investigating the role of computed tomography (CT) radiomics models that integrate clinical predictors, enabling personalized prediction of histopathology in metastatic non-seminomatous TGCT patients prior to PC-RPLND. In this retrospective study, we included a cohort of 122 patients. Using dedicated radiomics software, we segmented the targets and extracted quantitative features from the CT images. Subsequently, we employed feature selection techniques and developed radiomics-based machine learning models to predict histological subtypes. To ensure the robustness of our procedure, we implemented a 5-fold cross-validation approach. When evaluating the models' performance, we measured metrics such as the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, precision, and F-score. Our radiomics model based on the Support Vector Machine achieved an optimal average AUC of 0.945. The presented CT-based radiomics model can potentially serve as a non-invasive tool to predict histopathological outcomes, differentiating among fibrosis/necrosis, teratoma, and viable tumor in metastatic non-seminomatous TGCT before PC-RPLND. It has the potential to be considered a promising tool to mitigate the risk of over- or under-treatment in young patients, although multi-center validation is critical to confirm the clinical utility of the proposed radiomics workflow.

MRI Radiomics-Based Machine Learning Models for Ki67 Expression and Gleason Grade Group Prediction in Prostate Cancer.

The Ki67 index and the Gleason grade group (GGG) are vital prognostic indicators of prostate cancer (PCa). This study investigated the value of biparametric magnetic resonance imaging (bpMRI) radiomics feature-based machine learning (ML) models in predicting the Ki67 index and GGG of PCa. A total of 122 patients with pathologically proven PCa who had undergone preoperative MRI were retrospectively included. Radiomics features were extracted from T2-weighted imaging (T2WI), diffusion-weighted imaging (DWI), and apparent diffusion coefficient (ADC) maps. Then, recursive feature elimination (RFE) was applied to remove redundant features. ML models for predicting Ki67 expression and GGG were constructed based on bpMRI and different algorithms, including logistic regression (LR), support vector machine (SVM), random forest (RF), and K-nearest neighbor (KNN). The performances of different models were evaluated with receiver operating characteristic (ROC) analysis. In addition, a joint analysis of Ki67 expression and GGG was performed by assessing their Spearman correlation and calculating the diagnostic accuracy for both indices. The ML model based on LR and ADC + T2 (LR_ADC + T2, AUC = 0.8882) performed best in predicting Ki67 expression, and ADC_wavelet-LHH_firstorder_Maximum had the highest feature weighting. The SVM_DWI + T2 (AUC = 0.9248) performed best in predicting GGG, and DWI_wavelet HLL_glcm_SumAverage had the highest feature weighting. The Ki67 and GGG exhibited a weak positive correlation (r = 0.382, p < 0.001), and LR_ADC + DWI had the highest diagnostic accuracy in predicting both (0.6230). The proposed ML models are suitable for predicting both Ki67 expression and GGG in PCa. This algorithm could be used to identify indolent or invasive PCa with a noninvasive, repeatable, and accurate diagnostic method.