Keywords: Radiomics, Radiomics, Prostate, PSA

Motivation: In enlarged prostates, attributing increased PSA levels solely to benign hyperplasia is difficult, even with reassuring imaging features. This results in diagnostic ambiguity when PSA is elevated.

Goal(s): Create a radiomic signature for accurate PSA level prediction using only imaging characteristics

Approach: T2-weighted prostate images from 100 patients with PIRADS ≤2 were segmented into transitional and peripheral zones using 3D Slicer. Radiomic analysis of transitional zone segments identified MRI features associated with serum PSA.

Results: Principal component analysis identified one age-adjusted, independent predictor of PSA levels among seventeen radiomic features. This signature had a significant association with PSA (b=1.651, 95% CI: 1.07-2.24, p<0.001).

Impact: The predictive value of our component will allow physicians to identify PSA levels which are not representative of their imaging characteristics which may better inform the need for further investigation

Keywords: Radiomics, Liver, Radiomics, Segmentation

Motivation: MRI radiomics with T2-mapping has been proposed to detect changes of chronic liver disease (CLD) but may be hampered by liver vasculature.

Goal(s): Our goal is to evaluate radiomics features of T2 maps and the impact of vessel removal on CLD classification.

Approach: T2 liver maps from a clinical cohort were analyzed using radiomics features and a T2 thresholding-based feature, followed by feature selection and random forest classification. Experiments were conducted on both full liver T2 maps and T2 maps with vessel removal using the Frangi filter.

Results: Vessel removal significantly enhanced CLD classification performance as measured by mean AUC. 

Impact: Our findings have the potential for clinical translation to help establish MRI radiomics with T2-mapping for CLD diagnosis and treatment monitoring.

Keywords: Radiomics, Brain, Diagnosis/Prediction, Data Analysis

Motivation: Radiomic features can potentially help distinguish subtypes of IDH-mutant gliomas that appear similar on MRI.

Goal(s): The aim of this study was to evaluate whether imaging-based clustering of radiomic biomarkers of IDH-mutant gliomas may identify patterns or subgroups based on the 2021 CNS WHO classification.

Approach: Dimensionality reduction techniques were applied to radiomic features of 179 patients of different sequence combinations to analyze the high dimensional feature space. 

Results: FLAIR and T1 post-contrast imaging revealed unique clusters, and survival analysis suggested potential differences amongst clusters. However, further research with a larger dataset is needed to determine whether the observed differences are significant. 

Impact: This study analyzed quantitative imaging biomarkers to differentiate IDH-mutant gliomas according to the 2021 WHO classification. The findings suggest that radiomic features may hold insights into potential survival differences among subtypes. Larger-scale research is required to further investigate these findings. 

Keywords: Radiomics, Radiomics, rectal cancer;neoadjuvant therapy;overall survival

Motivation: Use MR radiomics features as a potential factor for predicting overall survival (OS) after neoadjuvant therap in rectal cancer.

Goal(s): To explore the ability of the intratumoral and peritumoral radiomics features to predict the OS.

Approach: A nomogram that combines clinical features, intratumor and peritumor radiomics features from T2WI using Machine Learning to predict OS was developed, validated using the Kaplan-Meier survival curve.

Results: There was a significant statistical difference in the label scores between the high-risk and low-risk groups divided by median survival time as the cutoff value. The C-index of the training and test cohort was 0.798, 0.772 respectively.

Impact: The radiomics model of peritumor radiomics features as the intratumor model can predict the OS in rectal cancer with neoadjuvant therapy. Combined with clinical features and the radiomics model of intratumor radiomics features, high accuracy can be obtained.

Keywords: Data Processing, Data Processing, Texture analysis, Cocaine use disorder

Motivation: Neuroimaging improves treatment response classification in cocaine use disorder (CUD), often relying on brain morphometry to track changes. Multiple studies report that MRI textural changes predict earlier changes than volumetric changes in Alzheimer’s disease. We aim to apply textural analysis to multi-timepoint CUD MRI data.

Goal(s): Our work investigates changes in the amygdala volume and texture in  CUD patients undergoing repetitive transcranial magnetic stimulation (rTMS). 

Approach:  We computed the volumetric and textural changes over time and compared the relative temporal sensitivities.

Results: The relative temporal sensitivity of textural changes in the amygdala is higher than volumetric changes in CUD patients undergoing rTMS.

Impact: Textural changes in MRI become noticeable before volumetric changes, offering additional insights into ongoing therapy complementary to clinical measures. This allows for personalized treatment, utilizing each individual's baseline data as their internal reference.

Keywords: Radiomics, Radiomics, Stroke

Motivation: Stroke is a major global health issue, necessitating early outcome prediction for optimal treatment. Brainstem stroke, often overlooked, requires dedicated predictive models due to its unique challenges.

Goal(s): Develop radiomics models to predict brainstem stroke outcomes, considering infarct edge and surrounding regions, improving prognosis, and simplifying clinical evaluation.

Approach: 474 patients were studied, and radiomics features were extracted from diffusion-weighted images. Machine learning models were trained using SVM, RF, KNN and AdaBoost algorithms.

Results: The RF model, based on the circle² region, exhibited the highest performance (AUC=0.84). Models in the circle region outperformed core.

Impact: Our specialized radiomics models offer a valuable tool for personalized brainstem stroke treatment planning, potentially enhancing patient outcomes.

Keywords: Radiomics, Liver

Motivation: Microvascular invasion (MVI) is a significant prognostic factor in intrahepatic cholangiocarcinoma (ICC).

Goal(s): This study aimed to develop an MR radiomics-based model for preoperative MVI status stratification and overall survival prediction.

Approach: Univariate and multivariate logistic regression identified independent clinical and imaging predictors. The radiomic model, utilizing robust features and a logistic regression classifier with the least absolute shrinkage and selection operator algorithm, was integrated into the imaging-radiomics (IR) model.

Results: The IR model demonstrated strong performance (AUCtraining=0.890, AUCvalidation=0.885, AUCtest=0.815), confirmed by calibration and decision curves, and its predicted survival closely matched histological methods.

Impact: The MRI-based radiomics models for preoperative MVI status and OS prediction holds the promise of enabling physicians to tailor medical regimens in ICC patients, optimizing their individual benefit.

Keywords: Analysis/Processing, Segmentation, Radiomics

Motivation: Machine learning (ML) models need to be periodically evaluated to combat ‘data drift’, where the target population changes over time

Goal(s): The goal is to present a framework for optimally selecting new datasets and updating ML models

Approach: We retrained a pancreas segmentation model in MRI scans. We selected 50 new cases to annotate using radiomics features, i.e., those unlabelled cases with most differing segmentations from those in the previous training set

Results: The system identified a diversity of failure cases, which flagged challenges in real-world data. The mean performance of the model improved after retraining with the additional cases

Impact: The proposed system yields a helpful guide for researchers and technicians for retraining machine learning models, particularly deep learning models for organ segmentation in MRI. Selecting an optimal new set of data to annotate produces time and cost savings

Keywords: Radiomics, Breast

Motivation: It's essential for breast and axillary conservation surgery decisions by accurately predicting axillary lymph node (ALN) status after neoadjuvant chemotherapy (NAC) in node-positive breast cancer.

Goal(s): We aimed to evaluate the performance of intratumor and peritumor radiomics signature from pretreatment DCE-MRI for predicting ALN pathologic complete response (pCR) after NAC in breast cancer patients.

Approach: We retrospectively collected 175 patients and integrated the clinicopathological features and DCE-MRI radiomics signature for prediction.

Results: The radiomics score was identified as one of independent predictor for ALN pCR. The factors in the optimal model include initial clinical N stage, ER status, HER2 status, and radiomics score.

Impact: The independent risk factors can provide valuable insights into the patients’ conditions before NAC. The nomogram model may help to identify the candidates who do not necessarily require ALN dissection, thereby facilitating personalized treatment strategies for node-positive breast cancer.

Keywords: Radiomics, Prostate, Repeatability, Optimization

Motivation: Radiomic feature extraction techniques may be combined with reduced k-space data acquisition, if their repeatability and clinical performance would stay in the same levels as with full data acquisition.

Goal(s): We evaluate radiomics for their potential to be used in keyhole imaging acquiring k-space only partially.

Approach: We utilized 78 patients with prostate cancer who underwent short-term test-retest prostate MRI examination. We calculated ADC parameter map with different portions of k-space, simulating keyhole acquisitions. We extracted radiomics, evaluating intra-class correlation coefficient ICC(3,1) changes and area under ROC curve (AUC).

Results: Repeatability and classification performance stayed in acceptable limits for some of the radiomics.

Impact: The technique is relative easy to implement, and thus may benefit clinical MR examinations in the near future.

Keywords: Analysis/Processing, Radiomics, glioma, machine learning, generalizability

Motivation: Isocitrate dehydrogenase (IDH) mutation plays a key role in the prognosis of gliomas. Several studies have detected the IDH mutation using radiomics. However, few studies focused on the generalizability of radiomics-based machine learning models.

Goal(s): To externally validate the ability of radiomics for noninvasive detection of IDH mutation using multi-site data.

Approach: Radiomics of T2w MRI of UCSF-PDGM dataset (Cohort 1) was used for training machine learning models, then externally validated at the local dataset (Cohort 2).

Results: T2w MRI-radiomics could identify the IDH mutation with an accuracy of 0.89 on Cohort 1, which was externally validated with an accuracy of  0.73.

Impact: External validation studies are important for investigating the generalizability of machine learning models. The models based on T2w MRI-radiomics resulted in 0.89 accuracy in the training dataset, with a slightly lower accuracy on external validation dataset for identifying IDH mutation.

Keywords: Radiomics, Radiomics, MRgRT, online adaptation

Motivation: MRI-guided radiotherapy (MRgRT) offers the advantage of superior soft-tissue image contrasts, particularly beneficial for daily online treatment plan adaptation in prostate cancer (PC). However, the decision between “adapt-to-position” (ATP) or “adapt-to-shape” (ATS) is subjective and complicated.

Goal(s): This study aims to use radiomics to predict the ATP or ATS adaptation for localized PC.

Approach: Daily MRI images from 210 fractions were included. 1023 radiomics features were extracted and used to build a logistic regression model for predicting ATP or ATS adaptations.

Results: The MRI radiomics model built was relatively good in objectively predicting ATP and ATS adaptations for MRgRT in localized PC.

Impact: Our study showed that MRI radiomics have promising predictive capabilities for determining online adaptation strategies for MRgRT in localized PC. This could enhance workflow efficiency and personalize care by providing quantitative and objective criteria for adaptation strategy determination in MRgRT.

Keywords: Radiomics, Tumor, nasopharyngeal carcinoma neoadjuvant chemotherapy

Motivation: Radiomics, as a non-invasive way for analysis of tumor heterogeneity, has been applied to the detection of therapeutic sensibility with satisfactory outcomes.

Goal(s): This study aimed to explore the predictive value of nomogram combing MRI-based radiomics and clinical factors in detecting tumor response to neoadjuvant therapy in nasopharyngeal carcinoma patients.

Approach: LASSO-logistic regression analysis was applied to select radiomics features. The “rms” package was used to construct nomogram and calibration curves. The ROC curves, calibration, and decision curves were performed to assess the performance of the models.

Results: The nomograms integrated radiomics scores with clinical factors outperformed the clinical-only or radiomics-only models.

Impact: The nomogram developed from MRI-based radiomics combined with clinical factors could serve as a reliable tool for non-invasively discriminating neoadjuvant chemotherapy responders from non-responders and provide a basis for personalized therapeutic regimens for LA-NPC patients.

Keywords: Radiomics, Radiomics

Motivation:  Pituitary adenomas (PAs) are a rare but clinically diverse group of tumors with varying hormone secretion profiles and clinical characteristics, comprising 15% of intracranial tumors. Typical classification of PAs relies on blood hormone levels as gold standard test, with a limited exploration into assessing hormone status using neuroimaging biomarkers.                                                                       

Goal(s): We aim to offer a practical MRI-based classification model, improving clinical PA management.
 
 

Approach: Our study developed a machine learning model using MRI radiomics as image biomarkers for the classification of PAs focusing on six subtypes.

Results: Our SVM model showed an accuracy of 0.65 based on MRI images. 

Impact: Our radiomics classification model promises to revolutionize MRI PA classification and diagnosis, enhancing clinical management and benefiting scientists, clinicians, and patients by enabling more accurate and efficient diagnostics and treatments.

Keywords: Analysis/Processing, Cancer, VASARI

Motivation: Diagnosis of molecular subtypes of IDH-mutant gliomas on MRI has presented a challenge in clinical practice. 

Goal(s): To classify IDH-mutant gliomas we compared different ML models using qualitative and quantitative features from preoperative MRI.

Approach: Three models were compared, using only qualitative VASARI features as scored by two blinded neuroradiologists, the other used quantitative features from FLAIR and T1Gd and finally combining both in a third model.

Results: The VASARI feature based model showed moderate diagnostic accuracies for different tumor entities, which was higher than the Radiomics only model. Combining both features improved results, emphasizing the importance of feature selection in clinical applications.

Impact: This study demonstrates the potential of machine learning models in enhancing the accuracy of IDH-mutant glioma classification on preoperative MRI images.

Keywords: Analysis/Processing, Data Processing, Harmonization

Motivation: Multi-center studies often suffer from non-biological variations due to different MRI scanners. This can adversely affect the comparability of MRI radiomic features and deep features.

Goal(s): This multi-center study aims to investigate the effectiveness of ComBat harmonization on radiomic and deep features from abdominal MRI data.

Approach: We retrieved 3,857 clinical T2-weighted MRI examinations of adult patients from three institutions. The ANOVA test and Cohen’s F score were applied as evaluation metrics.

Results: An average of 78.7% of radiomic features, and 74.9% of deep features had significant distribution differences. After ComBat harmonization, none of radiomic and deep features had significant difference.

Impact: This multi-center study showed that ComBat can effectively remove non-biological variations of radiomic and deep features from abdominal MRI studies acquired from different MRI scanners and institutions. Future multi-center studies should consider ComBat harmonization to improve data comparability.