Radiomics feature robustness as measured using an MRI phantom

Impact of harmonization on the reproducibility of MRI radiomic features when using different scanners, acquisition parameters, and image pre-processing techniques: a phantom study

This study investigated the impact of ComBat harmonization on the reproducibility of radiomic features extracted from magnetic resonance images (MRI) acquired on different scanners, using various data acquisition parameters and multiple image pre-processing techniques using a dedicated MRI phantom. Four scanners were used to acquire an MRI of a nonanatomic phantom as part of the TCIA RIDER database. In fast spin-echo inversion recovery (IR) sequences, several inversion durations were employed, including 50, 100, 250, 500, 750, 1000, 1500, 2000, 2500, and 3000 ms. In addition, a 3D fast spoiled gradient recalled echo (FSPGR) sequence was used to investigate several flip angles (FA): 2, 5, 10, 15, 20, 25, and 30 degrees. Nineteen phantom compartments were manually segmented. Different approaches were used to pre-process each image: Bin discretization, Wavelet filter, Laplacian of Gaussian, logarithm, square, square root, and gradient. Overall, 92 first-, second-, and higher-order statistical radiomic features were extracted. ComBat harmonization was also applied to the extracted radiomic features. Finally, the Intraclass Correlation Coefficient (ICC) and Kruskal-Wallis's (KW) tests were implemented to assess the robustness of radiomic features. The number of non-significant features in the KW test ranged between 0-5 and 29-74 for various scanners, 31-91 and 37-92 for three times tests, 0-33 to 34-90 for FAs, and 3-68 to 65-89 for IRs before and after ComBat harmonization, with different image pre-processing techniques, respectively. The number of features with ICC over 90% ranged between 0-8 and 6-60 for various scanners, 11-75 and 17-80 for three times tests, 3-83 to 9-84 for FAs, and 3-49 to 3-63 for IRs before and after ComBat harmonization, with different image pre-processing techniques, respectively. The use of various scanners, IRs, and FAs has a great impact on radiomic features. However, the majority of scanner-robust features is also robust to IR and FA. Among the effective parameters in MR images, several tests in one scanner have a negligible impact on radiomic features. Different scanners and acquisition parameters using various image pre-processing might affect radiomic features to a large extent. ComBat harmonization might significantly impact the reproducibility of MRI radiomic features.

Impact of SUSAN Denoising and ComBat Harmonization on Machine Learning Model Performance for Malignant Brain Neoplasms

BACKGROUND AND PURPOSE

Feature variability in radiomics studies due to technical and magnet strength parameters is well-known and may be addressed through various preprocessing methods. However, very few studies have evaluated the downstream impact of variable preprocessing on model classification performance in a multiclass setting. We sought to evaluate the impact of Smallest Univalue Segment Assimilating Nucleus (SUSAN) denoising and Combining Batches harmonization on model classification performance.

MATERIALS AND METHODS

A total of 493 cases (410 internal and 83 external data sets) of glioblastoma, intracranial metastatic disease, and primary CNS lymphoma underwent semiautomated 3D-segmentation post-baseline image processing (BIP) consisting of resampling, realignment, coregistration, skull-stripping, and image normalization. Post-BIP, 2 sets were generated, one with and another without SUSAN denoising. Radiomics features were extracted from both data sets and batch-corrected to produce 4 data sets: (a) BIP, (b) BIP with SUSAN denoising, (c) BIP with Combining Batches, and (d) BIP with both SUSAN denoising and Combining Batches harmonization. Performance was then summarized for models using a combination of 6 feature-selection techniques and 6 machine learning models across 4 mask-sequence combinations with features derived from 1 to 3 (multiparametric) MRI sequences.

RESULTS

Most top-performing models on the external test set used BIP+SUSAN denoising-derived features. Overall, the use of SUSAN denoising and Combining Batches harmonization led to a slight but generally consistent improvement in model performance on the external test set.

CONCLUSIONS

The use of image-preprocessing steps such as SUSAN denoising and Combining Batches harmonization may be more useful in a multi-institutional setting to improve model generalizability. Models derived from only T1 contrast-enhanced images showed comparable performance to models derived from multiparametric MRI.

Impact of MRI radiomic feature normalization for prognostic modelling in uterine endometrial and cervical cancers

Widespread clinical use of MRI radiomic tumor profiling for prognostication and treatment planning in cancers faces major obstacles due to limitations in standardization of radiomic features. The purpose of the current work was to assess the impact of different MRI scanning- and normalization protocols for the statistical analyses of tumor radiomic data in two patient cohorts with uterine endometrial-(EC) (n = 136) and cervical (CC) (n = 132) cancer. 1.5 T and 3 T, T1-weighted MRI 2 min post-contrast injection, T2-weighted turbo spin echo imaging, and diffusion-weighted imaging were acquired. Radiomic features were extracted from within manually segmented tumors in 3D and normalized either using z-score normalization or a linear regression model (LRM) accounting for linear dependencies with MRI acquisition parameters. Patients were clustered into two groups based on radiomic profile. Impact of MRI scanning parameters on cluster composition and prognostication were analyzed using Kruskal-Wallis tests, Kaplan-Meier plots, log-rank test, random survival forests and LASSO Cox regression with time-dependent area under curve (tdAUC) (α = 0.05). A large proportion of the radiomic features was statistically associated with MRI scanning protocol in both cohorts (EC: 162/385 [42%]; CC: 180/292 [62%]). A substantial number of EC (49/136 [36%]) and CC (50/132 [38%]) patients changed cluster when clustering was performed after z-score-versus LRM normalization. Prognostic modeling based on cluster groups yielded similar outputs for the two normalization methods in the EC/CC cohorts (log-rank test; z-score: p = 0.02/0.33; LRM: p = 0.01/0.45). Mean tdAUC for prognostic modeling of disease-specific survival (DSS) by the radiomic features in EC/CC was similar for the two normalization methods (random survival forests; z-score: mean tdAUC = 0.77/0.78; LRM: mean tdAUC = 0.80/0.75; LASSO Cox; z-score: mean tdAUC = 0.64/0.76; LRM: mean tdAUC = 0.76/0.75). Severe biases in tumor radiomics data due to MRI scanning parameters exist. Z-score normalization does not eliminate these biases, whereas LRM normalization effectively does. Still, radiomic cluster groups after z-score- and LRM normalization were similarly associated with DSS in EC and CC patients.

Radiomics score derived from T1-w/T2-w ratio image can predict motor symptom progression in Parkinson's disease

OBJECTIVES

To clarify the association between a radiomics score (Rad-score) derived from T1-weighted signal intensity to T2-weighted signal intensity (T1-w/T2-w) ratio images and the progression of motor symptoms in Parkinson's disease (PD).

MATERIALS AND METHODS

This retrospective study included patients with PD enrolled in the Parkinson's Progression Markers Initiative. The Movement Disorders Society-Unified Parkinson's Disease Rating Scale Part III score ≥ 33 and/or Hoehn and Yahr stage ≥ 3 indicated motor function decline. The Rad-score was constructed using radiomics features extracted from T1-w/T2-w ratio images. The Kaplan-Meier analysis and Cox regression analyses were used to assess the time differences in motor function decline between the high and low Rad-score groups.

RESULTS

A total of 171 patients with PD were divided into training (n = 101, mean age at baseline, 61.6 ± 9.3 years) and testing (n = 70, mean age at baseline, 61.6 ± 10 years). The patients in the high Rad-score group had a shorter time to motor function decline than those in the low Rad-score group in the training dataset (log-rank test, p < 0.001) and testing dataset (log-rank test, p < 0.001). The multivariate Cox regression using the Rad-score and clinical factors revealed a significant association between the Rad-score and motor function decline in the training dataset (HR = 2.368, 95%CI:1.423-3.943, p < 0.001) and testing dataset (HR = 2.931, 95%CI:1.472-5.837, p = 0.002).

CONCLUSION

Rad-scores based on radiomics features derived from T1-w/T2-w ratio images were associated with the progression of motor symptoms in PD.

CLINICAL RELEVANCE STATEMENT

The radiomics score derived from the T1-weighted/T2-weighted ratio images offers a predictive tool for assessing the progression of motor symptom in patients with PD.

KEY POINTS

Radiomics score derived from T1-weighted/T2-weighted ratio images is correlated with the motor symptoms of Parkinson's disease. A high radiomics score correlated with faster motor function decline in patients with Parkinson's disease. The proposed radiomics score offers predictive insight into the progression of motor symptoms of Parkinson's disease.

Radiomic profiles improve prognostication and reveal targets for therapy in cervical cancer

Cervical cancer (CC) is a major global health problem with 570,000 new cases and 266,000 deaths annually. Prognosis is poor for advanced stage disease, and few effective treatments exist. Preoperative diagnostic imaging is common in high-income countries and MRI measured tumor size routinely guides treatment allocation of cervical cancer patients. Recently, the role of MRI radiomics has been recognized. However, its potential to independently predict survival and treatment response requires further clarification. This retrospective cohort study demonstrates how non-invasive, preoperative, MRI radiomic profiling may improve prognostication and tailoring of treatments and follow-ups for cervical cancer patients. By unsupervised clustering based on 293 radiomic features from 132 patients, we identify three distinct clusters comprising patients with significantly different risk profiles, also when adjusting for FIGO stage and age. By linking their radiomic profiles to genomic alterations, we identify putative treatment targets for the different patient clusters (e.g., immunotherapy, CDK4/6 and YAP-TEAD inhibitors and p53 pathway targeting treatments).

Reproducible Radiomics Features from Multi-MRI-Scanner Test-Retest-Study: Influence on Performance and Generalizability of Models

BACKGROUND

Radiomics models trained on data from one center typically show a decline of performance when applied to data from external centers, hindering their introduction into large-scale clinical practice. Current expert recommendations suggest to use only reproducible radiomics features isolated by multiscanner test-retest experiments, which might help to overcome the problem of limited generalizability to external data.

PURPOSE

To evaluate the influence of using only a subset of robust radiomics features, defined in a prior in vivo multi-MRI-scanner test-retest-study, on the performance and generalizability of radiomics models.

STUDY TYPE

Retrospective.

POPULATION

Patients with monoclonal plasma cell disorders. Training set (117 MRIs from center 1); internal test set (42 MRIs from center 1); external test set (143 MRIs from center 2-8).

FIELD STRENGTH/SEQUENCE

1.5T and 3.0T; T1-weighted turbo spin echo.

ASSESSMENT

The task for the radiomics models was to predict plasma cell infiltration, determined by bone marrow biopsy, noninvasively from MRI. Radiomics machine learning models, including linear regressor, support vector regressor (SVR), and random forest regressor (RFR), were trained on data from center 1, using either all radiomics features, or using only reproducible radiomics features. Models were tested on an internal (center 1) and a multicentric external data set (center 2-8).

STATISTICAL TESTS

Pearson correlation coefficient r and mean absolute error (MAE) between predicted and actual plasma cell infiltration. Fisher's z-transformation, Wilcoxon signed-rank test, Wilcoxon rank-sum test; significance level P < 0.05.

RESULTS

When using only reproducible features compared with all features, the performance of the SVR on the external test set significantly improved (r = 0.43 vs. r = 0.18 and MAE = 22.6 vs. MAE = 28.2). For the RFR, the performance on the external test set deteriorated when using only reproducible instead of all radiomics features (r = 0.33 vs. r = 0.44, P = 0.29 and MAE = 21.9 vs. MAE = 20.5, P = 0.10).

CONCLUSION

Using only reproducible radiomics features improves the external performance of some, but not all machine learning models, and did not automatically lead to an improvement of the external performance of the overall best radiomics model.

LEVEL OF EVIDENCE: 3

TECHNICAL EFFICACY

Stage 2.

Contrast Agent Dynamics Determine Radiomics Profiles in Oncologic Imaging

BACKGROUND

The reproducibility of radiomics features extracted from CT and MRI examinations depends on several physiological and technical factors. The aim was to evaluate the impact of contrast agent timing on the stability of radiomics features using dynamic contrast-enhanced perfusion CT (dceCT) or MRI (dceMRI) in prostate and lung cancers.

METHODS

Radiomics features were extracted from dceCT or dceMRI images in patients with biopsy-proven peripheral prostate cancer (pzPC) or biopsy-proven non-small cell lung cancer (NSCLC), respectively. Features that showed significant differences between contrast phases were identified using linear mixed models. An L2-penalized logistic regression classifier was used to predict class labels for pzPC and unaffected prostate regions-of-interest (ROIs).

RESULTS

Nine pzPC and 28 NSCLC patients, who were imaged with dceCT and/or dceMRI, were included in this study. After normalizing for individual enhancement patterns by defining seven individual phases based on a reference vessel, 19, 467 and 128 out of 1204 CT features showed significant temporal dynamics in healthy prostate parenchyma, prostate tumors and lung tumors, respectively. CT radiomics-based classification accuracy of healthy and tumor ROIs was highly dependent on contrast agent phase. For dceMRI, 899 and 1027 out of 1118 features were significantly dependent on time after contrast agent injection for prostate and lung tumors.

CONCLUSIONS

CT and MRI radiomics features in both prostate and lung tumors are significantly affected by interindividual differences in contrast agent dynamics.

Dose-Incorporated Deep Ensemble Learning for Improving Brain Metastasis Stereotactic Radiosurgery Outcome Prediction

PURPOSE

To develop a novel deep ensemble learning model for accurate prediction of brain metastasis (BM) local control outcomes after stereotactic radiosurgery (SRS).

METHODS AND MATERIALS

A total of 114 brain metastases (BMs) from 82 patients were evaluated, including 26 BMs that developed biopsy-confirmed local failure post-SRS. The SRS spatial dose distribution (Dmap) of each BM was registered to the planning contrast-enhanced T1 (T1-CE) magnetic resonance imaging (MRI). Axial slices of the Dmap, T1-CE, and planning target volume (PTV) segmentation (PTVseg) intersecting the BM center were extracted within a fixed field of view determined by the 60% isodose volume in Dmap. A spherical projection was implemented to transform planar image content onto a spherical surface using multiple projection centers, and the resultant T1-CE/Dmap/PTVseg projections were stacked as a 3-channel variable. Four Visual Geometry Group (VGG-19) deep encoders were used in an ensemble design, with each submodel using a different spherical projection formula as input for BM outcome prediction. In each submodel, clinical features after positional encoding were fused with VGG-19 deep features to generate logit results. The ensemble's outcome was synthesized from the 4 submodel results via logistic regression. In total, 10 model versions with random validation sample assignments were trained to study model robustness. Performance was compared with (1) a single VGG-19 encoder, (2) an ensemble with a T1-CE MRI as the sole image input after projections, and (3) an ensemble with the same image input design without clinical feature inclusion.

RESULTS

The ensemble model achieved an excellent area under the receiver operating characteristic curve (AUCROC: 0.89 ± 0.02) with high sensitivity (0.82 ± 0.05), specificity (0.84 ± 0.11), and accuracy (0.84 ± 0.08) results. This outperformed the MRI-only VGG-19 encoder (sensitivity: 0.35 ± 0.01, AUCROC: 0.64 ± 0.08), the MRI-only deep ensemble (sensitivity: 0.60 ± 0.09, AUCROC: 0.68 ± 0.06), and the 3-channel ensemble without clinical feature fusion (sensitivity: 0.78 ± 0.08, AUCROC: 0.84 ± 0.03).

CONCLUSIONS

Facilitated by the spherical image projection method, a deep ensemble model incorporating Dmap and clinical variables demonstrated excellent performance in predicting BM post-SRS local failure. Our novel approach could improve other radiation therapy outcome models and warrants further evaluation.

Performance of a Region of Interest-based Algorithm in Diagnosing International Society of Urological Pathology Grade Group ≥2 Prostate Cancer on the MRI-FIRST Database-CAD-FIRST Study

BACKGROUND AND OBJECTIVE

Prostate multiparametric magnetic resonance imaging (MRI) shows high sensitivity for International Society of Urological Pathology grade group (GG) ≥2 cancers. Many artificial intelligence algorithms have shown promising results in diagnosing clinically significant prostate cancer on MRI. To assess a region-of-interest-based machine-learning algorithm aimed at characterising GG ≥2 prostate cancer on multiparametric MRI.

METHODS

The lesions targeted at biopsy in the MRI-FIRST dataset were retrospectively delineated and assessed using a previously developed algorithm. The Prostate Imaging-Reporting and Data System version 2 (PI-RADSv2) score assigned prospectively before biopsy and the algorithm score calculated retrospectively in the regions of interest were compared for diagnosing GG ≥2 cancer, using the areas under the curve (AUCs), and sensitivities and specificities calculated with predefined thresholds (PIRADSv2 scores ≥3 and ≥4; algorithm scores yielding 90% sensitivity in the training database). Ten predefined biopsy strategies were assessed retrospectively.

KEY FINDINGS AND LIMITATIONS

After excluding 19 patients, we analysed 232 patients imaged on 16 different scanners; 85 had GG ≥2 cancer at biopsy. At patient level, AUCs of the algorithm and PI-RADSv2 were 77% (95% confidence interval [CI]: 70-82) and 80% (CI: 74-85; p = 0.36), respectively. The algorithm's sensitivity and specificity were 86% (CI: 76-93) and 65% (CI: 54-73), respectively. PI-RADSv2 sensitivities and specificities were 95% (CI: 89-100) and 38% (CI: 26-47), and 89% (CI: 79-96) and 47% (CI: 35-57) for thresholds of ≥3 and ≥4, respectively. Using the PI-RADSv2 score to trigger a biopsy would have avoided 26-34% of biopsies while missing 5-11% of GG ≥2 cancers. Combining prostate-specific antigen density, the PI-RADSv2 and algorithm's scores would have avoided 44-47% of biopsies while missing 6-9% of GG ≥2 cancers. Limitations include the retrospective nature of the study and a lack of PI-RADS version 2.1 assessment.

CONCLUSIONS AND CLINICAL IMPLICATIONS

The algorithm provided robust results in the multicentre multiscanner MRI-FIRST database and could help select patients for biopsy.

PATIENT SUMMARY

An artificial intelligence-based algorithm aimed at diagnosing aggressive cancers on prostate magnetic resonance imaging showed results similar to expert human assessment in a prospectively acquired multicentre test database.

Achieving imaging and computational reproducibility on multiparametric MRI radiomics features in brain tumor diagnosis: phantom and clinical validation

OBJECTIVES

The Image Biomarker Standardization Initiative has helped improve the computational reproducibility of MRI radiomics features. Nonetheless, the MRI sequences and features with high imaging reproducibility are yet to be established. To determine reproducible multiparametric MRI radiomics features across test-retest, multi-scanner, and computational reproducibility comparisons, and to evaluate their clinical value in brain tumor diagnosis.

METHODS

To assess reproducibility, T1-weighted imaging (T1WI), T2-weighted imaging (T2WI), and diffusion-weighted imaging (DWI) were acquired from three 3-T MRI scanners using standardized phantom, and radiomics features were extracted using two computational algorithms. Reproducible radiomics features were selected when the concordance correlation coefficient value above 0.9 across multiple sessions, scanners, and computational algorithms. Random forest classifiers were trained with reproducible features (n = 117) and validated in a clinical cohort (n = 50) to evaluate whether features with high reproducibility improved the differentiation of glioblastoma from primary central nervous system lymphomas (PCNSLs).

RESULTS

Radiomics features from T2WI demonstrated higher repeatability (65-94%) than those from DWI (38-48%) or T1WI (2-92%). Across test-retest, multi-scanner, and computational comparisons, T2WI provided 41 reproducible features, DWI provided six, and T1WI provided two. The performance of the classification model with reproducible features was higher than that using non-reproducible features in both training set (AUC, 0.916 vs. 0.877) and validation set (AUC, 0.957 vs. 0.869).

CONCLUSION

Radiomics features with high reproducibility across multiple sessions, scanners, and computational algorithms were identified, and they showed higher diagnostic performance than non-reproducible radiomics features in the differentiation of glioblastoma from PCNSL.

CLINICAL RELEVANCE STATEMENT

By identifying the radiomics features showing higher multi-machine reproducibility, our results also demonstrated higher radiomics diagnostic performance in the differentiation of glioblastoma from PCNSL, paving the way for further research designs and clinical application in neuro-oncology.

KEY POINTS

• Highly reproducible radiomics features across multiple sessions, scanners, and computational algorithms were identified using phantom and applied to clinical diagnosis. • Radiomics features from T2-weighted imaging were more reproducible than those from T1-weighted and diffusion-weighted imaging. • Radiomics features with good reproducibility had better diagnostic performance for brain tumors than features with poor reproducibility.

The Value of Dual-Energy Computed Tomography-Based Radiomics in the Evaluation of Interstitial Fibers of Clear Cell Renal Carcinoma

OBJECTIVE

We investigated the potential of dual-energy computed tomography (DECT) radiomics in assessing cancer-associated fibroblasts in clear cell renal carcinoma (ccRCC).

METHODS

A retrospective analysis was conducted on 132 patients with ccRCC. The arterial and venous phase iodine-based material decomposition images (IMDIs), virtual non-contrast images, 70 keV, 100 keV, and 150 keV virtual monoenergetic images, and mixed energy images (MEIs) were obtained from the DECT datasets. On the Radcloud platform, radiomics feature extraction, feature selection, and model establishment were performed. Seven radiomics models were established using the support vector machine. The predictive performance was evaluated by utilizing receiver operating characteristic and the area under the curve (AUC) was calculated. Nomograms were constructed.

RESULTS

The combined model demonstrated high efficiency in evaluating pseudocapsule thickness with AUC, specificity, and sensitivity of 0.833, 0.870, and 0.750, respectively in the validation set, surpassing those of other models. The precision, F1-score, and Youden index were also higher for the combined model. For evaluating the number of collagen fibers, the combined model exhibited the highest AUC (0.741) among all models, with a specificity of 0.830 and a sensitivity of 0.330. The AUC in the 150 kv model and IMDI model were slightly lower than those in the combined model (0.728 and 0.710, respectively), with corresponding sensitivity and specificity of 0.560/0.780 and 0.670/0.830. The nomogram exhibited that Rad-score had good prediction efficiency.

CONCLUSION

DECT radiomics features have significant value in evaluating the interstitial fibers of ccRCC. The combined model of IMDI + MEI exhibits superior performance in assessing the thickness of the pseudocapsule, while the combined, 150 keV, and IMDI models demonstrate higher efficacy in evaluating collagen fiber number. Radiomics, combined with imaging features and clinical features, has excellent predictive performance. These findings offer crucial support for the clinical diagnosis, treatment, and prognosis of ccRCC and provide valuable insights into the application of DECT.

An Explainable MRI-Radiomic Quantum Neural Network to Differentiate Between Large Brain Metastases and High-Grade Glioma Using Quantum Annealing for Feature Selection

Solitary large brain metastases (LBM) and high-grade gliomas (HGG) are sometimes hard to differentiate on MRI. The management differs significantly between these two entities, and non-invasive methods that help differentiate between them are eagerly needed to avoid potentially morbid biopsies and surgical procedures. We explore herein the performance and interpretability of an MRI-radiomics variational quantum neural network (QNN) using a quantum-annealing mutual-information (MI) feature selection approach. We retrospectively included 423 patients with HGG and LBM (> 2 cm) who had a contrast-enhanced T1-weighted (CE-T1) MRI between 2012 and 2019. After exclusion, 72 HGG and 129 LBM were kept. Tumors were manually segmented, and a 5-mm peri-tumoral ring was created. MRI images were pre-processed, and 1813 radiomic features were extracted. A set of best features based on MI was selected. MI and conditional-MI were embedded into a quadratic unconstrained binary optimization (QUBO) formulation that was mapped to an Ising-model and submitted to D'Wave's quantum annealer to solve for the best combination of 10 features. The 10 selected features were embedded into a 2-qubits QNN using PennyLane library. The model was evaluated for balanced-accuracy (bACC) and area under the receiver operating characteristic curve (ROC-AUC) on the test set. The model performance was benchmarked against two classical models: dense neural networks (DNN) and extreme gradient boosting (XGB). Shapley values were calculated to interpret sample-wise predictions on the test set. The best 10-feature combination included 6 tumor and 4 ring features. For QNN, DNN, and XGB, respectively, training ROC-AUC was 0.86, 0.95, and 0.94; test ROC-AUC was 0.76, 0.75, and 0.79; and test bACC was 0.74, 0.73, and 0.72. The two most influential features were tumor Laplacian-of-Gaussian-GLRLM-Entropy and sphericity. We developed an accurate interpretable QNN model with quantum-informed feature selection to differentiate between LBM and HGG on CE-T1 brain MRI. The model performance is comparable to state-of-the-art classical models.

Radiomics for predicting survival in patients with locally advanced rectal cancer: a systematic review and meta-analysis

Background

Radiomics has recently received considerable research attention for providing potential prognostic biomarkers for locally advanced rectal cancer (LARC). We aimed to comprehensively evaluate the methodological quality and prognostic prediction value of radiomic studies for predicting survival outcomes in patients with LARC.

Methods

The Cochrane, Embase, Medline, and Web of Science databases were searched. The radiomics quality score (RQS), Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) checklist, the Image Biomarkers Standardization Initiative (IBSI) guideline, and the Prediction Model Risk of Bias Assessment Tool were used to assess the quality of the selected studies. A further meta-analysis of hazard ratio (HR) regarding disease-free survival (DFS) and overall survival (OS) was performed.

Results

Among the 358 studies reported, 15 studies were selected for our review. The mean RQS score was 7.73±4.61 (21.5% of the ideal score of 36). The overall TRIPOD adherence rate was 64.4% (251/390). Most of the included studies (60%) were assessed as having a high risk of bias (ROB) overall. The pooled estimates of the HRs were 3.14 [95% confidence interval (CI): 2.12-4.64, P<0.01] for DFS and 3.36 (95% CI: 1.74-6.49, P<0.01) for OS.

Conclusions

Radiomics has potential to noninvasively predict outcome in patients with LARC. However, the overall methodological quality of radiomics studies was low, and the adherence to the TRIPOD statement was moderate. Future radiomics research should put a greater focus on enhancing the methodological quality and considering the influence of higher-order features on reproducibility in radiomics.

Clinical and Magnetic Resonance Imaging Radiomics-Based Survival Prediction in Glioblastoma Using Multiparametric Magnetic Resonance Imaging

INTRODUCTION

Survival prediction in glioblastoma remains challenging, and identification of robust imaging markers could help with this relevant clinical problem. We evaluated multiparametric magnetic resonance imaging-derived radiomics to assess prediction of overall survival (OS) and progression-free survival (PFS).

METHODOLOGY

A retrospective, institutional review board-approved study was performed. There were 93 eligible patients, of which 55 underwent gross tumor resection and chemoradiation (GTR-CR). Overall survival and PFS were assessed in the entire cohort and the GTR-CR cohort using multiple machine learning pipelines. A model based on multiple clinical variables was also developed. Survival prediction was assessed using the radiomics-only, clinical-only, and the radiomics and clinical combined models.

RESULTS

For all patients combined, the clinical feature-derived model outperformed the best radiomics model for both OS (C-index, 0.706 vs 0.597; P < 0.0001) and PFS prediction (C-index, 0.675 vs 0.588; P < 0.001). Within the GTR-CR cohort, the radiomics model showed nonstatistically improved performance over the clinical model for predicting OS (C-index, 0.638 vs 0.588; P = 0.4). However, the radiomics model outperformed the clinical feature model for predicting PFS in GTR-CR cohort (C-index, 0.641 vs 0.550; P = 0.004). Combined clinical and radiomics model did not yield superior prediction when compared with the best model in each case.

CONCLUSIONS

When considering all patients, regardless of therapy, the radiomics-derived prediction of OS and PFS is inferior to that from a model derived from clinical features alone. However, in patients with GTR-CR, radiomics-only model outperforms clinical feature-derived model for predicting PFS.

Comparing effectiveness of image perturbation and test retest imaging in improving radiomic model reliability

Image perturbation is a promising technique to assess radiomic feature repeatability, but whether it can achieve the same effect as test-retest imaging on model reliability is unknown. This study aimed to compare radiomic model reliability based on repeatable features determined by the two methods using four different classifiers. A 191-patient public breast cancer dataset with 71 test-retest scans was used with pre-determined 117 training and 74 testing samples. We collected apparent diffusion coefficient images and manual tumor segmentations for radiomic feature extraction. Random translations, rotations, and contour randomizations were performed on the training images, and intra-class correlation coefficient (ICC) was used to filter high repeatable features. We evaluated model reliability in both internal generalizability and robustness, which were quantified by training and testing AUC and prediction ICC. Higher testing performance was found at higher feature ICC thresholds, but it dropped significantly at ICC = 0.95 for the test-retest model. Similar optimal reliability can be achieved with testing AUC = 0.7-0.8 and prediction ICC > 0.9 at the ICC threshold of 0.9. It is recommended to include feature repeatability analysis using image perturbation in any radiomic study when test-retest is not feasible, but care should be taken when deciding the optimal feature repeatability criteria.

Role of MRI radiomics for the prediction of MYCN amplification in neuroblastomas

OBJECTIVES

We evaluate MR radiomics and develop machine learning-based classifiers to predict MYCN amplification in neuroblastomas.

METHODS

A total of 120 patients with neuroblastomas and baseline MR imaging examination available were identified of whom 74 (mean age ± standard deviation [SD] of 6 years and 2 months ± 4 years and 9 months; 43 females and 31 males, 14 MYCN amplified) underwent imaging at our institution. This was therefore used to develop radiomics models. The model was tested in a cohort of children with the same diagnosis but imaged elsewhere (n = 46, mean age ± SD: 5 years 11 months ± 3 years 9 months, 26 females and 14 MYCN amplified). Whole tumour volumes of interest were adopted to extract first-order histogram and second-order radiomics features. Interclass correlation coefficient and maximum relevance and minimum redundancy algorithm were applied for feature selection. Logistic regression, support vector machine, and random forest were employed as the classifiers. Receiver operating characteristic (ROC) analysis was performed to evaluate the diagnostic accuracy of the classifiers on the external test set.

RESULTS

The logistic regression model and the random forest both showed an AUC of 0.75. The support vector machine classifier obtained an AUC of 0.78 on the test set with a sensitivity of 64% and a specificity of 72%.

CONCLUSION

The study provides preliminary retrospective evidence demonstrating the feasibility of MRI radiomics in predicting MYCN amplification in neuroblastomas. Future studies are needed to explore the correlation between other imaging features and genetic markers and to develop multiclass predictive models.

KEY POINTS

• MYCN amplification in neuroblastomas is an important determinant of disease prognosis. • Radiomics analysis of pre-treatment MR examinations can be used to predict MYCN amplification in neuroblastomas. • Radiomics machine learning models showed good generalisability to external test set, demonstrating reproducibility of the computational models.

Prognostic prediction of left ventricular myocardial noncompaction using machine learning and cardiac magnetic resonance radiomics

Background

Although there are many studies on the prognostic factors of left ventricular myocardial noncompaction (LVNC), the determinants are varied and not entirely consistent. This study aimed to build predictive models using radiomics features and machine learning to predict major adverse cardiovascular events (MACEs) in patients with LVNC.

Methods

In total, 96 patients with LVNC were included and randomly divided into training and test cohorts. A total of 105 cine cardiac magnetic resonance (CMR)-derived radiomics features and 35 clinical characteristics were extracted. Five different oversampling algorithms were compared for selection of the optimal imbalanced processing. Feature importance was assessed with extreme gradient boosting (XGBoost). We compared the performance of 5 machine learning classification methods with different sample:feature ratios to determine the optimal hybrid classification strategy. Subsequently, radiomics, clinical, and combined radiomics-clinical models were developed and compared.

Results

The machine learning pipeline included an adaptive synthetic (ADASYN) algorithm for imbalanced processing, XGBoost feature selection with a sample:feature ratio of 10, and support vector machine (SVM) modeling. The areas under the receiver operating characteristic curves (AUCs) of the radiomics model, clinical model, and combined model in the validation cohort were 0.87 (sensitivity 83.33%, specificity 64.29%), 0.65 (sensitivity 16.67%, specificity 78.57%), and 0.92 (specificity 33.33%, sensitivity 100.00%), respectively. The radiomics model performed similarly to the clinical and combined models (P=0.124 and P=0.621, respectively). The performance of the combined model was significantly better than that of the clinical model (P=0.003).

Conclusions

The machine learning-based cine CMR radiomics model performed well at predicting MACEs in patients with LVNC. Adding radiomics features offered incremental prognostic value over clinical factors alone.

Differentiation of Small Clear Renal Cell Carcinoma and Oncocytoma through Magnetic Resonance Imaging-Based Radiomics Analysis: Toward the End of Percutaneous Biopsy

PURPOSE

The aim of this study was to ascertain whether radiomics data can assist in differentiating small (<4 cm) clear cell renal cell carcinomas (ccRCCs) from small oncocytomas using T2-weighted magnetic resonance imaging (MRI).

MATERIAL AND METHODS

This retrospective study incorporated 48 tumors, 28 of which were ccRCCs and 20 were oncocytomas. All tumors were less than 4 cm in size and had undergone pre-biopsy or pre-surgery MRI. Following image pre-processing, 102 radiomics features were evaluated. A univariate analysis was performed using the Wilcoxon rank-sum test with Bonferroni correction. We compared multiple radiomics pipelines of normalization, feature selection, and machine learning (ML) algorithms, including random forest (RF), logistic regression (LR), AdaBoost, K-nearest neighbor, and support vector machine, using a supervised ML approach.

RESULTS

No statistically significant features were identified via the univariate analysis with Bonferroni correction. The most effective algorithm was identified using a pipeline incorporating standard normalization, RF-based feature selection, and LR, which achieved an area under the curve (AUC) of 83%, accuracy of 73%, sensitivity of 79%, and specificity of 65%. Subsequently, the most significant features were identified from this algorithm, and two groups of uncorrelated features were established based on Pearson correlation scores. Using these features, an algorithm was established after a pipeline of standard normalization and LR, achieving an AUC of 90%, an accuracy of 77%, sensitivity of 83%, and specificity of 69% for distinguishing ccRCCs from oncocytomas.

CONCLUSIONS

Radiomics analysis based on T2-weighted MRI can aid in distinguishing small ccRCCs from small oncocytomas. However, it is not superior to standard multiparameter renal MRI and does not yet allow us to dispense with percutaneous biopsy.

ComBat Harmonization of Myocardial Radiomic Features Sensitive to Cardiac MRI Acquisition Parameters

Purpose

To investigate the effect of ComBat harmonization methods on the robustness of cardiac MRI-derived radiomic features to variations in imaging parameters.

Materials and Methods

This Health Insurance Portability and Accountability Act-compliant retrospective study used a publicly available data set of 11 healthy controls (mean age, 33 years ± 16 [SD]; six men) and five patients (mean age, 52 years ± 16; four men). A single midventricular short-axis section was acquired with 3-T MRI using cine balanced steady-state free precision, T1-weighted, T2-weighted, T1 mapping, and T2 mapping imaging sequences. Each sequence was acquired using baseline parameters and after variations in flip angle, spatial resolution, section thickness, and parallel imaging. Image registration was performed for all sequences at a per-individual level. Manual myocardial contouring was performed, and 1652 radiomic features per sequence were extracted using baseline and variations in imaging parameters. Radiomic feature stability to change in imaging parameters was assessed using Cohen d sensitivity. The stability of radiomic features was assessed both without and after ComBat harmonization of radiomic features. Three ComBat methods were studied: parametric, nonparametric, and Gaussian mixture model (GMM).

Results

For all sequences combined, 51.4% of features were robust to changes in imaging parameters when no ComBat method was applied. ComBat harmonization substantially increased the number of stable features to 95.1% (95% CI: 94.9, 95.3) when parametric ComBat was used and 90.9% (95% CI: 90.6, 91.2) when nonparametric ComBat was used. GMM combat resulted in only 52.6% stable features.

Conclusion

ComBat harmonization improved the stability of radiomic features to changes in imaging parameters across all cardiac MRI sequences.Keywords: Cardiac MRI, Radiomics, ComBat, Harmonization Supplemental material is available for this article. © RSNA, 2023.

Advancements in MRI-Based Radiomics and Artificial Intelligence for Prostate Cancer: A Comprehensive Review and Future Prospects

The use of multiparametric magnetic resonance imaging (mpMRI) has become a common technique used in guiding biopsy and developing treatment plans for prostate lesions. While this technique is effective, non-invasive methods such as radiomics have gained popularity for extracting imaging features to develop predictive models for clinical tasks. The aim is to minimize invasive processes for improved management of prostate cancer (PCa). This study reviews recent research progress in MRI-based radiomics for PCa, including the radiomics pipeline and potential factors affecting personalized diagnosis. The integration of artificial intelligence (AI) with medical imaging is also discussed, in line with the development trend of radiogenomics and multi-omics. The survey highlights the need for more data from multiple institutions to avoid bias and generalize the predictive model. The AI-based radiomics model is considered a promising clinical tool with good prospects for application.

In Vivo Repeatability and Multiscanner Reproducibility of MRI Radiomics Features in Patients With Monoclonal Plasma Cell Disorders: A Prospective Bi-institutional Study

OBJECTIVES

Despite the extensive number of publications in the field of radiomics, radiomics algorithms barely enter large-scale clinical application. Supposedly, the low external generalizability of radiomics models is one of the main reasons, which hinders the translation from research to clinical application. The objectives of this study were to investigate reproducibility of radiomics features (RFs) in vivo under variation of patient positioning, magnetic resonance imaging (MRI) sequence, and MRI scanners, and to identify a subgroup of RFs that shows acceptable reproducibility across all different acquisition scenarios.

MATERIALS AND METHODS

Between November 30, 2020 and February 16, 2021, 55 patients with monoclonal plasma cell disorders were included in this prospective, bi-institutional, single-vendor study. Participants underwent one reference scan at a 1.5 T MRI scanner and several retest scans: once after simple repositioning, once with a second MRI protocol, once at another 1.5 T scanner, and once at a 3 T scanner. Radiomics feature from the bone marrow of the left hip bone were extracted, both from original scans and after different image normalizations. Intraclass correlation coefficient (ICC) was used to assess RF repeatability and reproducibility.

RESULTS

Fifty-five participants (mean age, 59 ± 7 years; 36 men) were enrolled. For T1-weighted images after muscle normalization, in the simple test-retest experiment, 110 (37%) of 295 RFs showed an ICC ≥0.8: 54 (61%) of 89 first-order features (FOFs), 35 (95%) of 37 volume and shape features, and 21 (12%) of 169 texture features (TFs). When the retest was performed with different technical settings, even after muscle normalization, the number of FOF/TF with an ICC ≥0.8 declined to 58/13 for the second protocol, 29/7 for the second 1.5 T scanner, and 49/7 for the 3 T scanner, respectively. Twenty-five (28%) of the 89 FOFs and 6 (4%) of the 169 TFs from muscle-normalized T1-weighted images showed an ICC ≥0.8 throughout all repeatability and reproducibility experiments.

CONCLUSIONS

In vivo, only few RFs are reproducible with different MRI sequences or different MRI scanners, even after application of a simple image normalization. Radiomics features selected by a repeatability experiment only are not necessarily suited to build radiomics models for multicenter clinical application. This study isolated a subset of RFs, which are robust to variations in MRI acquisition observed in scanners from 1 vendor, and therefore are candidates to build reproducible radiomics models for monoclonal plasma cell disorders for multicentric applications, at least when centers are equipped with scanners from this vendor.

MRI radiomics and potential applications to glioblastoma

MRI plays an important role in the evaluation of glioblastoma, both at initial diagnosis and follow up after treatment. Quantitative analysis via radiomics can augment the interpretation of MRI in terms of providing insights regarding the differential diagnosis, genotype, treatment response, and prognosis. The various MRI radiomic features of glioblastoma are reviewed in this article.

Clinical Radiobiology for Radiation Oncology

This chapter is focused on radiobiological aspects at the molecular, cellular, and tissue level which are relevant for the clinical use of ionizing radiation (IR) in cancer therapy. For radiation oncology, it is critical to find a balance, i.e., the therapeutic window, between the probability of tumor control and the probability of side effects caused by radiation injury to the healthy tissues and organs. An overview is given about modern precision radiotherapy (RT) techniques, which allow optimal sparing of healthy tissues. Biological factors determining the width of the therapeutic window are explained. The role of the six typical radiobiological phenomena determining the response of both malignant and normal tissues in the clinic, the 6R’s, which are Reoxygenation, Redistribution, Repopulation, Repair, Radiosensitivity, and Reactivation of the immune system, is discussed. Information is provided on tumor characteristics, for example, tumor type, growth kinetics, hypoxia, aberrant molecular signaling pathways, cancer stem cells and their impact on the response to RT. The role of the tumor microenvironment and microbiota is described and the effects of radiation on the immune system including the abscopal effect phenomenon are outlined. A summary is given on tumor diagnosis, response prediction via biomarkers, genetics, and radiomics, and ways to selectively enhance the RT response in tumors. Furthermore, we describe acute and late normal tissue reactions following exposure to radiation: cellular aspects, tissue kinetics, latency periods, permanent or transient injury, and histopathology. Details are also given on the differential effect on tumor and late responding healthy tissues following fractionated and low dose rate irradiation as well as the effect of whole-body exposure.

Radiomics in clinical trials: perspectives on standardization

The term biomarker is used to describe a biological measure of the disease behavior. The existing imaging biomarkers are associated with the known tissue biological characteristics and follow a well-established roadmap to be implemented in routine clinical practice. Recently, a new quantitative imaging analysis approach named radiomics has emerged. It refers to the extraction of a large number of advanced imaging features with high-throughput computing. Extensive research has demonstrated its value in predicting disease behavior, progression, and response to therapeutic options. However, there are numerous challenges to establishing it as a clinically viable solution, including lack of reproducibility and transparency. The data-driven nature also does not offer insights into the underpinning biology of the observed relationships. As such, additional effort is needed to establish it as a qualified biomarker to inform clinical decisions. Here we review the technical difficulties encountered in the clinical applications of radiomics and current effort in addressing some of these challenges in clinical trial designs. By addressing these challenges, the true potential of radiomics can be unleashed.

Minimising multi-centre radiomics variability through image normalisation: a pilot study

Radiomics is an emerging technique for the quantification of imaging data that has recently shown great promise for deeper phenotyping of cardiovascular disease. Thus far, the technique has been mostly applied in single-centre studies. However, one of the main difficulties in multi-centre imaging studies is the inherent variability of image characteristics due to centre differences. In this paper, a comprehensive analysis of radiomics variability under several image- and feature-based normalisation techniques was conducted using a multi-centre cardiovascular magnetic resonance dataset. 218 subjects divided into healthy (n = 112) and hypertrophic cardiomyopathy (n = 106, HCM) groups from five different centres were considered. First and second order texture radiomic features were extracted from three regions of interest, namely the left and right ventricular cavities and the left ventricular myocardium. Two methods were used to assess features’ variability. First, feature distributions were compared across centres to obtain a distribution similarity index. Second, two classification tasks were proposed to assess: (1) the amount of centre-related information encoded in normalised features (centre identification) and (2) the generalisation ability for a classification model when trained on these features (healthy versus HCM classification). The results showed that the feature-based harmonisation technique ComBat is able to remove the variability introduced by centre information from radiomic features, at the expense of slightly degrading classification performance. Piecewise linear histogram matching normalisation gave features with greater generalisation ability for classification ( balanced accuracy in between 0.78 ± 0.08 and 0.79 ± 0.09). Models trained with features from images without normalisation showed the worst performance overall ( balanced accuracy in between 0.45 ± 0.28 and 0.60 ± 0.22). In conclusion, centre-related information removal did not imply good generalisation ability for classification.

Uniqueness of radiomic features in non-small cell lung cancer

PURPOSE

The uniqueness of radiomic features, combined with their reproducibility, determines the reliability of radiomic studies. This study is to test the hypothesis that radiomic features extracted from a defined region of interest (ROI) are unique to the underlying structure (e.g., tumor).

APPROACH

Two cohorts of non-small cell lung cancer (NSCLC) patients were retrospectively retrieved from a GE and a Siemens CT scanner. The lung nodules (ROI) were delineated manually and radiomic features were extracted using IBEX. The same ROI was then translocated randomly to four other tissue regions of the same set of images: adipose, heart, lung beyond nodule, and muscle for radiomic feature extraction. Coefficient of variation (CV) within different ROIs and concordance correlation coefficient (CCC) between lung nodule and a given tissue region were calculated to test to determine feature uniqueness. The radiomic features were considered nonunique when (1) the CV < 10% and CCC > 0.85 for over 50% of patients; and (2) the CCC > 0.85 appeared in ≥2 tissue regions beyond the defined region.

RESULTS

In total, 14 patients from GE and 18 patients from Siemens are analyzed. The results show that 12 features fall below the 10% CV threshold for over 50% of patients in the GE cohort and 29 features in the Siemens cohort. According to CCC, 18 radiomic features in GE and 16 features in Siemens are identified as nonunique, with 11 overlapping features. Combining CV and CCC, 9 of 123 calculated features (7.3%) are identified as nonunique to a defined ROI.

CONCLUSIONS

Radiomic feature uniqueness should be considered to improve the reliability of radiomics study.

Phantom Study on the Robustness of MR Radiomics Features: Comparing the Applicability of 3D Printed and Biological Phantoms

The objectives of our study were to (a) evaluate the feasibility of using 3D printed phantoms in magnetic resonance imaging (MR) in assessing the robustness and repeatability of radiomic parameters and (b) to compare the results obtained from the 3D printed phantoms to metrics obtained in biological phantoms. To this end, three different 3D phantoms were printed: a Hilbert cube (5 × 5 × 5 cm³) and two cubic quick response (QR) code phantoms (a large phantom (large QR) (5 × 5 × 4 cm³) and a small phantom (small QR) (4 × 4 × 3 cm³)). All 3D printed and biological phantoms (kiwis, tomatoes, and onions) were scanned thrice on clinical 1.5 T and 3 T MR with 1 mm and 2 mm isotropic resolution. Subsequent analyses included analyses of several radiomics indices (RI), their repeatability and reliability were calculated using the coefficient of variation (CV), the relative percentage difference (RPD), and the interclass coefficient (ICC) parameters. Additionally, the readability of QR codes obtained from the MR images was examined with several mobile phones and algorithms. The best repeatability (CV ≤ 10%) is reported for the acquisition protocols with the highest spatial resolution. In general, the repeatability and reliability of RI were better in data obtained at 1.5 T (CV = 1.9) than at 3 T (CV = 2.11). Furthermore, we report good agreements between results obtained for the 3D phantoms and biological phantoms. Finally, analyses of the read-out rate of the QR code revealed better texture analyses for images with a spatial resolution of 1 mm than 2 mm. In conclusion, 3D printing techniques offer a unique solution to create textures for analyzing the reliability of radiomic data from MR scans.

The CT delta-radiomics based machine learning approach in evaluating multiple primary lung adenocarcinoma

Object

To evaluate the difference between multiple primary lung adenocarcinoma (MPLA) and solitary primary lung adenocarcinoma (SPLA) by delta-radiomics based machine learning algorithms in CT images.

Methods

A total of 1094 patients containing 268 MPLAs and 826 SPLAs were recruited for this retrospective study between 2014 to 2020. After the segmentation of volume of interest, the radiomic features were automatically calculated. The patients were categorized into the training set and testing set by a random proportion of 7:3. After feature selection, the relevant classifiers were constructed by the machine learning algorithms of Bayes, forest, k-nearest neighbor, logistic regression, support vector machine, and decision tree. The relative standard deviation (RSD) was calculated and the classification model with minimal RSD was chosen for delta-radiomics analysis to explore the variation of tumor during follow-up surveillance in the cohort of 225 MPLAs and 320 SPLAs. According to the different follow-up duration, it was divided into group A (3–12 months), group B (13–24 months), and group C (25–48 months). Then the corresponding delta-radiomics classifiers were developed to predict MPLAs. The area under the receiver operator characteristic curve (AUC) with 95% confidence interval (CI) was quantified to evaluate the efficiency of the model.

Results

To radiomics analysis, the forest classifier (FC-radio) with the minimal RSD showed the better stability with AUCs of 0.840 (95%CI, 0.810–0.867) and 0.670 (95%CI, 0.611–0.724) in the training and testing set. The AUCs of the forest classifier based on delta-radiomics (FC-delta) were higher than those of FC-radio. In addition, with the extension of follow-up duration, the performance of FC-delta in Group C were the best with AUCs of 0.998 (95%CI, 0.993–1.000) in the training set and 0.853 (95%CI, 0.752–0.940) in the testing set.

Conclusions

The machine-learning approach based on radiomics and delta-radiomics helped to differentiate SPLAs from MPLAs. The FC-delta with a longer follow-up duration could better distinguish between SPLAs and MPLAs.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-022-10036-1.

Challenges in Glioblastoma Radiomics and the Path to Clinical Implementation

Radiomics is a field of medical imaging analysis that focuses on the extraction of many quantitative imaging features related to shape, intensity and texture. These features are incorporated into models designed to predict important clinical or biological endpoints for patients. Attention for radiomics research has recently grown dramatically due to the increased use of imaging and the availability of large, publicly available imaging datasets. Glioblastoma multiforme (GBM) patients stand to benefit from this emerging research field as radiomics has the potential to assess the biological heterogeneity of the tumour, which contributes significantly to the inefficacy of current standard of care therapy. Radiomics models still require further development before they are implemented clinically in GBM patient management. Challenges relating to the standardisation of the radiomics process and the validation of radiomic models impede the progress of research towards clinical implementation. In this manuscript, we review the current state of radiomics in GBM, and we highlight the barriers to clinical implementation and discuss future validation studies needed to advance radiomics models towards clinical application.

Impact of image quality on radiomics applications

Radiomics features extracted from medical images have been widely reported to be useful in the patient specific outcome modeling for variety of assessment and prediction purposes. Successful application of radiomics features as imaging biomarkers, however, is dependent on the robustness of the approach to the variation in each step of the modeling workflow. Variation in the input image quality is one of the main sources that impacts the reproducibility of radiomics analysis when a model is applied to broader range of medical imaging data. The quality of medical image is generally affected by both the scanner related factors such as image acquisition/reconstruction settings and the patient related factors such as patient motion. This article aimed to review the published literatures in this field that reported the impact of various imaging factors on the radiomics features through the change in image quality. The literatures were categorized by different imaging modalities and also tabulated based on the imaging parameters and the class of radiomics features included in the study. Strategies for image quality standardization were discussed based on the relevant literatures and recommendations for reducing the impact of image quality variation on the radiomics in multi-institutional clinical trial were summarized at the end of this article.

The Potential and Emerging Role of Quantitative Imaging Biomarkers for Cancer Characterization

Simple Summary

Modern, personalized therapy approaches are increasingly changing advanced cancer into a chronic disease. Compared to imaging, novel omics methodologies in molecular biology have already achieved an individual characterization of cancerous lesions. With quantitative imaging biomarkers, analyzed by radiomics or deep learning, an imaging-based assessment of tumoral biology can be brought into clinical practice. Combining these with other non-invasive methods, e.g., liquid profiling, could allow for more individual decision making regarding therapies and applications.

Abstract

Similar to the transformation towards personalized oncology treatment, emerging techniques for evaluating oncologic imaging are fostering a transition from traditional response assessment towards more comprehensive cancer characterization via imaging. This development can be seen as key to the achievement of truly personalized and optimized cancer diagnosis and treatment. This review gives a methodological introduction for clinicians interested in the potential of quantitative imaging biomarkers, treating of radiomics models, texture visualization, convolutional neural networks and automated segmentation, in particular. Based on an introduction to these methods, clinical evidence for the corresponding imaging biomarkers—(i) dignity and etiology assessment; (ii) tumoral heterogeneity; (iii) aggressiveness and response; and (iv) targeting for biopsy and therapy—is summarized. Further requirements for the clinical implementation of these imaging biomarkers and the synergistic potential of personalized molecular cancer diagnostics and liquid profiling are discussed.

A Comparative Study on the Potential of Unsupervised Deep Learning-based Feature Selection in Radiomics

In Radiomics, deep learning-based systems for medical image analysis play an increasing role. However, due to the better explainability, feature-based systems are still preferred, especially by physicians. Often, high-dimensional data and low sample size pose different challenges (e.g. increased risk of overfitting) to machine learning systems. By removing irrelevant and redundant features from the data, feature selection is an effective way of pre-processing. The research in this study is focused on unsupervised deep learning-based methods for feature selection. Five recently proposed algorithms are compared regarding their applicability and efficiency on seven data sets in three different sample applications. It was found that deep learning-based feature selection leads to improved classification results compared to conventional methods, especially for small feature subsets. Clinical Relevance - The exploration of distinctive features and the ability to rank their importance without the need for outcome information is a potential field of application for unsupervised feature selection methods. Especially in multiparametric radiology, the number of features is increasing. The identification of new potential biomarkers is important both for treatment and prevention.

Repeatability and reproducibility of magnetic resonance imaging-based radiomic features in rectal cancer

Purpose: Radiomics of magnetic resonance images (MRIs) in rectal cancer can non-invasively characterize tumor heterogeneity with potential to discover new imaging biomarkers. However, for radiomics to be reliable, the imaging features measured must be stable and reproducible. The aim of this study is to quantify the repeatability and reproducibility of MRI-based radiomic features in rectal cancer. Approach: An MRI radiomics phantom was used to measure the longitudinal repeatability of radiomic features and the impact of post-processing changes related to image resolution and noise. Repeatability measurements in rectal cancers were also quantified in a cohort of 10 patients with test-retest imaging among two observers. Results: We found that many radiomic features, particularly from texture classes, were highly sensitive to changes in image resolution and noise. About 49% of features had coefficient of variations ≤ 10 % in longitudinal phantom measurements. About 75% of radiomic features in in vivo test-retest measurements had an intraclass correlation coefficient of ≥ 0.8 . We saw excellent interobserver agreement with mean Dice similarity coefficient of 0.95 ± 0.04 for test and retest scans. Conclusions: The results of this study show that even when using a consistent imaging protocol many radiomic features were unstable. Therefore, caution must be taken when selecting features for potential imaging biomarkers.

Radiomics and deep learning for myocardial scar screening in hypertrophic cardiomyopathy

BACKGROUND

Myocardial scar burden quantified using late gadolinium enhancement (LGE) cardiovascular magnetic resonance (CMR), has important prognostic value in hypertrophic cardiomyopathy (HCM). However, nearly 50% of HCM patients have no scar but undergo repeated gadolinium-based CMR over their life span. We sought to develop an artificial intelligence (AI)-based screening model using radiomics and deep learning (DL) features extracted from balanced steady state free precession (bSSFP) cine sequences to identify HCM patients without scar.

METHODS

We evaluated three AI-based screening models using bSSFP cine image features extracted by radiomics, DL, or combined DL-Radiomics. Images for 759 HCM patients (50 ± 16 years, 66% men) in a multi-center/vendor study were used to develop and test model performance. An external dataset of 100 HCM patients (53 ± 14 years, 70% men) was used to assess model generalizability. Model performance was evaluated using area-under-receiver-operating curve (AUC).

RESULTS

The DL-Radiomics model demonstrated higher AUC compared to DL and Radiomics in the internal (0.83 vs 0.77, p = 0.006 and 0.78, p = 0.05; n = 159) and external (0.74 vs 0.64, p = 0.006 and 0.71, p = 0.27; n = 100) datasets. The DL-Radiomics model correctly identified 43% and 28% of patients without scar in the internal and external datasets compared to 42% and 16% by Radiomics model and 42% and 23% by DL model, respectively.

CONCLUSIONS

A DL-Radiomics AI model using bSSFP cine images outperforms DL or Radiomics models alone as a scar screening tool prior to gadolinium administration. Despite its potential, the clinical utility of the model remains limited and further investigation is needed to improve the accuracy and generalizability.

Building reliable radiomic models using image perturbation

Radiomic model reliability is a central premise for its clinical translation. Presently, it is assessed using test–retest or external data, which, unfortunately, is often scarce in reality. Therefore, we aimed to develop a novel image perturbation-based method (IPBM) for the first of its kind toward building a reliable radiomic model. We first developed a radiomic prognostic model for head-and-neck cancer patients on a training (70%) and evaluated on a testing (30%) cohort using C-index. Subsequently, we applied the IPBM to CT images of both cohorts (Perturbed-Train and Perturbed-Test cohort) to generate 60 additional samples for both cohorts. Model reliability was assessed using intra-class correlation coefficient (ICC) to quantify consistency of the C-index among the 60 samples in the Perturbed-Train and Perturbed-Test cohorts. Besides, we re-trained the radiomic model using reliable RFs exclusively (ICC > 0.75) to validate the IPBM. Results showed moderate model reliability in Perturbed-Train (ICC: 0.565, 95%CI 0.518–0.615) and Perturbed-Test (ICC: 0.596, 95%CI 0.527–0.670) cohorts. An enhanced reliability of the re-trained model was observed in Perturbed-Train (ICC: 0.782, 95%CI 0.759–0.815) and Perturbed-Test (ICC: 0.825, 95%CI 0.782–0.867) cohorts, indicating validity of the IPBM. To conclude, we demonstrated capability of the IPBM toward building reliable radiomic models, providing community with a novel model reliability assessment strategy prior to prospective evaluation.

Impact of Parallel Acquisition Technology on the Robustness of Magnetic Resonance Imaging Radiomic Features

OBJECTIVE

The aim of this study was to investigate the impact of integrated parallel acquisition technology (iPAT) on the robustness of magnetic resonance imaging radiomic features.

METHODS

A phantom and 6 healthy volunteers were scanned on a clinical 3-T system using T1-weighted (S1), T1-weighted fluid-attenuated (S2), T2-weighted fluid-attenuated (S3), and T2-weighted (S4); 2 iPAT flavors (generalized autocalibration partially parallel acquisitions and modified sensitivity encoding [mSENSE]) and their different acceleration factors R. Radiomic features were extracted, and their robustness was assessed using coefficient of variation (CV), and differences between sequences and region of interest (ROI) were evaluated using the χ2 test.

RESULTS

One volunteer was excluded because of movement during imaging acquisition. Generalized autocalibration partially parallel acquisitions provided more radiomic features with excellent robustness than mSENSE. Radiomic features with excellent robustness, unaffected by iPAT across different sequences and ROIs, in 92 radiomic features for phantom and healthy volunteers are 6.5% and 2.2%. For phantom, difference in the robustness degree between 4 sequences/P-ROIs was significant according to χ2 test; S2 and S3 could provide more excellent robust radiomic features than S1 and S4, and P-ROI3 filled with the biggest polystyrene particles could provide the most radiomic features with excellent robustness than the other P-ROIs. For healthy volunteers, only the difference in the degree of robustness between the 4 V-ROIs was significant, and V-ROI3 in white matter region of the left frontal lobe, which was located at periphery in image, could provide the most robust radiomic features compared with other V-ROIs.

CONCLUSIONS

Integrated parallel acquisition technology had a significant impact on the robustness of radiomic features. Generalized autocalibration partially parallel acquisitions delivered a more robust substrate for radiomic analyses than mSENSE.

Additively manufactured, solid object structures for adjustable image contrast in Magnetic Resonance Imaging

The choice of materials challenges the development of Magnetic Resonance Imaging (MRI) phantoms and, to date, is mainly limited to water-filled compartments or gel-based components. Recently, solid materials have been introduced through additive manufacturing (AM) to mimic complex geometrical structures. Nonetheless, no such manufactured solid materials are available with controllable MRI contrast to mimic organ substructures or lesion heterogeneities. Here, we present a novel AM design that allows MRI contrast manipulation by varying the partial volume contribution to a ROI/voxel of MRI-visible material within an imaging object. Two sets of 11 cubes and three replicates of a spherical tumour model were designed and printed using AM. Most samples presented varying MRI-contrast in standard MRI sequences, based mainly on spin density and partial volume signal variation. A smooth and continuous MRI-contrast gradient could be generated in a single-compartment tumour model. This concept supports the development of more complex MRI phantoms that mimic the appearance of heterogeneous tumour tissues.

The stability of oncologic MRI radiomic features and the potential role of deep learning: a review

The use of MRI radiomic models for the diagnosis, prognosis and treatment response prediction of tumors has been increasingly reported in literature. However, its widespread adoption in clinics is hampered by issues related to features stability. In the MRI radiomic workflow, the main factors that affect radiomic features computation can be found in the image acquisition and reconstruction phase, in the image pre-processing steps, and in the segmentation of the region of interest on which radiomic indices are extracted. Deep Neural Networks (DNNs), having shown their potentiality in the medical image processing and analysis field, can be seen as an attractive strategy to partially overcome the issues related to radiomic stability and mitigate their impact. In fact, DNN approaches can be prospectively integrated in the MRI radiomic workflow to improve image quality, obtain accurate and reproducible segmentations and generate standardized images. In this review, DNN methods that can be included in the image processing steps of the radiomic workflow are described and discussed, in the light of a detailed analysis of the literature in the context of MRI radiomic reliability.

Comparative performance of fully-automated and semi-automated artificial intelligence methods for the detection of clinically significant prostate cancer on MRI: a systematic review

OBJECTIVES

We systematically reviewed the current literature evaluating the ability of fully-automated deep learning (DL) and semi-automated traditional machine learning (TML) MRI-based artificial intelligence (AI) methods to differentiate clinically significant prostate cancer (csPCa) from indolent PCa (iPCa) and benign conditions.

METHODS

We performed a computerised bibliographic search of studies indexed in MEDLINE/PubMed, arXiv, medRxiv, and bioRxiv between 1 January 2016 and 31 July 2021. Two reviewers performed the title/abstract and full-text screening. The remaining papers were screened by four reviewers using the Checklist for Artificial Intelligence in Medical Imaging (CLAIM) for DL studies and Radiomics Quality Score (RQS) for TML studies. Papers that fulfilled the pre-defined screening requirements underwent full CLAIM/RQS evaluation alongside the risk of bias assessment using QUADAS-2, both conducted by the same four reviewers. Standard measures of discrimination were extracted for the developed predictive models.

RESULTS

17/28 papers (five DL and twelve TML) passed the quality screening and were subject to a full CLAIM/RQS/QUADAS-2 assessment, which revealed a substantial study heterogeneity that precluded us from performing quantitative analysis as part of this review. The mean RQS of TML papers was 11/36, and a total of five papers had a high risk of bias. AUCs of DL and TML papers with low risk of bias ranged between 0.80-0.89 and 0.75-0.88, respectively.

CONCLUSION

We observed comparable performance of the two classes of AI methods and identified a number of common methodological limitations and biases that future studies will need to address to ensure the generalisability of the developed models.

Local tuning of radiomics-based model for predicting pathological response to neoadjuvant chemoradiotherapy in locally advanced rectal cancer

PURPOSE

This study aims to further enhance a validated radiomics-based model for predicting pathologic complete response (pCR) after chemo‑radiotherapy in locally advanced rectal cancer (LARC) for use in clinical practice.

METHODS

A generalized linear model (GLM) to predict pCR in LARC patients previously trained in Europe and validated with an external inter-continental cohort (59 patients), was first examined with further 88 intercontinental patient datasets to assess its reproducibility; then new radiomics and clinical features, and validation methods were investigated to build a new model for enhancing the pCR prediction for patients admitted to our department. The patients were divided into training group (75%) and validation group (25%) according to their demographic. The least absolute shrinkage and selection operator (LASSO) logistic regression was used to reduce the dimensionality of the extracted features of the training group and select the optimal ones; the performance of the reference GLM and enhanced models was compared through the area under curve (AUC) of the receiver operating characteristics.

RESULTS

The value of AUC of the reference model was 0.831 (95% CI, 0.701-0.961), and 0.828 (95% CI, 0.700-0.956) in the original and new validation cohorts, respectively, showing a reproducibility in the applicability of the GLM model. Eight features were found to be significant with LASSO and used to establish an enhanced model. The AUC of the enhanced model of 0.926 (95% CI, 0.859-0.993) for training, and 0.926 (95% CI, 0.767-1.00) for the validation group shows better performance than the reference model.

CONCLUSIONS

The GLM model shows good reproducibility in predicting pCR in LARC; the enhanced model has the potential to improve prediction accuracy and may be a candidate in clinical practice.

Robustness and reproducibility of radiomics in T2 weighted images from magnetic resonance image guided linear accelerator in a phantom study

PURPOSE

Quantitative radiomics features extracted from medical images have been shown to provide value in predicting clinical outcomes. The study for robustness and reproducibility of radiomics features obtained with magnetic resonance image guided linear accelerator (MR-Linac) is insufficient. The objective of this work was to investigate the stability of radiomics features extracted from T2-weighted images of MR-Linac for five common effect factors.

MATERIALS AND METHOD

In this work, ten jellies, five fruits/vegetables, and a dynamic phantom were used to evaluate the impact of test-retest, intraobserver, varied thicknesses, radiation, and motion. These phantoms were scanned on a 1.5 T MRI system of MR-Linac. For test-retest data, the phantoms were scanned twice with repositioning within 15 min. To assess for intraobserver comparison, the segmentation of MR images was repeated by one observer in a double-blind manner. Three slice thicknesses (1.2 mm, 2.4 mm, and 4.8 mm) were used to select robust features that were insensitive to different thicknesses. The effect of radiation on features was studied by acquiring images when the beam was on. Common movement images of patients during radiotherapy were simulated by a dynamic phantom with five motion states to study the motion effect. A total of 1409 radiomics features, including shape features, first-order features, and texture features, were extracted from the original, wavelet, square, logarithmic, exponential and gradient images. The robustness and reproducibility features were evaluated using the concordance correlation coefficient (CCC).

RESULT

The intraobserver group had the most robust features (936/1079, 86.7%), while the group of motion effects had the lowest robustness (56/936, 6.0%), followed by the group of different thickness cohorts (374/936, 40.0%). The stability of features in the test-retest and radiation groups was 1072 of 1312 (81.7%) and 810 of 936 (86.5%), respectively. Overall, 25 of 1409 (2.4%) radiomics features remained robust in all five tests, mostly focusing on the image type of the wavelet. The number of stable features extracted from when the beam was on was less than that extracted when the beam was off. Shape features were the most robust of all of the features in all of the groups, excluding the motion group.

CONCLUSION

Compared with other factors fewer features remained robust to the effect of motion. This result emphasizes the need to consider the effect of respiration motion. The study for T2-weighted images from MR-Linac under different conditions will help us to build a robust predictive model applicable for radiotherapy.

Sources of variation in multicenter rectal MRI data and their effect on radiomics feature reproducibility

OBJECTIVES

To investigate sources of variation in a multicenter rectal cancer MRI dataset focusing on hardware and image acquisition, segmentation methodology, and radiomics feature extraction software.

METHODS

T2W and DWI/ADC MRIs from 649 rectal cancer patients were retrospectively acquired in 9 centers. Fifty-two imaging features (14 first-order/6 shape/32 higher-order) were extracted from each scan using whole-volume (expert/non-expert) and single-slice segmentations using two different software packages (PyRadiomics/CapTk). Influence of hardware, acquisition, and patient-intrinsic factors (age/gender/cTN-stage) on ADC was assessed using linear regression. Feature reproducibility was assessed between segmentation methods and software packages using the intraclass correlation coefficient.

RESULTS

Image features differed significantly (p < 0.001) between centers with more substantial variations in ADC compared to T2W-MRI. In total, 64.3% of the variation in mean ADC was explained by differences in hardware and acquisition, compared to 0.4% by patient-intrinsic factors. Feature reproducibility between expert and non-expert segmentations was good to excellent (median ICC 0.89-0.90). Reproducibility for single-slice versus whole-volume segmentations was substantially poorer (median ICC 0.40-0.58). Between software packages, reproducibility was good to excellent (median ICC 0.99) for most features (first-order/shape/GLCM/GLRLM) but poor for higher-order (GLSZM/NGTDM) features (median ICC 0.00-0.41).

CONCLUSIONS

Significant variations are present in multicenter MRI data, particularly related to differences in hardware and acquisition, which will likely negatively influence subsequent analysis if not corrected for. Segmentation variations had a minor impact when using whole volume segmentations. Between software packages, higher-order features were less reproducible and caution is warranted when implementing these in prediction models.

KEY POINTS

• Features derived from T2W-MRI and in particular ADC differ significantly between centers when performing multicenter data analysis. • Variations in ADC are mainly (> 60%) caused by hardware and image acquisition differences and less so (< 1%) by patient- or tumor-intrinsic variations. • Features derived using different image segmentations (expert/non-expert) were reproducible, provided that whole-volume segmentations were used. When using different feature extraction software packages with similar settings, higher-order features were less reproducible.

Test-Retest Data for the Assessment of Breast MRI Radiomic Feature Repeatability

Background

Radiomic features extracted from breast MRI have potential for diagnostic, prognostic, and predictive purposes. However, before they can be used as biomarkers in clinical decision support systems, features need to be repeatable and reproducible.

Objective

Identify repeatable radiomics features within breast tissue on prospectively collected MRI exams through multiple test–retest measurements.

Study Type

Prospective.

Population

11 healthy female volunteers.

Field Strength/Sequence

1.5 T; MRI exams, comprising T2‐weighted turbo spin‐echo (T2W) sequence, native T1‐weighted turbo gradient‐echo (T1W) sequence, diffusion‐weighted imaging (DWI) sequence using b‐values 0/150/800, and corresponding derived ADC maps.

Assessment

18 MRI exams (three test–retest settings, repeated on 2 days) per healthy volunteer were examined on an identical scanner using a fixed clinical breast protocol. For each scan, 91 features were extracted from the 3D manually segmented right breast using Pyradiomics, before and after image preprocessing. Image preprocessing consisted of 1) bias field correction (BFC); 2) z‐score normalization with and without BFC; 3) grayscale discretization using 32 and 64 bins with and without BFC; and 4) z‐score normalization + grayscale discretization using 32 and 64 bins with and without BFC.

Statistical Tests

Features' repeatability was assessed using concordance correlation coefficient(CCC) for each pair, i.e. each MRI was compared to each of the remaining 17 MRI with a cut‐off value of CCC > 0.90.

Results

Images without preprocessing produced the highest number of repeatable features for both T1W sequence and ADC maps with 15 of 91 (16.5%) and 8 of 91 (8.8%) repeatable features, respectively. Preprocessed images produced between 4 of 91 (4.4%) and 14 of 91 (15.4%), and 6 of 91 (6.6%) and 7 of 91 (7.7%) repeatable features, respectively for T1W and ADC maps. Z‐score normalization produced highest number of repeatable features, 26 of 91 (28.6%) in T2W sequences, in these images, no preprocessing produced 11 of 91 (12.1%) repeatable features.

Data Conclusion

Radiomic features extracted from T1W, T2W sequences and ADC maps from breast MRI exams showed a varying number of repeatable features, depending on the sequence. Effects of different preprocessing procedures on repeatability of features were different for each sequence.

Level of Evidence

2

Technical Efficacy Stage

1

Making Radiomics More Reproducible across Scanner and Imaging Protocol Variations: A Review of Harmonization Methods

Radiomics converts medical images into mineable data via a high-throughput extraction of quantitative features used for clinical decision support. However, these radiomic features are susceptible to variation across scanners, acquisition protocols, and reconstruction settings. Various investigations have assessed the reproducibility and validation of radiomic features across these discrepancies. In this narrative review, we combine systematic keyword searches with prior domain knowledge to discuss various harmonization solutions to make the radiomic features more reproducible across various scanners and protocol settings. Different harmonization solutions are discussed and divided into two main categories: image domain and feature domain. The image domain category comprises methods such as the standardization of image acquisition, post-processing of raw sensor-level image data, data augmentation techniques, and style transfer. The feature domain category consists of methods such as the identification of reproducible features and normalization techniques such as statistical normalization, intensity harmonization, ComBat and its derivatives, and normalization using deep learning. We also reflect upon the importance of deep learning solutions for addressing variability across multi-centric radiomic studies especially using generative adversarial networks (GANs), neural style transfer (NST) techniques, or a combination of both. We cover a broader range of methods especially GANs and NST methods in more detail than previous reviews.

Systematic Review on the Association of Radiomics with Tumor Biological Endpoints

Radiomics supposes an alternative non-invasive tumor characterization tool, which has experienced increased interest with the advent of more powerful computers and more sophisticated machine learning algorithms. Nonetheless, the incorporation of radiomics in cancer clinical-decision support systems still necessitates a thorough analysis of its relationship with tumor biology. Herein, we present a systematic review focusing on the clinical evidence of radiomics as a surrogate method for tumor molecular profile characterization. An extensive literature review was conducted in PubMed, including papers on radiomics and a selected set of clinically relevant and commonly used tumor molecular markers. We summarized our findings based on different cancer entities, additionally evaluating the effect of different modalities for the prediction of biomarkers at each tumor site. Results suggest the existence of an association between the studied biomarkers and radiomics from different modalities and different tumor sites, even though a larger number of multi-center studies are required to further validate the reported outcomes.

Impact of Preprocessing and Harmonization Methods on the Removal of Scanner Effects in Brain MRI Radiomic Features

Simple Summary

As a rapid-development research field, radiomics-based analysis has been applied to many clinical problems. However, the reproducibility of the radiomics studies remain challenging especially when data suffers from scanner effects, a kind of non-biological variations introduced by different image acquiring settings. This study aims to investigate how the image preprocessing methods (N4 bias field correction and image resampling) and the harmonization methods (intensity normalization methods working on images and ComBat method working on radiomic features) help to remove the scanner effects and improve the radiomics reproducibility in brain MRI radiomics.

Abstract

In brain MRI radiomics studies, the non-biological variations introduced by different image acquisition settings, namely scanner effects, affect the reliability and reproducibility of the radiomics results. This paper assesses how the preprocessing methods (including N4 bias field correction and image resampling) and the harmonization methods (either the six intensity normalization methods working on brain MRI images or the ComBat method working on radiomic features) help to remove the scanner effects and improve the radiomic feature reproducibility in brain MRI radiomics. The analyses were based on in vitro datasets (homogeneous and heterogeneous phantom data) and in vivo datasets (brain MRI images collected from healthy volunteers and clinical patients with brain tumors). The results show that the ComBat method is essential and vital to remove scanner effects in brain MRI radiomic studies. Moreover, the intensity normalization methods, while not able to remove scanner effects at the radiomic feature level, still yield more comparable MRI images and improve the robustness of the harmonized features to the choice among ComBat implementations.

Observing deep radiomics for the classification of glioma grades

Deep learning is a promising method for medical image analysis because it can automatically acquire meaningful representations from raw data. However, a technical challenge lies in the difficulty of determining which types of internal representation are associated with a specific task, because feature vectors can vary dynamically according to individual inputs. Here, based on the magnetic resonance imaging (MRI) of gliomas, we propose a novel method to extract a shareable set of feature vectors that encode various parts in tumor imaging phenotypes. By applying vector quantization to latent representations, features extracted by an encoder are replaced with a fixed set of feature vectors. Hence, the set of feature vectors can be used in downstream tasks as imaging markers, which we call deep radiomics. Using deep radiomics, a classifier is established using logistic regression to predict the glioma grade with 90% accuracy. We also devise an algorithm to visualize the image region encoded by each feature vector, and demonstrate that the classification model preferentially relies on feature vectors associated with the presence or absence of contrast enhancement in tumor regions. Our proposal provides a data-driven approach to enhance the understanding of the imaging appearance of gliomas.