Multisite evaluation of radiomic feature reproducibility and discriminability for identifying peripheral zone prostate tumors on MRI

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.

Solving the Pervasive Problem of Protocol Non-Compliance in MRI using an Open-Source tool mrQA

Pooling data across diverse sources acquired by multisite consortia requires compliance with a predefined reference protocol i.e., ensuring different sites and scanners for a given project have used identical or compatible MR physics parameter values. Traditionally, this has been an arduous and manual process due to difficulties in working with the complicated DICOM standard and lack of resources allocated towards protocol compliance. Moreover, issues of protocol compliance is often overlooked for lack of realization that parameter values are routinely improvised/modified locally at various sites. The inconsistencies in acquisition protocols can reduce SNR, statistical power, and in the worst case, may invalidate the results altogether. An open-source tool, mrQA was developed to automatically assess protocol compliance on standard dataset formats such as DICOM and BIDS, and to study the patterns of non-compliance in over 20 open neuroimaging datasets, including the large ABCD study. The results demonstrate that the lack of compliance is rather pervasive. The frequent sources of non-compliance include but are not limited to deviations in Repetition Time, Echo Time, Flip Angle, and Phase Encoding Direction. It was also observed that GE and Philips scanners exhibited higher rates of non-compliance relative to the Siemens scanners in the ABCD dataset. Continuous monitoring for protocol compliance is strongly recommended before any pre/post-processing, ideally right after the acquisition, to avoid the silent propagation of severe/subtle issues. Although, this study focuses on neuroimaging datasets, the proposed tool mrQA can work with any DICOM-based datasets.

A Physics-Based Algorithm to Universally Standardize Routinely Obtained Clinical T2-Weighted Images

RATIONALE AND OBJECTIVES

MR images can be challenging for machine learning and other large-scale analyses because most clinical images, for example, T2-weighted (T2w) images, reflect not only the biologically relevant T2 of tissue but also hardware and acquisition parameters that vary from site to site. Quantitative T2 mapping avoids these confounds because it quantitatively isolates the biological parameter of interest, thus representing a universal standardization across sites. However, efforts to incorporate quantitative mapping sequences into routine clinical practice have seen slow adoption. Here we show, for the first time, that the routine T2w complex raw dataset can be successfully regarded as a quantitative mapping sequence that can be reconstructed with classical optimization methods and physics-based constraints.

MATERIALS AND METHODS

While previous constrained reconstruction methods are unable to reconstruct a T2 map based on this data, the expanding-constrained alternating minimization for parameter mapping (e-CAMP), which employs stepwise initialization, a linearized version of the exponential model and a phase conjugacy constraint, is demonstrated to provide useful quantitative maps directly from a vendor T2w single image data.

RESULTS

This paper introduces the method and demonstrates its performance using simulations, retrospectively undersampled brain images, and prospectively acquired T2w images taken on both phantom and brain.

CONCLUSION

Because T2w scans are included in nearly every protocol, this approach could open the door to creating large, standardized datasets without requiring widespread changes in clinical protocols.

Leveraging physiology and artificial intelligence to deliver advancements in health care

Artificial intelligence in health care has experienced remarkable innovation and progress in the last decade. Significant advancements can be attributed to the utilization of artificial intelligence to transform physiology data to advance health care. In this review, we explore how past work has shaped the field and defined future challenges and directions. In particular, we focus on three areas of development. First, we give an overview of artificial intelligence, with special attention to the most relevant artificial intelligence models. We then detail how physiology data have been harnessed by artificial intelligence to advance the main areas of health care: automating existing health care tasks, increasing access to care, and augmenting health care capabilities. Finally, we discuss emerging concerns surrounding the use of individual physiology data and detail an increasingly important consideration for the field, namely the challenges of deploying artificial intelligence models to achieve meaningful clinical impact.

Artificial intelligence algorithms aimed at characterizing or detecting prostate cancer on MRI: How accurate are they when tested on independent cohorts? – A systematic review

PURPOSE

The purpose of this study was to perform a systematic review of the literature on the diagnostic performance, in independent test cohorts, of artificial intelligence (AI)-based algorithms aimed at characterizing/detecting prostate cancer on magnetic resonance imaging (MRI).

MATERIALS AND METHODS

Medline, Embase and Web of Science were searched for studies published between January 2018 and September 2022, using a histological reference standard, and assessing prostate cancer characterization/detection by AI-based MRI algorithms in test cohorts composed of more than 40 patients and with at least one of the following independency criteria as compared to the training cohort: different institution, different population type, different MRI vendor, different magnetic field strength or strict temporal splitting.

RESULTS

Thirty-five studies were selected. The overall risk of bias was low. However, 23 studies did not use predefined diagnostic thresholds, which may have optimistically biased the results. Test cohorts fulfilled one to three of the five independency criteria. The diagnostic performance of the algorithms used as standalones was good, challenging that of human reading. In the 12 studies with predefined diagnostic thresholds, radiomics-based computer-aided diagnosis systems (assessing regions-of-interest drawn by the radiologist) tended to provide more robust results than deep learning-based computer-aided detection systems (providing probability maps). Two of the six studies comparing unassisted and assisted reading showed significant improvement due to the algorithm, mostly by reducing false positive findings.

CONCLUSION

Prostate MRI AI-based algorithms showed promising results, especially for the relatively simple task of characterizing predefined lesions. The best management of discrepancies between human reading and algorithm findings still needs to be defined.

Early stage NSCLS patients' prognostic prediction with multi-information using transformer and graph neural network model

Background

We proposed a population graph with Transformer-generated and clinical features for the purpose of predicting overall survival (OS) and recurrence-free survival (RFS) for patients with early stage non-small cell lung carcinomas and to compare this model with traditional models.

Methods

The study included 1705 patients with lung cancer (stages I and II), and a public data set for external validation (n=127). We proposed a graph with edges representing non-imaging patient characteristics and nodes representing imaging tumour region characteristics generated by a pretrained Vision Transformer. The model was compared with a TNM model and a ResNet-Graph model. To evaluate the models' performance, the area under the receiver operator characteristic curve (ROC-AUC) was calculated for both OS and RFS prediction. The Kaplan-Meier method was used to generate prognostic and survival estimates for low- and high-risk groups, along with net reclassification improvement (NRI), integrated discrimination improvement (IDI), and decision curve analysis. An additional subanalysis was conducted to examine the relationship between clinical data and imaging features associated with risk prediction.

Results

Our model achieved AUC values of 0.785 (95% confidence interval [CI]: 0.716-0.855) and 0.695 (95% CI: 0.603-0.787) on the testing and external data sets for OS prediction, and 0.726 (95% CI: 0.653-0.800) and 0.700 (95% CI: 0.615-0.785) for RFS prediction. Additional survival analyses indicated that our model outperformed the present TNM and ResNet-Graph models in terms of net benefit for survival prediction.

Conclusions

Our Transformer-Graph model was effective at predicting survival in patients with early stage lung cancer, which was constructed using both imaging and non-imaging clinical features. Some high-risk patients were distinguishable by using a similarity score function defined by non-imaging characteristics such as age, gender, histology type, and tumour location, while Transformer-generated features demonstrated additional benefits for patients whose non-imaging characteristics were non-discriminatory for survival outcomes.

Funding

The study was supported by the National Natural Science Foundation of China (91959126, 8210071009), and Science and Technology Commission of Shanghai Municipality (20XD1403000, 21YF1438200).

Data harmonisation for information fusion in digital healthcare: A state-of-the-art systematic review, meta-analysis and future research directions

Removing the bias and variance of multicentre data has always been a challenge in large scale digital healthcare studies, which requires the ability to integrate clinical features extracted from data acquired by different scanners and protocols to improve stability and robustness. Previous studies have described various computational approaches to fuse single modality multicentre datasets. However, these surveys rarely focused on evaluation metrics and lacked a checklist for computational data harmonisation studies. In this systematic review, we summarise the computational data harmonisation approaches for multi-modality data in the digital healthcare field, including harmonisation strategies and evaluation metrics based on different theories. In addition, a comprehensive checklist that summarises common practices for data harmonisation studies is proposed to guide researchers to report their research findings more effectively. Last but not least, flowcharts presenting possible ways for methodology and metric selection are proposed and the limitations of different methods have been surveyed for future research.

Prostate Surface Distension and Tumor Texture Descriptors From Pre-Treatment MRI Are Associated With Biochemical Recurrence Following Radical Prostatectomy: Preliminary Findings

Objective

To derive and evaluate the association of prostate shape distension descriptors from T2-weighted MRI (T2WI) with prostate cancer (PCa) biochemical recurrence (BCR) post-radical prostatectomy (RP) independently and in conjunction with texture radiomics of PCa.

Methods

This retrospective study comprised 133 PCa patients from two institutions who underwent 3T-MRI prior to RP and were followed up with PSA measurements for ≥3 years. A 3D shape atlas-based approach was adopted to derive prostate shape distension descriptors from T2WI, and these descriptors were used to train a random forest classifier (C_S) to predict BCR. Texture radiomics was derived within PCa regions of interest from T2WI and ADC maps, and another machine learning classifier (C_R) was trained for BCR. An integrated classifier C_S+R was then trained using predictions from C_S and C_R. These models were trained on D₁ (N = 71, 27 BCR+) and evaluated on independent hold-out set D₂ (N = 62, 12 BCR+). C_S+R was compared against pre-RP, post-RP clinical variables, and extant nomograms for BCR-free survival (bFS) at 3 years.

Results

C_S+R resulted in a higher AUC (0.75) compared to C_R (0.70, p = 0.04) and C_S (0.69, p = 0.01) on D₂ in predicting BCR. On univariable analysis, C_S+R achieved a higher hazard ratio (2.89, 95% CI 0.35–12.81, p < 0.01) compared to other pre-RP clinical variables for bFS. C_S+R, pathologic Gleason grade, extraprostatic extension, and positive surgical margins were associated with bFS (p < 0.05). C_S+R resulted in a higher C-index (0.76 ± 0.06) compared to CAPRA (0.69 ± 0.09, p < 0.01) and Decipher risk (0.59 ± 0.06, p < 0.01); however, it was comparable to post-RP CAPRA-S (0.75 ± 0.02, p = 0.07).

Conclusions

Radiomic shape descriptors quantifying prostate surface distension complement texture radiomics of prostate cancer on MRI and result in an improved association with biochemical recurrence post-radical prostatectomy.

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.

A novel radiomics signature based on T2-weighted imaging accurately predicts hepatic inflammation in individuals with biopsy-proven nonalcoholic fatty liver disease: a derivation and independent validation study

Background

Currently, there are no effective methods for assessing hepatic inflammation without resorting to histological examination of liver tissue obtained by biopsy. T2-weighted images (T2WI) are routinely obtained from liver magnetic resonance imaging (MRI) scan sequences. We aimed to establish a radiomics signature based on T2WI (T2-RS) for assessment of hepatic inflammation in people with nonalcoholic fatty liver disease (NAFLD).

Methods

A total of 203 individuals with biopsy-confirmed NAFLD from two independent Chinese cohorts with liver MRI examination were enrolled in this study. The hepatic inflammatory activity score (IAS) was calculated by the unweighted sum of the histologic scores for lobular inflammation and ballooning. One thousand and thirty-two radiomics features were extracted from the localized region of interest (ROI) in the right liver lobe of T2WI and, subsequently, selected by minimum redundancy maximum relevance and least absolute shrinkage and selection operator (LASSO) methods. The T2-RS was calculated by adding the selected features weighted by their coefficients.

Results

Eighteen radiomics features from Laplacian of Gaussian, wavelet, and original images were selected for establishing T2-RS. The T2-RS value differed significantly between groups with increasing grades of hepatic inflammation (P<0.01). The T2-RS yielded an area under the receiver operating characteristic (ROC) curve (AUROC) of 0.80 [95% confidence interval (CI): 0.71-0.89] for predicting hepatic inflammation in the training cohort with excellent calibration. The AUROCs of T2-RS in the internal cohort and external validation cohorts were 0.77 (0.61-0.93) and 0.75 (0.63-0.84), respectively.

Conclusions

The T2-RS derived from radiomics analysis of T2WI shows promising utility for predicting hepatic inflammation in individuals with NAFLD.

Predicting cancer outcomes with radiomics and artificial intelligence in radiology

The successful use of artificial intelligence (AI) for diagnostic purposes has prompted the application of AI-based cancer imaging analysis to address other, more complex, clinical needs. In this Perspective, we discuss the next generation of challenges in clinical decision-making that AI tools can solve using radiology images, such as prognostication of outcome across multiple cancers, prediction of response to various treatment modalities, discrimination of benign treatment confounders from true progression, identification of unusual response patterns and prediction of the mutational and molecular profile of tumours. We describe the evolution of and opportunities for AI in oncology imaging, focusing on hand-crafted radiomic approaches and deep learning-derived representations, with examples of their application for decision support. We also address the challenges faced on the path to clinical adoption, including data curation and annotation, interpretability, and regulatory and reimbursement issues. We hope to demystify AI in radiology for clinicians by helping them to understand its limitations and challenges, as well as the opportunities it provides as a decision-support tool in cancer management.

The Reproducibility of Deep Learning-Based Segmentation of the Prostate Gland and Zones on T2-Weighted MR Images

Volume of interest segmentation is an essential step in computer-aided detection and diagnosis (CAD) systems. Deep learning (DL)-based methods provide good performance for prostate segmentation, but little is known about the reproducibility of these methods. In this work, an in-house collected dataset from 244 patients was used to investigate the intra-patient reproducibility of 14 shape features for DL-based segmentation methods of the whole prostate gland (WP), peripheral zone (PZ), and the remaining prostate zones (non-PZ) on T2-weighted (T2W) magnetic resonance (MR) images compared to manual segmentations. The DL-based segmentation was performed using three different convolutional neural networks (CNNs): V-Net, nnU-Net-2D, and nnU-Net-3D. The two-way random, single score intra-class correlation coefficient (ICC) was used to measure the inter-scan reproducibility of each feature for each CNN and the manual segmentation. We found that the reproducibility of the investigated methods is comparable to manual for all CNNs (14/14 features), except for V-Net in PZ (7/14 features). The ICC score for segmentation volume was found to be 0.888, 0.607, 0.819, and 0.903 in PZ; 0.988, 0.967, 0.986, and 0.983 in non-PZ; 0.982, 0.975, 0.973, and 0.984 in WP for manual, V-Net, nnU-Net-2D, and nnU-Net-3D, respectively. The results of this work show the feasibility of embedding DL-based segmentation in CAD systems, based on multiple T2W MR scans of the prostate, which is an important step towards the clinical implementation.

Prospective Evaluation of Repeatability and Robustness of Radiomic Descriptors in Healthy Brain Tissue Regions In Vivo Across Systematic Variations in T2-Weighted Magnetic Resonance Imaging Acquisition Parameters

BACKGROUND

Radiomic descriptors from magnetic resonance imaging (MRI) are promising for disease diagnosis and characterization but may be sensitive to differences in imaging parameters.

OBJECTIVE

To evaluate the repeatability and robustness of radiomic descriptors within healthy brain tissue regions on prospectively acquired MRI scans; in a test-retest setting, under controlled systematic variations of MRI acquisition parameters, and after postprocessing.

STUDY TYPE

Prospective.

SUBJECTS

Fifteen healthy participants.

FIELD STRENGTH/SEQUENCE

A 3.0 T, axial T₂ -weighted 2D turbo spin-echo pulse sequence, 181 scans acquired (2 test/retest reference scans and 12 with systematic variations in contrast weighting, resolution, and acceleration per participant; removing scans with artifacts).

ASSESSMENT

One hundred and forty-six radiomic descriptors were extracted from a contiguous 2D region of white matter in each scan, before and after postprocessing.

STATISTICAL TESTS

Repeatability was assessed in a test/retest setting and between manual and automated annotations for the reference scan. Robustness was evaluated between the reference scan and each group of variant scans (contrast weighting, resolution, and acceleration). Both repeatability and robustness were quantified as the proportion of radiomic descriptors that fell into distinct ranges of the concordance correlation coefficient (CCC): excellent (CCC > 0.85), good (0.7 ≤ CCC ≤ 0.85), moderate (0.5 ≤ CCC < 0.7), and poor (CCC < 0.5); for unprocessed and postprocessed scans separately.

RESULTS

Good to excellent repeatability was observed for 52% of radiomic descriptors between test/retest scans and 48% of descriptors between automated vs. manual annotations, respectively. Contrast weighting (TR/TE) changes were associated with the largest proportion of highly robust radiomic descriptors (21%, after processing). Image resolution changes resulted in the largest proportion of poorly robust radiomic descriptors (97%, before postprocessing). Postprocessing of images with only resolution/acceleration differences resulted in 73% of radiomic descriptors showing poor robustness.

DATA CONCLUSIONS

Many radiomic descriptors appear to be nonrobust across variations in MR contrast weighting, resolution, and acceleration, as well in test-retest settings, depending on feature formulation and postprocessing.

EVIDENCE LEVEL

2 TECHNICAL EFFICACY: Stage 2.

Radiomic Features of Multiparametric MRI Present Stable Associations With Analogous Histological Features in Patients With Brain Cancer

Magnetic resonance (MR)-derived radiomic features have shown substantial predictive utility in modeling different prognostic factors of glioblastoma and other brain cancers. However, the biological relationship underpinning these predictive models has been largely unstudied, and the generalizability of these models had been called into question. Here, we examine the localized relationship between MR-derived radiomic features and histology-derived “histomic” features using a data set of 16 patients with brain cancer. Tile-based radiomic features were collected on T1, post-contrast T1, FLAIR, and diffusion-weighted imaging (DWI)-derived apparent diffusion coefficient (ADC) images acquired before patient death, with analogous histomic features collected for autopsy samples coregistered to the magnetic resonance imaging. Features were collected for each original image, as well as a 3D wavelet decomposition of each image, resulting in 837 features per MR and histology image. Correlative analyses were used to assess the degree of association between radiomic–histomic pairs for each magnetic resonance imaging. The influence of several confounds was also assessed using linear mixed-effect models for the normalized radiomic–histomic distance, testing for main effects of different acquisition field strengths. Results as a whole were largely heterogeneous, but several features showed substantial associations with their histomic analogs, particularly those derived from the FLAIR and postcontrast T1W images. These features with the strongest association typically presented as stable across field strengths as well. These data suggest that a subset of radiomic features can consistently capture texture information on underlying tissue histology.

Computer extracted gland features from H&E predicts prostate cancer recurrence comparably to a genomic companion diagnostic test: a large multi-site study

Existing tools for post-radical prostatectomy (RP) prostate cancer biochemical recurrence (BCR) prognosis rely on human pathologist-derived parameters such as tumor grade, with the resulting inter-reviewer variability. Genomic companion diagnostic tests such as Decipher tend to be tissue destructive, expensive, and not routinely available in most centers. We present a tissue non-destructive method for automated BCR prognosis, termed "Histotyping", that employs computational image analysis of morphologic patterns of prostate tissue from a single, routinely acquired hematoxylin and eosin slide. Patients from two institutions (n = 214) were used to train Histotyping for identifying high-risk patients based on six features of glandular morphology extracted from RP specimens. Histotyping was validated for post-RP BCR prognosis on a separate set of n = 675 patients from five institutions and compared against Decipher on n = 167 patients. Histotyping was prognostic of BCR in the validation set (p < 0.001, univariable hazard ratio [HR] = 2.83, 95% confidence interval [CI]: 2.03-3.93, concordance index [c-index] = 0.68, median years-to-BCR: 1.7). Histotyping was also prognostic in clinically stratified subsets, such as patients with Gleason grade group 3 (HR = 4.09) and negative surgical margins (HR = 3.26). Histotyping was prognostic independent of grade group, margin status, pathological stage, and preoperative prostate-specific antigen (PSA) (multivariable p < 0.001, HR = 2.09, 95% CI: 1.40-3.10, n = 648). The combination of Histotyping, grade group, and preoperative PSA outperformed Decipher (c-index = 0.75 vs. 0.70, n = 167). These results suggest that a prognostic classifier for prostate cancer based on digital images could serve as an alternative or complement to molecular-based companion diagnostic tests.

Distinguishing granulomas from adenocarcinomas by integrating stable and discriminating radiomic features on non-contrast computed tomography scans

OBJECTIVE

To identify stable and discriminating radiomic features on non-contrast CT scans to develop more generalisable radiomic classifiers for distinguishing granulomas from adenocarcinomas.

METHODS

In total, 412 patients with adenocarcinomas and granulomas from three institutions were retrospectively included. Segmentations of the lung nodules were performed manually by an expert radiologist in a 2D axial view. Radiomic features were extracted from intra- and perinodular regions. A total of 145 patients were used as part of the training set (Str), whereas 205 patients were used as part of test set I (Ste1) and 62 patients were used as part of independent test set II (Ste2). To mitigate the variation of CT acquisition parameters, we defined 'stable' radiomic features as those for which the feature expression remains relatively unchanged between different sites, as assessed using a Wilcoxon rank-sum test. These stable features were used to develop more generalisable radiomic classifiers that were more resilient to variations in lung CT scans. Features were ranked based on two criteria, firstly based on discriminability (i.e. maximising AUC) alone and subsequently based on maximising both feature stability and discriminability. Different machine-learning classifiers (Linear discriminant analysis, Quadratic discriminant analysis, Support vector machines and random forest) were trained with features selected using the two different criteria and then compared on the two independent test sets for distinguishing granulomas from adenocarcinomas, in terms of area under the receiver operating characteristic curve.

RESULTS

In the test sets, classifiers constructed using the criteria involving maximising feature stability and discriminability simultaneously achieved higher AUC compared with the discriminating alone criteria (Ste1 [n = 205]: maximum AUCs of 0.85versus . 0.80; p-value = 0.047 and Ste2 [n = 62]: maximum AUCs of 0.87 versus. 0.79; p-value = 0.021). These differences held for features extracted from scans with <3 mm slice thickness (AUC = 0.88 versus. 0.80; p-value = 0.039, n = 100) and for the ≥3 mm cases (AUC = 0.81 versus. 0.76; p-value = 0.034, n = 105). In both experiments, shape and peritumoural texture features had a higher stability compared with intratumoural texture features.

CONCLUSIONS

Our study suggests that explicitly accounting for both stability and discriminability results in more generalisable radiomic classifiers to distinguish adenocarcinomas from granulomas on non-contrast CT scans. Our results also showed that peritumoural texture and shape features were less affected by the scanner parameters compared with intratumoural texture features; however, they were also less discriminating compared with intratumoural features.

Utility of T2-weighted MRI texture analysis in assessment of peripheral zone prostate cancer aggressiveness: a single-arm, multicenter study

T₂-weighted (T₂W) MRI provides high spatial resolution and tissue-specific contrast, but it is predominantly used for qualitative evaluation of prostate anatomy and anomalies. This retrospective multicenter study evaluated the potential of T₂W image-derived textural features for quantitative assessment of peripheral zone prostate cancer (PCa) aggressiveness. A standardized preoperative multiparametric MRI was performed on 87 PCa patients across 6 institutions. T₂W intensity and apparent diffusion coefficient (ADC) histogram, and T₂W textural features were computed from tumor volumes annotated based on whole-mount histology. Spearman correlations were used to evaluate association between textural features and PCa grade groups (i.e. 1–5). Feature utility in differentiating and classifying low-(grade group 1) vs. intermediate/high-(grade group ≥ 2) aggressive cancers was evaluated using Mann–Whitney U-tests, and a support vector machine classifier employing “hold-one-institution-out” cross-validation scheme, respectively. Textural features indicating image homogeneity and disorder/complexity correlated significantly (p < 0.05) with PCa grade groups. In the intermediate/high-aggressive cancers, textural homogeneity and disorder/complexity were significantly lower and higher, respectively, compared to the low-aggressive cancers. The mean classification accuracy across the centers was highest for the combined ADC and T₂W intensity-textural features (84%) compared to ADC histogram (75%), T₂W histogram (72%), T₂W textural (72%) features alone or T₂W histogram and texture (77%), T₂W and ADC histogram (79%) combined. Texture analysis of T₂W images provides quantitative information or features that are associated with peripheral zone PCa aggressiveness and can augment their classification.

Influence of Magnetic Field Strength on Magnetic Resonance Imaging Radiomics Features in Brain Imaging, an In Vitro and In Vivo Study

BACKGROUND

The development and clinical adoption of quantitative imaging biomarkers (radiomics) has established the need for the identification of parameters altering radiomics reproducibility. The aim of this study was to assess the impact of magnetic field strength on magnetic resonance imaging (MRI) radiomics features in neuroradiology clinical practice.

METHODS

T1 3D SPGR sequence was acquired on two phantoms and 10 healthy volunteers with two clinical MR devices from the same manufacturer using two different magnetic fields (1.5 and 3T). Phantoms varied in terms of gadolinium concentrations and textural heterogeneity. 27 regions of interest were segmented (phantom: 21, volunteers: 6) using the LIFEX software. 34 features were analyzed.

RESULTS

In the phantom dataset, 10 (67%) out of 15 radiomics features were significantly different when measured at 1.5T or 3T (student's t-test, p < 0.05). Gray levels resampling, and pixel size also influence part of texture features. These findings were validated in healthy volunteers.

CONCLUSIONS

According to daily used protocols for clinical examinations, radiomic features extracted on 1.5T should not be used interchangeably with 3T when evaluating texture features. Such confounding factor should be adjusted when adapting the results of a study to a different platform, or when designing a multicentric trial.

A multicenter study on radiomic features from T2 -weighted images of a customized MR pelvic phantom setting the basis for robust radiomic models in clinics

PURPOSE

To investigate the repeatability and reproducibility of radiomic features extracted from MR images and provide a workflow to identify robust features.

METHODS

T₂ -weighted images of a pelvic phantom were acquired on three scanners of two manufacturers and two magnetic field strengths. The repeatability and reproducibility of features were assessed by the intraclass correlation coefficient and the concordance correlation coefficient, respectively, and by the within-subject coefficient of variation, considering repeated acquisitions with and without phantom repositioning, and with different scanner and acquisition parameters. The features showing intraclass correlation coefficient or concordance correlation coefficient >0.9 were selected, and their dependence on shape information (Spearman's ρ > 0.8) analyzed. They were classified for their ability to distinguish textures, after shuffling voxel intensities of images.

RESULTS

From 944 two-dimensional features, 79.9% to 96.4% showed excellent repeatability in fixed position across all scanners. A much lower range (11.2% to 85.4%) was obtained after phantom repositioning. Three-dimensional extraction did not improve repeatability performance. Excellent reproducibility between scanners was observed in 4.6% to 15.6% of the features, at fixed imaging parameters. In addition, 82.4% to 94.9% of the features showed excellent agreement when extracted from images acquired with echo times 5 ms apart, but decreased with increasing echo-time intervals, and 90.7% of the features exhibited excellent reproducibility for changes in pulse repetition time. Of nonshape features, 2.0% was identified as providing only shape information.

CONCLUSION

We showed that radiomic features are affected by MRI protocols and propose a general workflow to identify repeatable, reproducible, and informative radiomic features to ensure robustness of clinical studies.

Correction for Magnetic Field Inhomogeneities and Normalization of Voxel Values Are Needed to Better Reveal the Potential of MR Radiomic Features in Lung Cancer

Purpose: To design and validate a preprocessing procedure dedicated to T2-weighted MR images of lung cancers so as to improve the ability of radiomic features to distinguish between adenocarcinoma and other histological types.

Materials and Methods: A discovery set of 52 patients with advanced lung cancer who underwent T2-weighted MR imaging at 3 Tesla in a single center study from August 2017 to May 2019 was used. Findings were then validated using a validation set of 19 additional patients included from May to October 2019. Tumor type was obtained from the pathology report after trans-thoracic needle biopsy, metastatic lymph node or metastasis samples, or surgical excisions. MR images were preprocessed using N4ITK bias field correction and by normalizing voxel intensities with fat as a reference region. Segmentation and extraction of radiomic features were performed with LIFEx software on the raw images, on the N4ITK-corrected images and on the fully preprocessed images. Two analyses were conducted where radiomic features were extracted: (1) from the whole tumor volume (3D analysis); (2) from all slices encompassing the tumor (2D analysis). Receiver operating characteristic (ROC) analysis was used to identify features that could distinguish between adenocarcinoma and other histological types. Sham experiments were also designed to control the number of false positive findings.

Results: There were 31 (12) adenocarcinomas and 21 (7) other histological types in the discovery (validation) set. In 2D, preprocessing increased the number of discriminant radiomic features from 8 without preprocessing to 22 with preprocessing. 2D analysis yielded more features able to identify adenocarcinoma than 3D analysis (12 discriminant radiomic features after preprocessing in 3D). Preprocessing did not increase false positive findings as no discriminant features were identified in any of the sham experiments. The greatest sensitivity of the 2D analysis applied to preprocessed data was confirmed in the validation set.

Conclusion: Correction for magnetic field inhomogeneities and normalization of voxel values are essential to reveal the full potential of radiomic features to identify the tumor histological type from MR T2-weighted images, with classification performance similar to those reported in PET/CT and in multiphase CT in lung cancers.

Repeatability of radiomics and machine learning for DWI: Short-term repeatability study of 112 patients with prostate cancer

PURPOSE

To evaluate repeatability of prostate DWI-derived radiomics and machine learning methods for prostate cancer (PCa) characterization.

METHODS

A total of 112 patients with diagnosed PCa underwent 2 prostate MRI examinations (Scan1 and Scan2) performed on the same day. DWI was performed using 12 b-values (0-2000 s/mm² ), post-processed using kurtosis function, and PCa areas were annotated using whole mount prostatectomy sections. A total of 1694 radiomic features including Sobel, Kirch, Gradient, Zernike Moments, Gabor, Haralick, CoLIAGe, Haar wavelet coefficients, 3D analogue to Laws features, 2D contours, and corner detectors were calculated. Radiomics and 4 feature pruning methods (area under the receiver operator characteristic curve, maximum relevance minimum redundancy, Spearman's ρ, Wilcoxon rank-sum) were evaluated in terms of Scan1-Scan2 repeatability using intraclass correlation coefficient (ICC)(3,1). Classification performance for clinically significant and insignificant PCa with Gleason grade groups 1 versus >1 was evaluated by area under the receiver operator characteristic curve in unseen random 30% data split.

RESULTS

The ICC(3,1) values for conventional radiomics and feature pruning methods were in the range of 0.28-0.90. The machine learning classifications varied between Scan1 and Scan2 with % of same class labels between Scan1 and Scan2 in the range of 61-81%. Surface-to-volume ratio and corner detector-based features were among the most represented features with high repeatability, ICC(3,1) >0.75, consistently high ranking using all 4 feature pruning methods, and classification performance with area under the receiver operator characteristic curve >0.70.

CONCLUSION

Surface-to-volume ratio and corner detectors for prostate DWI led to good classification of unseen data and performed similarly in Scan1 and Scan2 in contrast to multiple conventional radiomic features.