Test-retest repeatability of a deep learning architecture in detecting and segmenting clinically significant prostate cancer on apparent diffusion coefficient (ADC) maps

An integrated radiology-pathology machine learning classifier for outcome prediction following radical prostatectomy: Preliminary findings

Objectives

To evaluate the added benefit of integrating features from pre-treatment MRI (radiomics) and digitized post-surgical pathology slides (pathomics) in prostate cancer (PCa) patients for prognosticating outcomes post radical-prostatectomy (RP) including a) rising prostate specific antigen (PSA), and b) extraprostatic-extension (EPE).

Methods

Multi-institutional data (N = 58) of PCa patients who underwent pre-treatment 3-T MRI prior to RP were included in this retrospective study. Radiomic and pathomic features were extracted from PCa regions on MRI and RP specimens delineated by expert clinicians. On training set (D1, N = 44), Cox Proportional-Hazards models MR, MP and MRaP were trained using radiomics, pathomics, and their combination, respectively, to prognosticate rising PSA (PSA > 0.03 ng/mL). Top features from MRaP were used to train a model to predict EPE on D1 and test on external dataset (D2, N = 14). C-index, Kalplan-Meier curves were used for survival analysis, and area under ROC (AUC) was used for EPE. MRaP was compared with the existing post-treatment risk-calculator, CAPRA (MC).

Results

Patients had median follow-up of 34 months. MRaP (c-index = 0.685 ± 0.05) significantly outperformed MR (c-index = 0.646 ± 0.05), MP (c-index = 0.631 ± 0.06) and MC (c-index = 0.601 ± 0.071) (p < 0.0001). Cross-validated Kaplan-Meier curves showed significant separation among risk groups for rising PSA for MRaP (p < 0.005, Hazard Ratio (HR) = 11.36) as compared to MR (p = 0.64, HR = 1.33), MP (p = 0.19, HR = 2.82) and MC (p = 0.10, HR = 3.05). Integrated radio-pathomic model MRaP (AUC = 0.80) outperformed MR (AUC = 0.57) and MP (AUC = 0.76) in predicting EPE on external-data (D2).

Conclusions

Results from this preliminary study suggest that a combination of radiomic and pathomic features can better predict post-surgical outcomes (rising PSA and EPE) compared to either of them individually as well as extant prognostic nomogram (CAPRA).

Advantage of whole-mount histopathology in prostate cancer: current applications and future prospects

BACKGROUND

Whole-mount histopathology (WMH) has been a powerful tool to investigate the characteristics of prostate cancer. However, the latest advancement of WMH was yet under summarization. In this review, we offer a comprehensive exposition of current research utilizing WMH in diagnosing and treating prostate cancer (PCa), and summarize the clinical advantages of WMH and outlines potential on future prospects.

METHODS

An extensive PubMed search was conducted until February 26, 2023, with the search term "prostate", "whole-mount", "large format histology", which was limited to the last 4 years. Publications included were restricted to those in English. Other papers were also cited to contribute a better understanding.

RESULTS

WMH exhibits an enhanced legibility for pathologists, which improved the efficacy of pathologic examination and provide educational value. It simplifies the histopathological registration with medical images, which serves as a convincing reference standard for imaging indicator investigation and medical image-based artificial intelligence (AI). Additionally, WMH provides comprehensive histopathological information for tumor volume estimation, post-treatment evaluation, and provides direct pathological data for AI readers. It also offers complete spatial context for the location estimation of both intraprostatic and extraprostatic cancerous region.

CONCLUSIONS

WMH provides unique benefits in several aspects of clinical diagnosis and treatment of PCa. The utilization of WMH technique facilitates the development and refinement of various clinical technologies. We believe that WMH will play an important role in future clinical applications.

Deep-Learning Models for Detection and Localization of Visible Clinically Significant Prostate Cancer on Multi-Parametric MRI

BACKGROUND

Deep learning for diagnosing clinically significant prostate cancer (csPCa) is feasible but needs further evaluation in patients with prostate-specific antigen (PSA) levels of 4-10 ng/mL.

PURPOSE

To explore diffusion-weighted imaging (DWI), alone and in combination with T2-weighted imaging (T2WI), for deep-learning-based models to detect and localize visible csPCa.

STUDY TYPE

Retrospective.

POPULATION

One thousand six hundred twenty-eight patients with systematic and cognitive-targeted biopsy-confirmation (1007 csPCa, 621 non-csPCa) were divided into model development (N = 1428) and hold-out test (N = 200) datasets.

FIELD STRENGTH/SEQUENCE

DWI with diffusion-weighted single-shot gradient echo planar imaging sequence and T2WI with T2-weighted fast spin echo sequence at 3.0-T and 1.5-T.

ASSESSMENT

The ground truth of csPCa was annotated by two radiologists in consensus. A diffusion model, DWI and apparent diffusion coefficient (ADC) as input, and a biparametric model (DWI, ADC, and T2WI as input) were trained based on U-Net. Three radiologists provided the PI-RADS (version 2.1) assessment. The performances were determined at the lesion, location, and the patient level.

STATISTICAL TESTS

The performance was evaluated using the areas under the ROC curves (AUCs), sensitivity, specificity, and accuracy. A P value <0.05 was considered statistically significant.

RESULTS

The lesion-level sensitivities of the diffusion model, the biparametric model, and the PI-RADS assessment were 89.0%, 85.3%, and 90.8% (P = 0.289-0.754). At the patient level, the diffusion model had significantly higher sensitivity than the biparametric model (96.0% vs. 90.0%), while there was no significant difference in specificity (77.0%. vs. 85.0%, P = 0.096). For location analysis, there were no significant differences in AUCs between the models (sextant-level, 0.895 vs. 0.893, P = 0.777; zone-level, 0.931 vs. 0.917, P = 0.282), and both models had significantly higher AUCs than the PI-RADS assessment (sextant-level, 0.734; zone-level, 0.863).

DATA CONCLUSION

The diffusion model achieved the best performance in detecting and localizing csPCa in patients with PSA levels of 4-10 ng/mL.

EVIDENCE LEVEL

3 TECHNICAL EFFICACY: Stage 2.

Automated deep-learning system in the assessment of MRI-visible prostate cancer: comparison of advanced zoomed diffusion-weighted imaging and conventional technique

BACKGROUND

Deep-learning-based computer-aided diagnosis (DL-CAD) systems using MRI for prostate cancer (PCa) detection have demonstrated good performance. Nevertheless, DL-CAD systems are vulnerable to high heterogeneities in DWI, which can interfere with DL-CAD assessments and impair performance. This study aims to compare PCa detection of DL-CAD between zoomed-field-of-view echo-planar DWI (z-DWI) and full-field-of-view DWI (f-DWI) and find the risk factors affecting DL-CAD diagnostic efficiency.

METHODS

This retrospective study enrolled 354 consecutive participants who underwent MRI including T2WI, f-DWI, and z-DWI because of clinically suspected PCa. A DL-CAD was used to compare the performance of f-DWI and z-DWI both on a patient level and lesion level. We used the area under the curve (AUC) of receiver operating characteristics analysis and alternative free-response receiver operating characteristics analysis to compare the performances of DL-CAD using f- DWI and z-DWI. The risk factors affecting the DL-CAD were analyzed using logistic regression analyses. P values less than 0.05 were considered statistically significant.

RESULTS

DL-CAD with z-DWI had a significantly better overall accuracy than that with f-DWI both on patient level and lesion level (AUC_patient: 0.89 vs. 0.86; AUC_lesion: 0.86 vs. 0.76; P < .001). The contrast-to-noise ratio (CNR) of lesions in DWI was an independent risk factor of false positives (odds ratio [OR] = 1.12; P < .001). Rectal susceptibility artifacts, lesion diameter, and apparent diffusion coefficients (ADC) were independent risk factors of both false positives (OR_{rectal susceptibility artifact} = 5.46; OR_diameter, = 1.12; OR_ADC = 0.998; all P < .001) and false negatives (OR_{rectal susceptibility artifact} = 3.31; OR_diameter = 0.82; OR_ADC = 1.007; all P ≤ .03) of DL-CAD.

CONCLUSIONS

Z-DWI has potential to improve the detection performance of a prostate MRI based DL-CAD.

TRIAL REGISTRATION

ChiCTR, NO. ChiCTR2100041834 . Registered 7 January 2021.

Improving the repeatability of deep learning models with Monte Carlo dropout

The integration of artificial intelligence into clinical workflows requires reliable and robust models. Repeatability is a key attribute of model robustness. Ideal repeatable models output predictions without variation during independent tests carried out under similar conditions. However, slight variations, though not ideal, may be unavoidable and acceptable in practice. During model development and evaluation, much attention is given to classification performance while model repeatability is rarely assessed, leading to the development of models that are unusable in clinical practice. In this work, we evaluate the repeatability of four model types (binary classification, multi-class classification, ordinal classification, and regression) on images that were acquired from the same patient during the same visit. We study the each model's performance on four medical image classification tasks from public and private datasets: knee osteoarthritis, cervical cancer screening, breast density estimation, and retinopathy of prematurity. Repeatability is measured and compared on ResNet and DenseNet architectures. Moreover, we assess the impact of sampling Monte Carlo dropout predictions at test time on classification performance and repeatability. Leveraging Monte Carlo predictions significantly increases repeatability, in particular at the class boundaries, for all tasks on the binary, multi-class, and ordinal models leading to an average reduction of the 95% limits of agreement by 16% points and of the class disagreement rate by 7% points. The classification accuracy improves in most settings along with the repeatability. Our results suggest that beyond about 20 Monte Carlo iterations, there is no further gain in repeatability. In addition to the higher test-retest agreement, Monte Carlo predictions are better calibrated which leads to output probabilities reflecting more accurately the true likelihood of being correctly classified.

Pseudoprospective Paraclinical Interaction of Radiology Residents With a Deep Learning System for Prostate Cancer Detection: Experience, Performance, and Identification of the Need for Intermittent Recalibration

OBJECTIVES

The aim of this study was to estimate the prospective utility of a previously retrospectively validated convolutional neural network (CNN) for prostate cancer (PC) detection on prostate magnetic resonance imaging (MRI).

MATERIALS AND METHODS

The biparametric (T2-weighted and diffusion-weighted) portion of clinical multiparametric prostate MRI from consecutive men included between November 2019 and September 2020 was fully automatically and individually analyzed by a CNN briefly after image acquisition (pseudoprospective design). Radiology residents performed 2 research Prostate Imaging Reporting and Data System (PI-RADS) assessments of the multiparametric dataset independent from clinical reporting (paraclinical design) before and after review of the CNN results and completed a survey. Presence of clinically significant PC was determined by the presence of an International Society of Urological Pathology grade 2 or higher PC on combined targeted and extended systematic transperineal MRI/transrectal ultrasound fusion biopsy. Sensitivities and specificities on a patient and prostate sextant basis were compared using the McNemar test and compared with the receiver operating characteristic (ROC) curve of CNN. Survey results were summarized as absolute counts and percentages.

RESULTS

A total of 201 men were included. The CNN achieved an ROC area under the curve of 0.77 on a patient basis. Using PI-RADS ≥3-emulating probability threshold (c3), CNN had a patient-based sensitivity of 81.8% and specificity of 54.8%, not statistically different from the current clinical routine PI-RADS ≥4 assessment at 90.9% and 54.8%, respectively ( P = 0.30/ P = 1.0). In general, residents achieved similar sensitivity and specificity before and after CNN review. On a prostate sextant basis, clinical assessment possessed the highest ROC area under the curve of 0.82, higher than CNN (AUC = 0.76, P = 0.21) and significantly higher than resident performance before and after CNN review (AUC = 0.76 / 0.76, P ≤ 0.03). The resident survey indicated CNN to be helpful and clinically useful.

CONCLUSIONS

Pseudoprospective paraclinical integration of fully automated CNN-based detection of suspicious lesions on prostate multiparametric MRI was demonstrated and showed good acceptance among residents, whereas no significant improvement in resident performance was found. General CNN performance was preserved despite an observed shift in CNN calibration, identifying the requirement for continuous quality control and recalibration.

Stable and Discriminatory Radiomic Features from the Tumor and Its Habitat Associated with Progression-Free Survival in Glioblastoma: A Multi-Institutional Study

BACKGROUND AND PURPOSE

Glioblastoma is an aggressive brain tumor, with no validated prognostic biomarkers for survival before surgical resection. Although recent approaches have demonstrated the prognostic ability of tumor habitat (constituting necrotic core, enhancing lesion, T2/FLAIR hyperintensity subcompartments) derived radiomic features for glioblastoma survival on treatment-naive MR imaging scans, radiomic features are known to be sensitive to MR imaging acquisitions across sites and scanners. In this study, we sought to identify the radiomic features that are both stable across sites and discriminatory of poor and improved progression-free survival in glioblastoma tumors.

MATERIALS AND METHODS

We used 150 treatment-naive glioblastoma MR imaging scans (Gadolinium-T1w, T2w, FLAIR) obtained from 5 sites. For every tumor subcompartment (enhancing tumor, peritumoral FLAIR-hyperintensities, necrosis), a total of 316 three-dimensional radiomic features were extracted. The training cohort constituted studies from 4 sites (n = 93) to select the most stable and discriminatory radiomic features for every tumor subcompartment. These features were used on a hold-out cohort (n = 57) to evaluate their ability to discriminate patients with poor survival from those with improved survival.

RESULTS

Incorporating the most stable and discriminatory features within a linear discriminant analysis classifier yielded areas under the curve of 0.71, 0.73, and 0.76 on the test set for distinguishing poor and improved survival compared with discriminatory features alone (areas under the curve of 0.65, 0.54, 0.62) from the necrotic core, enhancing tumor, and peritumoral T2/FLAIR hyperintensity, respectively.

CONCLUSIONS

Incorporating stable and discriminatory radiomic features extracted from tumors and associated habitats across multisite MR imaging sequences may yield robust prognostic classifiers of patient survival in glioblastoma tumors.

Role of Deep Learning in Prostate Cancer Management: Past, Present and Future Based on a Comprehensive Literature Review

This review aims to present the applications of deep learning (DL) in prostate cancer diagnosis and treatment. Computer vision is becoming an increasingly large part of our daily lives due to advancements in technology. These advancements in computational power have allowed more extensive and more complex DL models to be trained on large datasets. Urologists have found these technologies help them in their work, and many such models have been developed to aid in the identification, treatment and surgical practices in prostate cancer. This review will present a systematic outline and summary of these deep learning models and technologies used for prostate cancer management. A literature search was carried out for English language articles over the last two decades from 2000–2021, and present in Scopus, MEDLINE, Clinicaltrials.gov, Science Direct, Web of Science and Google Scholar. A total of 224 articles were identified on the initial search. After screening, 64 articles were identified as related to applications in urology, from which 24 articles were identified to be solely related to the diagnosis and treatment of prostate cancer. The constant improvement in DL models should drive more research focusing on deep learning applications. The focus should be on improving models to the stage where they are ready to be implemented in clinical practice. Future research should prioritize developing models that can train on encrypted images, allowing increased data sharing and accessibility.

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.

Current Value of Biparametric Prostate MRI with Machine-Learning or Deep-Learning in the Detection, Grading, and Characterization of Prostate Cancer: A Systematic Review

Prostate cancer detection with magnetic resonance imaging is based on a standardized MRI-protocol according to the PI-RADS guidelines including morphologic imaging, diffusion weighted imaging, and perfusion. To facilitate data acquisition and analysis the contrast-enhanced perfusion is often omitted resulting in a biparametric prostate MRI protocol. The intention of this review is to analyze the current value of biparametric prostate MRI in combination with methods of machine-learning and deep learning in the detection, grading, and characterization of prostate cancer; if available a direct comparison with human radiologist performance was performed. PubMed was systematically queried and 29 appropriate studies were identified and retrieved. The data show that detection of clinically significant prostate cancer and differentiation of prostate cancer from non-cancerous tissue using machine-learning and deep learning is feasible with promising results. Some techniques of machine-learning and deep-learning currently seem to be equally good as human radiologists in terms of classification of single lesion according to the PIRADS score.

The Primacy of High B-Value 3T-DWI Radiomics in the Prediction of Clinically Significant Prostate Cancer

Predicting clinically significant prostate cancer (csPCa) is crucial in PCa management. 3T-magnetic resonance (MR) systems may have a novel role in quantitative imaging and early csPCa prediction, accordingly. In this study, we develop a radiomic model for predicting csPCa based solely on native b2000 diffusion weighted imaging (DWI_b2000) and debate the effectiveness of apparent diffusion coefficient (ADC) in the same task. In total, 105 patients were retrospectively enrolled between January–November 2020, with confirmed csPCa or ncsPCa based on biopsy. DWI_b2000 and ADC images acquired with a 3T-MRI were analyzed by computing 84 local first-order radiomic features (RFs). Two predictive models were built based on DWI_b2000 and ADC, separately. Relevant RFs were selected through LASSO, a support vector machine (SVM) classifier was trained using repeated 3-fold cross validation (CV) and validated on a holdout set. The SVM models rely on a single couple of uncorrelated RFs (ρ < 0.15) selected through Wilcoxon rank-sum test (p ≤ 0.05) with Holm–Bonferroni correction. On the holdout set, while the ADC model yielded AUC = 0.76 (95% CI, 0.63–0.96), the DWI_b2000 model reached AUC = 0.84 (95% CI, 0.63–0.90), with specificity = 75%, sensitivity = 90%, and informedness = 0.65. This study establishes the primary role of 3T-DWI_b2000 in PCa quantitative analyses, whilst ADC can remain the leading sequence for detection.