Evaluation of a Cascaded Deep Learning-based Algorithm for Prostate Lesion Detection at Biparametric MRI.

Background Multiparametric MRI (mpMRI) improves prostate cancer (PCa) detection compared with systematic biopsy, but its interpretation is prone to interreader variation, which results in performance inconsistency. Artificial intelligence (AI) models can assist in mpMRI interpretation, but large training data sets and extensive model testing are required. Purpose To evaluate a biparametric MRI AI algorithm for intraprostatic lesion detection and segmentation and to compare its performance with radiologist readings and biopsy results. Materials and Methods This secondary analysis of a prospective registry included consecutive patients with suspected or known PCa who underwent mpMRI, US-guided systematic biopsy, or combined systematic and MRI/US fusion-guided biopsy between April 2019 and September 2022. All lesions were prospectively evaluated using Prostate Imaging Reporting and Data System version 2.1. The lesion- and participant-level performance of a previously developed cascaded deep learning algorithm was compared with histopathologic outcomes and radiologist readings using sensitivity, positive predictive value (PPV), and Dice similarity coefficient (DSC). Results A total of 658 male participants (median age, 67 years [IQR, 61-71 years]) with 1029 MRI-visible lesions were included. At histopathologic analysis, 45% (294 of 658) of participants had lesions of International Society of Urological Pathology (ISUP) grade group (GG) 2 or higher. The algorithm identified 96% (282 of 294; 95% CI: 94%, 98%) of all participants with clinically significant PCa, whereas the radiologist identified 98% (287 of 294; 95% CI: 96%, 99%; P = .23). The algorithm identified 84% (103 of 122), 96% (152 of 159), 96% (47 of 49), 95% (38 of 40), and 98% (45 of 46) of participants with ISUP GG 1, 2, 3, 4, and 5 lesions, respectively. In the lesion-level analysis using radiologist ground truth, the detection sensitivity was 55% (569 of 1029; 95% CI: 52%, 58%), and the PPV was 57% (535 of 934; 95% CI: 54%, 61%). The mean number of false-positive lesions per participant was 0.61 (range, 0-3). The lesion segmentation DSC was 0.29. Conclusion The AI algorithm detected cancer-suspicious lesions on biparametric MRI scans with a performance comparable to that of an experienced radiologist. Moreover, the algorithm reliably predicted clinically significant lesions at histopathologic examination. ClinicalTrials.gov Identifier: NCT03354416 © RSNA, 2024 Supplemental material is available for this article.

Non-Invasive Tumor Grade Evaluation in Von Hippel-Lindau-Associated Clear Cell Renal Cell Carcinoma: A Magnetic Resonance Imaging-Based Study.

BACKGROUND: Pathology grading is an essential step for the treatment and evaluation of the prognosis in patients with clear cell renal cell carcinoma (ccRCC). PURPOSE: To investigate the utility of texture analysis in evaluating Fuhrman grades of renal tumors in patients with Von Hippel-Lindau (VHL)-associated ccRCC, aiming to improve non-invasive diagnosis and personalized treatment. STUDY TYPE: Retrospective analysis of a prospectively maintained cohort. POPULATION: One hundred and thirty-six patients, 84 (61%) males and 52 (39%) females with pathology-proven ccRCC with a mean age of 52.8 ± 12.7 from 2010 to 2023. FIELD STRENGTH AND SEQUENCES: 1.5 and 3 T MRIs. Segmentations were performed on the T1-weighted 3-minute delayed sequence and then registered on pre-contrast, T1-weighted arterial and venous sequences. ASSESSMENT: A total of 404 lesions, 345 low-grade tumors, and 59 high-grade tumors were segmented using ITK-SNAP on a T1-weighted 3-minute delayed sequence of MRI. Radiomics features were extracted from pre-contrast, T1-weighted arterial, venous, and delayed post-contrast sequences. Preprocessing techniques were employed to address class imbalances. Features were then rescaled to normalize the numeric values. We developed a stacked model combining random forest and XGBoost to assess tumor grades using radiomics signatures. STATISTICAL TESTS: The model's performance was evaluated using positive predictive value (PPV), sensitivity, F1 score, area under the curve of receiver operating characteristic curve, and Matthews correlation coefficient. Using Monte Carlo technique, the average performance of 100 benchmarks of 85% train and 15% test was reported. RESULTS: The best model displayed an accuracy of 0.79. For low-grade tumor detection, a sensitivity of 0.79, a PPV of 0.95, and an F1 score of 0.86 were obtained. For high-grade tumor detection, a sensitivity of 0.78, PPV of 0.39, and F1 score of 0.52 were reported. DATA CONCLUSION: Radiomics analysis shows promise in classifying pathology grades non-invasively for patients with VHL-associated ccRCC, potentially leading to better diagnosis and personalized treatment. LEVEL OF EVIDENCE: 1 TECHNICAL EFFICACY: Stage 2.

PI-RADS Version 2.0 Versus Version 2.1: Comparison of Prostate Cancer Gleason Grade Upgrade and Downgrade Rates From MRI-Targeted Biopsy to Radical Prostatectomy.

BACKGROUND. Precise risk stratification through MRI/ultrasound (US) fusion-guided targeted biopsy (TBx) can guide optimal prostate cancer (PCa) management. OBJECTIVE. The purpose of this study was to compare PI-RADS version 2.0 (v2.0) and PI-RADS version 2.1 (v2.1) in terms of the rates of International Society of Urological Pathology (ISUP) grade group (GG) upgrade and downgrade from TBx to radical prostatectomy (RP). METHODS. This study entailed a retrospective post hoc analysis of patients who underwent 3-T prostate MRI at a single institution from May 2015 to March 2023 as part of three prospective clinical trials. Trial participants who underwent MRI followed by MRI/US fusion-guided TBx and RP within a 1-year interval were identified. A single genitourinary radiologist performed clinical interpretations of the MRI examinations using PI-RADS v2.0 from May 2015 to March 2019 and PI-RADS v2.1 from April 2019 to March 2023. Upgrade and downgrade rates from TBx to RP were compared using chi-square tests. Clinically significant cancer was defined as ISUP GG2 or greater. RESULTS. The final analysis included 308 patients (median age, 65 years; median PSA density, 0.16 ng/mL2). The v2.0 group (n = 177) and v2.1 group (n = 131) showed no significant difference in terms of upgrade rate (29% vs 22%, respectively; p = .15), downgrade rate (19% vs 21%, p = .76), clinically significant upgrade rate (14% vs 10%, p = .27), or clinically significant downgrade rate (1% vs 1%, p > .99). The upgrade rate and downgrade rate were also not significantly different between the v2.0 and v2.1 groups when stratifying by index lesion PI-RADS category or index lesion zone, as well as when assessed only in patients without a prior PCa diagnosis (all p > .01). Among patients with GG2 or GG3 at RP (n = 121 for v2.0; n = 103 for v2.1), the concordance rate between TBx and RP was not significantly different between the v2.0 and v2.1 groups (53% vs 57%, p = .51). CONCLUSION. Upgrade and downgrade rates from TBx to RP were not significantly different between patients whose MRI examinations were clinically interpreted using v2.0 or v2.1. CLINICAL IMPACT. Implementation of the most recent PI-RADS update did not improve the incongruence in PCa grade assessment between TBx and surgery.

Deep Learning-Based T2-weighted MR Image Quality Assessment and Its Impact on Prostate Cancer Detection Rates.

BACKGROUND: Image quality evaluation of prostate MRI is important for successful implementation of MRI into localized prostate cancer diagnosis. PURPOSE: To examine the impact of image quality on prostate cancer detection using an in-house previously developed artificial intelligence (AI) algorithm. STUDY TYPE: Retrospective. SUBJECTS: 615 consecutive patients (median age 67 [interquartile range [IQR]: 61-71] years) with elevated serum PSA (median PSA 6.6 [IQR: 4.6-9.8] ng/mL) prior to prostate biopsy. FIELD STRENGTH/SEQUENCE: 3.0T/T2-weighted turbo-spin-echo MRI, high b-value echo-planar diffusion-weighted imaging, and gradient recalled echo dynamic contrast-enhanced. ASSESSMENTS: Scans were prospectively evaluated during clinical readout using PI-RADSv2.1 by one genitourinary radiologist with 17 years of experience. For each patient, T2-weighted images (T2WIs) were classified as high-quality or low-quality based on evaluation of both general distortions (eg, motion, distortion, noise, and aliasing) and perceptual distortions (eg, obscured delineation of prostatic capsule, prostatic zones, and excess rectal gas) by a previously developed in-house AI algorithm. Patients with PI-RADS category 1 underwent 12-core ultrasound-guided systematic biopsy while those with PI-RADS category 2-5 underwent combined systematic and targeted biopsies. Patient-level cancer detection rates (CDRs) were calculated for clinically significant prostate cancer (csPCa, International Society of Urological Pathology Grade Group ≥2) by each biopsy method and compared between high- and low-quality images in each PI-RADS category. STATISTICAL TESTS: Fisher's exact test. Bootstrap 95% confidence intervals (CI). A P value <0.05 was considered statistically significant. RESULTS: 385 (63%) T2WIs were classified as high-quality and 230 (37%) as low-quality by AI. Targeted biopsy with high-quality T2WIs resulted in significantly higher clinically significant CDR than low-quality images for PI-RADS category 4 lesions (52% [95% CI: 43-61] vs. 32% [95% CI: 22-42]). For combined biopsy, there was no significant difference in patient-level CDRs for PI-RADS 4 between high- and low-quality T2WIs (56% [95% CI: 47-64] vs. 44% [95% CI: 34-55]; P = 0.09). DATA CONCLUSION: Higher quality T2WIs were associated with better targeted biopsy clinically significant cancer detection performance for PI-RADS 4 lesions. Combined biopsy might be needed when T2WI is lower quality. LEVEL OF EVIDENCE: 2 TECHNICAL EFFICACY: Stage 1.

Comparison of MRI-Based Staging and Pathologic Staging for Predicting Biochemical Recurrence of Prostate Cancer After Radical Prostatectomy.

BACKGROUND. Currently most clinical models for predicting biochemical recurrence (BCR) of prostate cancer (PCa) after radical prostatectomy (RP) incorporate staging information from RP specimens, creating a gap in preoperative risk assessment. OBJECTIVE. The purpose of our study was to compare the utility of presurgical staging information from MRI and postsurgical staging information from RP pathology in predicting BCR in patients with PCa. METHODS. This retrospective study included 604 patients (median age, 60 years) with PCa who underwent prostate MRI before RP from June 2007 to December 2018. A single genitourinary radiologist assessed MRI examinations for extraprostatic extension (EPE) and seminal vesicle invasion (SVI) during clinical interpretations. The utility of EPE and SVI on MRI and RP pathology for BCR prediction was assessed through Kaplan-Meier and Cox proportional hazards analyses. Established clinical BCR prediction models, including the University of California San Francisco Cancer of the Prostate Risk Assessment (UCSF-CAPRA) model and the Cancer of the Prostate Risk Assessment Postsurgical (CAPRA-S) model, were evaluated in a subset of 374 patients with available Gleason grade groups from biopsy and RP pathology; two CAPRA-MRI models (CAPRA-S model with modifications to replace RP pathologic staging features with MRI staging features) were also assessed. RESULTS. Univariable predictors of BCR included EPE on MRI (HR = 3.6), SVI on MRI (HR = 4.4), EPE on RP pathology (HR = 5.0), and SVI on RP pathology (HR = 4.6) (all p < .001). Three-year BCR-free survival (RFS) rates for patients without versus with EPE were 84% versus 59% for MRI and 89% versus 58% for RP pathology, and 3-year RFS rates for patients without versus with SVI were 82% versus 50% for MRI and 83% versus 54% for RP histology (all p < .001). For patients with T3 disease on RP pathology, 3-year RFS rates were 67% and 41% for patients without and with T3 disease on MRI. AUCs of CAPRA models, including CAPRA-MRI models, ranged from 0.743 to 0.778. AUCs were not significantly different between CAPRA-S and CAPRA-MRI models (p > .05). RFS rates were significantly different between low- and intermediate-risk groups for only CAPRA-MRI models (80% vs 51% and 74% vs 44%; both p < .001). CONCLUSION. Presurgical MRI-based staging features perform comparably to postsurgical pathologic staging features for predicting BCR. CLINICAL IMPACT. MRI staging can preoperatively identify patients at high BCR risk, helping to inform early clinical decision-making. TRIAL REGISTRATION. ClinicalTrials.gov NCT00026884 and NCT02594202.

Multicenter Multireader Evaluation of an Artificial Intelligence-Based Attention Mapping System for the Detection of Prostate Cancer With Multiparametric MRI.

OBJECTIVE. The purpose of this study was to evaluate in a multicenter dataset the performance of an artificial intelligence (AI) detection system with attention mapping compared with multiparametric MRI (mpMRI) interpretation in the detection of prostate cancer. MATERIALS AND METHODS. MRI examinations from five institutions were included in this study and were evaluated by nine readers. In the first round, readers evaluated mpMRI studies using the Prostate Imaging Reporting and Data System version 2. After 4 weeks, images were again presented to readers along with the AI-based detection system output. Readers accepted or rejected lesions within four AI-generated attention map boxes. Additional lesions outside of boxes were excluded from detection and categorization. The performances of readers using the mpMRI-only and AI-assisted approaches were compared. RESULTS. The study population included 152 case patients and 84 control patients with 274 pathologically proven cancer lesions. The lesion-based AUC was 74.9% for MRI and 77.5% for AI with no significant difference (p = 0.095). The sensitivity for overall detection of cancer lesions was higher for AI than for mpMRI but did not reach statistical significance (57.4% vs 53.6%, p = 0.073). However, for transition zone lesions, sensitivity was higher for AI than for MRI (61.8% vs 50.8%, p = 0.001). Reading time was longer for AI than for MRI (4.66 vs 4.03 minutes, p < 0.001). There was moderate interreader agreement for AI and MRI with no significant difference (58.7% vs 58.5%, p = 0.966). CONCLUSION. Overall sensitivity was only minimally improved by use of the AI system. Significant improvement was achieved, however, in the detection of transition zone lesions with use of the AI system at the cost of a mean of 40 seconds of additional reading time.

Interreader Variability of Prostate Imaging Reporting and Data System Version 2 in Detecting and Assessing Prostate Cancer Lesions at Prostate MRI.

OBJECTIVE. The purpose of this study was to evaluate agreement among radiologists in detecting and assessing prostate cancer at multiparametric MRI using Prostate Imaging Reporting and Data System version 2 (PI-RADSv2). MATERIALS AND METHODS. Treatment-naïve patients underwent 3-T multipara-metric MRI between April 2012 and June 2015. Among the 163 patients evaluated, 110 underwent prostatectomy after MRI and 53 had normal MRI findings and transrectal ultrasound-guided biopsy results. Nine radiologists participated (three each with high, intermediate, and low levels of experience). Readers interpreted images of 58 patients on average (range, 56-60) using PI-RADSv2. Prostatectomy specimens registered to MRI were ground truth. Interob-server agreement was evaluated with the index of specific agreement for lesion detection and kappa and proportion of agreement for PI-RADS category assignment. RESULTS. The radiologists detected 336 lesions. Sensitivity for index lesions was 80.9% (95% CI, 75.1-85.9%), comparable across reader experience (p = 0.392). Patient-level specificity was experience dependent; highly experienced readers had 84.0% specificity versus 55.2% for all others (p < 0.001). Interobserver agreement was excellent for detecting index lesions (index of specific agreement, 0.871; 95% CI, 0.798-0.923). Agreement on PI-RADSv2 category assignment of index lesions was moderate (κ = 0.419; 95% CI, 0.238-0.595). For individual category assignments, proportion of agreement was slight for PI-RADS category 3 (0.208; 95% CI, 0.086-0.284) but substantial for PI-RADS category 4 (0.674; 95% CI, 0.540-0.776). However, proportion of agreement for T2-weighted PI-RADS 4 in the transition zone was 0.250 (95% CI, 0.108-0.372). Proportion of agreement for category assignment of index lesions on dynamic contrast-enhanced MR images was 0.822 (95% CI, 0.728-0.903), on T2-weighted MR images was 0.515 (95% CI, 0.430-0623), and on DW images was 0.586 (95% CI, 0.495-0.682). Proportion of agreement for dominant lesion was excellent (0.828; 95% CI, 0.742-0.913). CONCLUSION. Radiologists across experience levels had excellent agreement for detecting index lesions and moderate agreement for category assignment of lesions using PI-RADS. Future iterations of PI-RADS should clarify PI-RADS 3 and PI-RADS 4 in the transition zone.

Computer-aided diagnosis prior to conventional interpretation of prostate mpMRI: an international multi-reader study.

OBJECTIVES: To evaluate if computer-aided diagnosis (CAD) prior to prostate multi-parametric MRI (mpMRI) can improve sensitivity and agreement between radiologists. METHODS: Nine radiologists (three each high, intermediate, low experience) from eight institutions participated. A total of 163 patients with 3-T mpMRI from 4/2012 to 6/2015 were included: 110 cancer patients with prostatectomy after mpMRI, 53 patients with no lesions on mpMRI and negative TRUS-guided biopsy. Readers were blinded to all outcomes and detected lesions per PI-RADSv2 on mpMRI. After 5 weeks, readers re-evaluated patients using CAD to detect lesions. Prostatectomy specimens registered to MRI were ground truth with index lesions defined on pathology. Sensitivity, specificity and agreement were calculated per patient, lesion level and zone-peripheral (PZ) and transition (TZ). RESULTS: Index lesion sensitivity was 78.2% for mpMRI alone and 86.3% for CAD-assisted mpMRI (p = 0.013). Sensitivity was comparable for TZ lesions (78.7% vs 78.1%; p = 0.929); CAD improved PZ lesion sensitivity (84% vs 94%; p = 0.003). Improved sensitivity came from lesions scored PI-RADS < 3 as index lesion sensitivity was comparable at PI-RADS ≥ 3 (77.6% vs 78.1%; p = 0.859). Per patient specificity was 57.1% for CAD and 70.4% for mpMRI (p = 0.003). CAD improved agreement between all readers (56.9% vs 71.8%; p < 0.001). CONCLUSIONS: CAD-assisted mpMRI improved sensitivity and agreement, but decreased specificity, between radiologists of varying experience. KEY POINTS: • Computer-aided diagnosis (CAD) assists clinicians in detecting prostate cancer on MRI. • CAD assistance improves agreement between radiologists in detecting prostate cancer lesions. • However, this CAD system induces more false positives, particularly for less-experienced clinicians and in the transition zone. • CAD assists radiologists in detecting cancer missed on MRI, suggesting a path for improved diagnostic confidence.

Validation of the Dominant Sequence Paradigm and Role of Dynamic Contrast-enhanced Imaging in PI-RADS Version 2.

Purpose To validate the dominant pulse sequence paradigm and limited role of dynamic contrast material-enhanced magnetic resonance (MR) imaging in the Prostate Imaging Reporting and Data System (PI-RADS) version 2 for prostate multiparametric MR imaging by using data from a multireader study. Materials and Methods This HIPAA-compliant retrospective interpretation of prospectively acquired data was approved by the local ethics committee. Patients were treatment-naïve with endorectal coil 3-T multiparametric MR imaging. A total of 163 patients were evaluated, 110 with prostatectomy after multiparametric MR imaging and 53 with negative multiparametric MR imaging and systematic biopsy findings. Nine radiologists participated in this study and interpreted images in 58 patients, on average (range, 56-60 patients). Lesions were detected with PI-RADS version 2 and were compared with whole-mount prostatectomy findings. Probability of cancer detection for overall, T2-weighted, and diffusion-weighted (DW) imaging PI-RADS scores was calculated in the peripheral zone (PZ) and transition zone (TZ) by using generalized estimating equations. To determine dominant pulse sequence and benefit of dynamic contrast-enhanced (DCE) imaging, odds ratios (ORs) were calculated as the ratio of odds of cancer of two consecutive scores by logistic regression. Results A total of 654 lesions (420 in the PZ) were detected. The probability of cancer detection for PI-RADS category 2, 3, 4, and 5 lesions was 15.7%, 33.1%, 70.5%, and 90.7%, respectively. DW imaging outperformed T2-weighted imaging in the PZ (OR, 3.49 vs 2.45; P = .008). T2-weighted imaging performed better but did not clearly outperform DW imaging in the TZ (OR, 4.79 vs 3.77; P = .494). Lesions classified as PI-RADS category 3 at DW MR imaging and as positive at DCE imaging in the PZ showed a higher probability of cancer detection than did DCE-negative PI-RADS category 3 lesions (67.8% vs 40.0%, P = .02). The addition of DCE imaging to DW imaging in the PZ was beneficial (OR, 2.0; P = .027), with an increase in the probability of cancer detection of 15.7%, 16.0%, and 9.2% for PI-RADS category 2, 3, and 4 lesions, respectively. Conclusion DW imaging outperforms T2-weighted imaging in the PZ; T2-weighted imaging did not show a significant difference when compared with DW imaging in the TZ by PI-RADS version 2 criteria. The addition of DCE imaging to DW imaging scores in the PZ yields meaningful improvements in probability of cancer detection. © RSNA, 2017 An earlier incorrect version of this article appeared online. This article was corrected on July 27, 2017. Online supplemental material is available for this article.

Deep Learning-Based Interpretable AI for Prostate T2W MRI Quality Evaluation.

RATIONALE AND OBJECTIVES: Prostate MRI quality is essential in guiding prostate biopsies. However, assessment of MRI quality is subjective with variation. Quality degradation sources exert varying impacts based on the sequence under consideration, such as T2W versus DWI. As a result, employing sequence-specific techniques for quality assessment could yield more advantageous outcomes. This study aims to develop an AI tool that offers a more consistent evaluation of T2W prostate MRI quality, efficiently identifying suboptimal scans while minimizing user bias. MATERIALS AND METHODS: This retrospective study included 1046 patients from three cohorts (ProstateX [n = 347], All-comer in-house [n = 602], enriched bad-quality MRI in-house [n = 97]) scanned between January 2011 and May 2022. An expert reader assigned T2W MRIs a quality score. A train-validation-test split of 70:15:15 was applied, ensuring equal distribution of MRI scanners and protocols across all partitions. T2W quality AI classification model was based on 3D DenseNet121 architecture using MONAI framework. In addition to multiclassification, binary classification was utilized (Classes 0/1 vs. 2). A score of 0 was given to scans considered non-diagnostic or unusable, a score of 1 was given to those with acceptable diagnostic quality with some usability but with some quality distortions present, and a score of 2 was given to those considered optimal diagnostic quality and usability. Partial occlusion sensitivity maps were generated for anatomical correlation. Three body radiologists assessed reproducibility within a subgroup of 60 test cases using weighted Cohen Kappa. RESULTS: The best validation multiclass accuracy of 77.1% (121/157) was achieved during training. In the test dataset, multiclassification accuracy was 73.9% (116/157), whereas binary accuracy was 84.7% (133/157). Sub-class sensitivity for binary quality distortion classification for class 0 was 100% (18/18), and sub-class specificity for T2W classification of absence/minimal quality distortions for class 2 was 90.5% (95/105). All three readers showed moderate to substantial agreement with ground truth (R1-R3 κ = 0.588, κ = 0.649, κ = 0.487, respectively), moderate to substantial agreement with each other (R1-R2 κ = 0.599, R1-R3 κ = 0.612, R2-R3 κ = 0.685), fair to moderate agreement with AI (R1-R3 κ = 0.445, κ = 0.410, κ = 0.292, respectively). AI showed substantial agreement with ground truth (κ = 0.704). 3D quality heatmap evaluation revealed that the most critical non-diagnostic quality imaging features from an AI perspective related to obscuration of the rectoprostatic space (94.4%, 17/18). CONCLUSION: The 3D AI model can assess T2W prostate MRI quality with moderate accuracy and translate whole sequence-level classification labels into 3D voxel-level quality heatmaps for interpretation. Image quality has a significant downstream impact on ruling out clinically significant cancers. AI may be able to help with reproducible identification of MRI sequences requiring re-acquisition with explainability.

Kidney scoring surveillance: predictive machine learning models for clear cell renal cell carcinoma growth using MRI.

INTRODUCTION: Classification of clear cell renal cell carcinoma (ccRCC) growth rates in patients with Von Hippel-Lindau (VHL) syndrome has several ramifications for tumor monitoring and surgical planning. Using two separate machine-learning algorithms, we sought to produce models to predict ccRCC growth rate classes based on qualitative MRI-derived characteristics. MATERIAL AND METHODS: We used a prospectively maintained database of patients with VHL who underwent surgical resection for ccRCC between January 2015 and June 2022. We employed a threshold growth rate of 0.5 cm per year to categorize ccRCC tumors into two distinct groups-'slow-growing' and 'fast-growing'. Utilizing a questionnaire of qualitative imaging features, two radiologists assessed each lesion on different MRI sequences. Two machine-learning models, a stacked ensemble technique and a decision tree algorithm, were used to predict the tumor growth rate classes. Positive predictive value (PPV), sensitivity, and F1-score were used to evaluate the performance of the models. RESULTS: This study comprises 55 patients with VHL with 128 ccRCC tumors. Patients' median age was 48 years, and 28 patients were males. Each patient had an average of two tumors, with a median size of 2.1 cm and a median growth rate of 0.35 cm/year. The overall performance of the stacked and DT model had 0.77 ± 0.05 and 0.71 ± 0.06 accuracies, respectively. The best stacked model achieved a PPV of 0.92, a sensitivity of 0.91, and an F1-score of 0.90. CONCLUSION: This study provides valuable insight into the potential of machine-learning analysis for the determination of renal tumor growth rate in patients with VHL. This finding could be utilized as an assistive tool for the individualized screening and follow-up of this population.

Deep learning-based whole-body PSMA PET/CT attenuation correction utilizing Pix-2-Pix GAN.

PURPOSE: Sequential PET/CT studies oncology patients can undergo during their treatment follow-up course is limited by radiation dosage. We propose an artificial intelligence (AI) tool to produce attenuation-corrected PET (AC-PET) images from non-attenuation-corrected PET (NAC-PET) images to reduce need for low-dose CT scans. METHODS: A deep learning algorithm based on 2D Pix-2-Pix generative adversarial network (GAN) architecture was developed from paired AC-PET and NAC-PET images. 18F-DCFPyL PSMA PET-CT studies from 302 prostate cancer patients, split into training, validation, and testing cohorts (n = 183, 60, 59, respectively). Models were trained with two normalization strategies: Standard Uptake Value (SUV)-based and SUV-Nyul-based. Scan-level performance was evaluated by normalized mean square error (NMSE), mean absolute error (MAE), structural similarity index (SSIM), and peak signal-to-noise ratio (PSNR). Lesion-level analysis was performed in regions-of-interest prospectively from nuclear medicine physicians. SUV metrics were evaluated using intraclass correlation coefficient (ICC), repeatability coefficient (RC), and linear mixed-effects modeling. RESULTS: Median NMSE, MAE, SSIM, and PSNR were 13.26%, 3.59%, 0.891, and 26.82, respectively, in the independent test cohort. ICC for SUVmax and SUVmean were 0.88 and 0.89, which indicated a high correlation between original and AI-generated quantitative imaging markers. Lesion location, density (Hounsfield units), and lesion uptake were all shown to impact relative error in generated SUV metrics (all p < 0.05). CONCLUSION: The Pix-2-Pix GAN model for generating AC-PET demonstrates SUV metrics that highly correlate with original images. AI-generated PET images show clinical potential for reducing the need for CT scans for attenuation correction while preserving quantitative markers and image quality.

Using YOLO v7 to Detect Kidney in Magnetic Resonance Imaging.

Introduction: This study explores the use of the latest You Only Look Once (YOLO V7) object detection method to enhance kidney detection in medical imaging by training and testing a modified YOLO V7 on medical image formats. Methods: Study includes 878 patients with various subtypes of renal cell carcinoma (RCC) and 206 patients with normal kidneys. A total of 5657 MRI scans for 1084 patients were retrieved. 326 patients with 1034 tumors recruited from a retrospective maintained database, and bounding boxes were drawn around their tumors. A primary model was trained on 80% of annotated cases, with 20% saved for testing (primary test set). The best primary model was then used to identify tumors in the remaining 861 patients and bounding box coordinates were generated on their scans using the model. Ten benchmark training sets were created with generated coordinates on not-segmented patients. The final model used to predict the kidney in the primary test set. We reported the positive predictive value (PPV), sensitivity, and mean average precision (mAP). Results: The primary training set showed an average PPV of 0.94 ± 0.01, sensitivity of 0.87 ± 0.04, and mAP of 0.91 ± 0.02. The best primary model yielded a PPV of 0.97, sensitivity of 0.92, and mAP of 0.95. The final model demonstrated an average PPV of 0.95 ± 0.03, sensitivity of 0.98 ± 0.004, and mAP of 0.95 ± 0.01. Conclusion: Using a semi-supervised approach with a medical image library, we developed a high-performing model for kidney detection. Further external validation is required to assess the model's generalizability.

Deep learning algorithm (YOLOv7) for automated renal mass detection on contrast-enhanced MRI: a 2D and 2.5D evaluation of results.

INTRODUCTION: Accurate diagnosis and treatment of kidney tumors greatly benefit from automated solutions for detection and classification on MRI. In this study, we explore the application of a deep learning algorithm, YOLOv7, for detecting kidney tumors on contrast-enhanced MRI. MATERIAL AND METHODS: We assessed the performance of YOLOv7 tumor detection on excretory phase MRIs in a large institutional cohort of patients with RCC. Tumors were segmented on MRI using ITK-SNAP and converted to bounding boxes. The cohort was randomly divided into ten benchmarks for training and testing the YOLOv7 algorithm. The model was evaluated using both 2-dimensional and a novel in-house developed 2.5-dimensional approach. Performance measures included F1, Positive Predictive Value (PPV), Sensitivity, F1 curve, PPV-Sensitivity curve, Intersection over Union (IoU), and mean average PPV (mAP). RESULTS: A total of 326 patients with 1034 tumors with 7 different pathologies were analyzed across ten benchmarks. The average 2D evaluation results were as follows: Positive Predictive Value (PPV) of 0.69 ± 0.05, sensitivity of 0.39 ± 0.02, and F1 score of 0.43 ± 0.03. For the 2.5D evaluation, the average results included a PPV of 0.72 ± 0.06, sensitivity of 0.61 ± 0.06, and F1 score of 0.66 ± 0.04. The best model performance demonstrated a 2.5D PPV of 0.75, sensitivity of 0.69, and F1 score of 0.72. CONCLUSION: Using computer vision for tumor identification is a cutting-edge and rapidly expanding subject. In this work, we showed that YOLOv7 can be utilized in the detection of kidney cancers.

Deep learning-based image quality assessment: impact on detection accuracy of prostate cancer extraprostatic extension on MRI.

OBJECTIVE: To assess impact of image quality on prostate cancer extraprostatic extension (EPE) detection on MRI using a deep learning-based AI algorithm. MATERIALS AND METHODS: This retrospective, single institution study included patients who were imaged with mpMRI and subsequently underwent radical prostatectomy from June 2007 to August 2022. One genitourinary radiologist prospectively evaluated each patient using the NCI EPE grading system. Each T2WI was classified as low- or high-quality by a previously developed AI algorithm. Fisher's exact tests were performed to compare EPE detection metrics between low- and high-quality images. Univariable and multivariable analyses were conducted to assess the predictive value of image quality for pathological EPE. RESULTS: A total of 773 consecutive patients (median age 61 [IQR 56-67] years) were evaluated. At radical prostatectomy, 23% (180/773) of patients had EPE at pathology, and 41% (131/318) of positive EPE calls on mpMRI were confirmed to have EPE. The AI algorithm classified 36% (280/773) of T2WIs as low-quality and 64% (493/773) as high-quality. For EPE grade ≥ 1, high-quality T2WI significantly improved specificity for EPE detection (72% [95% CI 67-76%] vs. 63% [95% CI 56-69%], P = 0.03), but did not significantly affect sensitivity (72% [95% CI 62-80%] vs. 75% [95% CI 63-85%]), positive predictive value (44% [95% CI 39-49%] vs. 38% [95% CI 32-43%]), or negative predictive value (89% [95% CI 86-92%] vs. 89% [95% CI 85-93%]). Sensitivity, specificity, PPV, and NPV for EPE grades ≥ 2 and ≥ 3 did not show significant differences attributable to imaging quality. For NCI EPE grade 1, high-quality images (OR 3.05, 95% CI 1.54-5.86; P < 0.001) demonstrated a stronger association with pathologic EPE than low-quality images (OR 1.76, 95% CI 0.63-4.24; P = 0.24). CONCLUSION: Our study successfully employed a deep learning-based AI algorithm to classify image quality of prostate MRI and demonstrated that better quality T2WI was associated with more accurate prediction of EPE at final pathology.

Automated prostate gland segmentation in challenging clinical cases: comparison of three artificial intelligence methods.

OBJECTIVE: Automated methods for prostate segmentation on MRI are typically developed under ideal scanning and anatomical conditions. This study evaluates three different prostate segmentation AI algorithms in a challenging population of patients with prior treatments, variable anatomic characteristics, complex clinical history, or atypical MRI acquisition parameters. MATERIALS AND METHODS: A single institution retrospective database was queried for the following conditions at prostate MRI: prior prostate-specific oncologic treatment, transurethral resection of the prostate (TURP), abdominal perineal resection (APR), hip prosthesis (HP), diversity of prostate volumes (large ≥ 150 cc, small ≤ 25 cc), whole gland tumor burden, magnet strength, noted poor quality, and various scanners (outside/vendors). Final inclusion criteria required availability of axial T2-weighted (T2W) sequence and corresponding prostate organ segmentation from an expert radiologist. Three previously developed algorithms were evaluated: (1) deep learning (DL)-based model, (2) commercially available shape-based model, and (3) federated DL-based model. Dice Similarity Coefficient (DSC) was calculated compared to expert. DSC by model and scan factors were evaluated with Wilcox signed-rank test and linear mixed effects (LMER) model. RESULTS: 683 scans (651 patients) met inclusion criteria (mean prostate volume 60.1 cc [9.05-329 cc]). Overall DSC scores for models 1, 2, and 3 were 0.916 (0.707-0.971), 0.873 (0-0.997), and 0.894 (0.025-0.961), respectively, with DL-based models demonstrating significantly higher performance (p < 0.01). In sub-group analysis by factors, Model 1 outperformed Model 2 (all p < 0.05) and Model 3 (all p < 0.001). Performance of all models was negatively impacted by prostate volume and poor signal quality (p < 0.01). Shape-based factors influenced DL models (p < 0.001) while signal factors influenced all (p < 0.001). CONCLUSION: Factors affecting anatomical and signal conditions of the prostate gland can adversely impact both DL and non-deep learning-based segmentation models.

Deep Learning-Based Detection and Classification of Bone Lesions on Staging Computed Tomography in Prostate Cancer: A Development Study.

RATIONALE AND OBJECTIVES: Efficiently detecting and characterizing metastatic bone lesions on staging CT is crucial for prostate cancer (PCa) care. However, it demands significant expert time and additional imaging such as PET/CT. We aimed to develop an ensemble of two automated deep learning AI models for 1) bone lesion detection and segmentation and 2) benign vs. metastatic lesion classification on staging CTs and to compare its performance with radiologists. MATERIALS AND METHODS: This retrospective study developed two AI models using 297 staging CT scans (81 metastatic) with 4601 benign and 1911 metastatic lesions in PCa patients. Metastases were validated by follow-up scans, bone biopsy, or PET/CT. Segmentation AI (3DAISeg) was developed using the lesion contours delineated by a radiologist. 3DAISeg performance was evaluated with the Dice similarity coefficient, and classification AI (3DAIClass) performance on AI and radiologist contours was assessed with F1-score and accuracy. Training/validation/testing data partitions of 70:15:15 were used. A multi-reader study was performed with two junior and two senior radiologists within a subset of the testing dataset (n = 36). RESULTS: In 45 unseen staging CT scans (12 metastatic PCa) with 669 benign and 364 metastatic lesions, 3DAISeg detected 73.1% of metastatic (266/364) and 72.4% of benign lesions (484/669). Each scan averaged 12 extra segmentations (range: 1-31). All metastatic scans had at least one detected metastatic lesion, achieving a 100% patient-level detection. The mean Dice score for 3DAISeg was 0.53 (median: 0.59, range: 0-0.87). The F1 for 3DAIClass was 94.8% (radiologist contours) and 92.4% (3DAISeg contours), with a median false positive of 0 (range: 0-3). Using radiologist contours, 3DAIClass had PPV and NPV rates comparable to junior and senior radiologists: PPV (semi-automated approach AI 40.0% vs. Juniors 32.0% vs. Seniors 50.0%) and NPV (AI 96.2% vs. Juniors 95.7% vs. Seniors 91.9%). When using 3DAISeg, 3DAIClass mimicked junior radiologists in PPV (pure-AI 20.0% vs. Juniors 32.0% vs. Seniors 50.0%) but surpassed seniors in NPV (pure-AI 93.8% vs. Juniors 95.7% vs. Seniors 91.9%). CONCLUSION: Our lesion detection and classification AI model performs on par with junior and senior radiologists in discerning benign and metastatic lesions on staging CTs obtained for PCa.

The Low Rate of Adherence to Checklist for Artificial Intelligence in Medical Imaging Criteria Among Published Prostate MRI Artificial Intelligence Algorithms.

OBJECTIVE: To determine the rigor, generalizability, and reproducibility of published classification and detection artificial intelligence (AI) models for prostate cancer (PCa) on MRI using the Checklist for Artificial Intelligence in Medical Imaging (CLAIM) guidelines, a 42-item checklist that is considered a measure of best practice for presenting and reviewing medical imaging AI research. MATERIALS AND METHODS: This review searched English literature for studies proposing PCa AI detection and classification models on MRI. Each study was evaluated with the CLAIM checklist. The additional outcomes for which data were sought included measures of AI model performance (eg, area under the curve [AUC], sensitivity, specificity, free-response operating characteristic curves), training and validation and testing group sample size, AI approach, detection versus classification AI, public data set utilization, MRI sequences used, and definition of gold standard for ground truth. The percentage of CLAIM checklist fulfillment was used to stratify studies into quartiles. Wilcoxon's rank-sum test was used for pair-wise comparisons. RESULTS: In all, 75 studies were identified, and 53 studies qualified for analysis. The original CLAIM items that most studies did not fulfill includes item 12 (77% no): de-identification methods; item 13 (68% no): handling missing data; item 15 (47% no): rationale for choosing ground truth reference standard; item 18 (55% no): measurements of inter- and intrareader variability; item 31 (60% no): inclusion of validated interpretability maps; item 37 (92% no): inclusion of failure analysis to elucidate AI model weaknesses. An AUC score versus percentage CLAIM fulfillment quartile revealed a significant difference of the mean AUC scores between quartile 1 versus quartile 2 (0.78 versus 0.86, P = .034) and quartile 1 versus quartile 4 (0.78 versus 0.89, P = .003) scores. Based on additional information and outcome metrics gathered in this study, additional measures of best practice are defined. These new items include disclosure of public dataset usage, ground truth definition in comparison to other referenced works in the defined task, and sample size power calculation. CONCLUSION: A large proportion of AI studies do not fulfill key items in CLAIM guidelines within their methods and results sections. The percentage of CLAIM checklist fulfillment is weakly associated with improved AI model performance. Additions or supplementations to CLAIM are recommended to improve publishing standards and aid reviewers in determining study rigor.

Deep learning-based decision forest for hereditary clear cell renal cell carcinoma segmentation on MRI.

BACKGROUND: von Hippel-Lindau syndrome (VHL) is an autosomal dominant hereditary syndrome with an increased predisposition of developing numerous cysts and tumors, almost exclusively clear cell renal cell carcinoma (ccRCC). Considering the lifelong surveillance in such patients to monitor the disease, patients with VHL are preferentially imaged using MRI to eliminate radiation exposure. PURPOSE: Segmentation of kidney and tumor structures on MRI in VHL patients is useful in lesion characterization (e.g., cyst vs. tumor), volumetric lesion analysis, and tumor growth prediction. However, automated tasks such as ccRCC segmentation on MRI is sparsely studied. We develop segmentation methodology for ccRCC on T1 weighted precontrast, corticomedullary, nephrogenic, and excretory contrast phase MRI. METHODS: We applied a new neural network approache using a novel differentiable decision forest, called hinge forest (HF), to segment kidney parenchyma, cyst, and ccRCC tumors in 117 images from 115 patients. This data set represented an unprecedented 504 ccRCCs with 1171 cystic lesions obtained at five different MRI scanners. The HF architecture was compared with U-Net on 10 randomized splits with 75% used for training and 25% used for testing. Both methods were trained with Adam using default parameters ( α = 0.001 , ß 1 = 0.9 , ß 2 = 0.999 $\alpha = 0.001,\ \beta _1 = 0.9,\ \beta _2 = 0.999$ ) over 1000 epochs. We further demonstrated some interpretability of our HF method by exploiting decision tree structure. RESULTS: The HF achieved an average kidney, cyst, and tumor Dice similarity coefficient (DSC) of 0.75 ± 0.03, 0.44 ± 0.05, 0.53 ± 0.04, respectively, while U-Net achieved an average kidney, cyst, and tumor DSC of 0.78 ± 0.02, 0.41 ± 0.04, 0.46 ± 0.05, respectively. The HF significantly outperformed U-Net on tumors while U-Net significantly outperformed HF when segmenting kidney parenchymas ( α < 0.01 $\alpha < 0.01$ ). CONCLUSIONS: For the task of ccRCC segmentation, the HF can offer better segmentation performance compared to the traditional U-Net architecture. The leaf maps can glean hints about deep learning features that might prove to be useful in other automated tasks such as tumor characterization.

Automatic segmentation of clear cell renal cell tumors, kidney, and cysts in patients with von Hippel-Lindau syndrome using U-net architecture on magnetic resonance images.

We demonstrate automated segmentation of clear cell renal cell carcinomas (ccRCC), cysts, and surrounding normal kidney parenchyma in patients with von Hippel-Lindau (VHL) syndrome using convolutional neural networks (CNN) on Magnetic Resonance Imaging (MRI). We queried 115 VHL patients and 117 scans (3 patients have two separate scans) with 504 ccRCCs and 1171 cysts from 2015 to 2021. Lesions were manually segmented on T1 excretory phase, co-registered on all contrast-enhanced T1 sequences and used to train 2D and 3D U-Net. The U-Net performance was evaluated on 10 randomized splits of the cohort. The models were evaluated using the dice similarity coefficient (DSC). Our 2D U-Net achieved an average ccRCC lesion detection Area under the curve (AUC) of 0.88 and DSC scores of 0.78, 0.40, and 0.46 for segmentation of the kidney, cysts, and tumors, respectively. Our 3D U-Net achieved an average ccRCC lesion detection AUC of 0.79 and DSC scores of 0.67, 0.32, and 0.34 for kidney, cysts, and tumors, respectively. We demonstrated good detection and moderate segmentation results using U-Net for ccRCC on MRI. Automatic detection and segmentation of normal renal parenchyma, cysts, and masses may assist radiologists in quantifying the burden of disease in patients with VHL.