Abstract:Objective: To develop and evaluate a deep radiomics model for clinically significant prostate cancer (csPCa, grade group >= 2) detection and compare its performance to Prostate Imaging Reporting and Data System (PI-RADS) assessment in a multicenter cohort. Materials and Methods: This retrospective study analyzed biparametric (T2W and DW) prostate MRI sequences of 615 patients (mean age, 63.1 +/- 7 years) from four datasets acquired between 2010 and 2020: PROSTATEx challenge, Prostate158 challenge, PCaMAP trial, and an in-house (NTNU/St. Olavs Hospital) dataset. With expert annotations as ground truth, a deep radiomics model was trained, including nnU-Net segmentation of the prostate gland, voxel-wise radiomic feature extraction, extreme gradient boost classification, and post-processing of tumor probability maps into csPCa detection maps. Training involved 5-fold cross-validation using the PROSTATEx (n=199), Prostate158 (n=138), and PCaMAP (n=78) datasets, and testing on the in-house (n=200) dataset. Patient- and lesion-level performance were compared to PI-RADS using area under ROC curve (AUROC [95% CI]), sensitivity, and specificity analysis. Results: On the test data, the radiologist achieved a patient-level AUROC of 0.94 [0.91-0.98] with 94% (75/80) sensitivity and 77% (92/120) specificity at PI-RADS >= 3. The deep radiomics model at a tumor probability cut-off >= 0.76 achieved 0.91 [0.86-0.95] AUROC with 90% (72/80) sensitivity and 73% (87/120) specificity, not significantly different (p = 0.068) from PI-RADS. On the lesion level, PI-RADS cut-off >= 3 had 84% (91/108) sensitivity at 0.2 (40/200) false positives per patient, while deep radiomics attained 68% (73/108) sensitivity at the same false positive rate. Conclusion: Deep radiomics machine learning model achieved comparable performance to PI-RADS assessment in csPCa detection at the patient-level but not at the lesion-level.

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is to detect clinically significant prostate cancer (csPCa) using deep radiomics technology in a multi - center clinical environment and compare its performance with the commonly used Prostate Imaging Reporting and Data System (PI - RADS) assessment in clinical practice. Specifically, the researchers hope to improve the detection accuracy of csPCa by developing and evaluating a machine - learning model based on deep radiomics, and verify its practicality and effectiveness in actual clinical applications. In addition, they also hope to explore whether this model can replace or assist the traditional PI - RADS assessment method, thereby providing clinicians with a more efficient and reliable diagnostic tool. ### Research Background 1. **The Importance of MRI in Prostate Cancer Management**: - MRI has become an indispensable part of the prostate cancer management pathway and can guide targeted biopsy and improve the accuracy of diagnosis. 2. **Limitations of PI - RADS Assessment**: - Although PI - RADS has standardized the evaluation and interpretation of prostate MRI, it depends on highly professional experience and subjective judgment, and misjudgment may occur in benign lesions. - With the aging of the population and the popularization of MRI screening, manual PI - RADS assessment is difficult to meet the increasing imaging needs. 3. **The Potential of AI in Prostate Cancer Detection**: - Deep learning and radiomics have shown great potential in prostate cancer detection and classification, but deep - learning methods lack interpretability, which may affect their clinical applications. ### Research Objectives - Develop a fully - automatic machine - learning model that combines deep learning and radiomics for detecting csPCa. - Evaluate the performance of this model in a multi - center cohort and compare it with PI - RADS assessment. - Provide an AI - assisted diagnostic tool with strong interpretability and high transparency to support clinical decision - making. ### Method Overview - **Data Set**: Use four independent data sets (PROSTATEx, Prostate158, PCaMAP, and an internal data set), which altogether contain biparametric prostate MRI sequences (T2 - weighted and diffusion - weighted images) of 615 patients. - **Model Development**: Use nnU - Net for prostate segmentation, extract radiomics features, and use XGBoost for classification to generate tumor probability maps (TPmap). - **Performance Evaluation**: Evaluate the model performance at the patient and lesion levels through indicators such as AUROC, sensitivity, and specificity, and compare it with PI - RADS assessment. ### Main Results - At the patient level, the AUROC of the deep radiomics model is 0.91 (95% CI: 0.86 - 0.95), the sensitivity is 90% (72/80), and the specificity is 73% (87/120), which is comparable to PI - RADS assessment (AUROC is 0.94, sensitivity is 94%, and specificity is 77%). - At the lesion level, the sensitivity of PI - RADS assessment is 84%, while the sensitivity of the deep radiomics model is 68%, indicating that there is still room for improvement in lesion detection. ### Conclusions - The performance of the deep radiomics model at the patient level is comparable to that of PI - RADS assessment and has potential clinical application value. - This model provides a higher degree of automation and interpretability, which helps to improve the efficiency and accuracy of prostate cancer detection. Through these efforts, the researchers hope to promote the wide application of AI technology in prostate cancer diagnosis and improve the quality and efficiency of clinical diagnosis and treatment.

Deep Radiomics Detection of Clinically Significant Prostate Cancer on Multicenter MRI: Initial Comparison to PI-RADS Assessment

Radiomics from multisite MRI and clinical data to predict clinically significant prostate cancer

Magnetic resonance imaging radiomics-based prediction of clinically significant prostate cancer in equivocal PI-RADS 3 lesions in the transitional zone

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

Utility of Clinical-Radiomic Model to Identify Clinically Significant Prostate Cancer in Biparametric MRI PI-RADS V2.1 Category 3 Lesions

Fully Automated Deep Learning Model to Detect Clinically Significant Prostate Cancer at MRI

Deep learning model for the detection of prostate cancer and classification of clinically significant disease using multiparametric MRI in comparison to PI-RADs score

MRI Based Radiomics Compared With the PI-RADS V2.1 in the Prediction of Clinically Significant Prostate Cancer: Biparametric vs Multiparametric MRI

Autonomous Detection and Classification of PI-RADS Lesions in an MRI Screening Population Incorporating Multicenter-Labeled Deep Learning and Biparametric Imaging: Proof of Concept

Deep learning AI and Restriction Spectrum Imaging for patient-level detection of clinically significant prostate cancer on MRI

A Deep Learning Approach for Predicting Clinically Significant Prostate Cancer: A Retrospective, Multicentre Study

Deep‐Learning‐Based Artificial Intelligence for PI‐RADS Classification to Assist Multiparametric Prostate MRI Interpretation: A Development Study

MRI T2w Radiomics-Based Machine Learning Models in Imaging Simulated Biopsy Add Diagnostic Value to PI-RADS in Predicting Prostate Cancer: A Retrospective Diagnostic Study

External Validation of a Previously Developed Deep Learning-based Prostate Lesion Detection Algorithm on Paired External and In-House Biparametric MRI Scans

Biparametric MRI-based radiomics classifiers for the detection of prostate cancer in patients with PSA serum levels of 4∼10 ng/mL

Deep Learning in Prostate Cancer Diagnosis Using Multiparametric Magnetic Resonance Imaging With Whole-Mount Histopathology Referenced Delineations

MRI-Based Surrogate Imaging Markers of Aggressiveness in Prostate Cancer: Development of a Machine Learning Model Based on Radiomic Features

Fully automated detection and localization of clinically significant prostate cancer on MR images using a cascaded convolutional neural network

Using an Artificial Intelligence Model to Detect and Localize Visible Clinically Significant Prostate Cancer in Prostate Magnetic Resonance Imaging: a Multicenter External Validation Study.

Predicting Clinically Significant Prostate Cancer with a Deep Learning Approach: a Multicentre Retrospective Study

Use of Radiomics to Improve Diagnostic Performance of PI-RADS v2.1 in Prostate Cancer