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

Aug 5, 2026

Upload by Ezra

This retrospective diagnostic study evaluates radiomics-based machine learning using MRI T2-weighted (T2w) images to predict prostate cancer before biopsy and to complement PI-RADS. Models were developed with 820 lesions from TCIA and externally validated with 83 lesions from Hong Kong Queen Mary Hospital. Radiomic features were extracted, selected via CV-LARS, and 18 prediction models across three algorithms were compared with PI-RADS using AUC, sensitivity, specificity, PPV, and NPV. The logistic regression model incorporating three radiomic features outperformed PI-RADS in external validation.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages14

Reading Time35 min

AccessFree to download

Introduction

Study summary and motivation

Methods overview (data, preprocessing, radiomics, modeling)

Model performance comparison (internal and external validation)

Conclusions

cancers Article 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 Jia-Cheng Liu 1,†, Xiao-Hao Ruan 1,†, Tsun-Tsun Chun 2, Chi Yao 2, Da Huang 1, Hoi-Lung Wong 3, Chun-Ting Lai 3, Chiu-Fung Tsang 3, Sze-Ho Ho 3, Tsui-Lin Ng 3, Dan-Feng Xu 1,* and Rong Na 2,3, * Citation: Liu, J.-C.; Ruan, X.-H.; Chun, T.-T.; Yao, C.; Huang, D.; Wong, H.-L.; Lai, C.-T.; Tsang, C.-F.; Ho, S.-H.; Ng, T.-L.; et al. 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. Cancers 2024, 16, 2944. [https://](https://)[ ](https://)[doi.org/10.3390/cancers16172944](doi.org/10.3390/cancers16172944) Academic Editors: Michał Strzelecki, Adam Piórkowski, Rafał Obuchowicz and Karolina Nurzynska Received: 26 July 2024 Revised: 16 August 2024 Accepted: 19 August 2024 Published: 23 August 2024 Copyright: © 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license ([https://](https://)[ ](https://)[creativecommons.org/licenses/by/](creativecommons.org/licenses/by/)[ ](creativecommons.org/licenses/by/)[4.0/](4.0/)) . 1 Department of Urology, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai 200025, China 2 Department of Surgery, School of Clinical Medicine, LKS Faculty of Medicine, The University of Hong Kong, Hong Kong, China 3 Department of Surgery, Queen Mary Hospital, Hong Kong, China; [lct729@ha.org.hk](lct729@ha.org.hk) (C.-T.L.) * Correspondence: [xdf12036@rjh.com.cn](xdf12036@rjh.com.cn) (D.-F.X.); narong.hs@gmail.com or yungna@hku.hk (R.N.); Tel.: +86-021-64370045 (D.-F.X.); +852-22554310 (R.N.)† These authors contributed equally to this work. Simple Summary: Prostate mpMRI is currently the most widely used image diagnosis approach to detect prostate cancer, while the PI-RADS system was developed to standardize and improve the accuracy of suspicious lesion identification on MRI. However, there still remain several limitations including inter-individual inconsistencies and naked-eye insufficiency. This study aims to apply AI technology to image interpretation to enhance diagnostic efficiency and explore the use of T2-weighted image-based stimulated biopsy in predicting prostate cancer (PCa) . Using 820 lesions from The Cancer Imaging Archive database and 83 lesions from Hong Kong Queen Mary Hospital, we constructed 18 machine-learning models based on three algorithms and conducted both internal and external validation. We found that the logistic regression-based model provides additional diagnostic value to the PI-RADS in predicting PCa. Abstract: Background: Currently, prostate cancer (PCa) prebiopsy medical image diagnosis mainly relies on mpMRI and PI-RADS scores. However, PI-RADS has its limitations, such as inter-and intra-radiologist variability and the potential for imperceptible features. The primary objective of this study is to evaluate the effectiveness of a machine learning model based on radiomics analysis of MRIT2-weighted (T2w) images for predicting PCa in prebiopsy cases. Method: A retrospective analysis was conducted using 820 lesions (363 cases, 457 controls) from The Cancer Imaging Archive (TCIA) Database for model development and validation. An additional 83 lesions (30 cases, 53 controls) from Hong Kong Queen Mary Hospital were used for independent external validation. The MRI T2w images were preprocessed, and radiomic features were extracted. Feature selection was performed using Cross Validation Least Angle Regression (CV-LARS). Using three different machine learning algorithms, a total of 18 prediction models and 3 shape control models were developed. The performance of the models, including the area under

FAQs

What is the main purpose of the study?

To assess whether radiomics-based machine learning from MRI T2w images can predict prostate cancer in prebiopsy cases and add diagnostic value beyond PI-RADS.

How were the machine-learning models built and validated?

The models were developed using 820 lesions from TCIA and externally validated with 83 lesions from Hong Kong Queen Mary Hospital, with radiomic feature extraction, CV-LARS feature selection, and comparisons using AUC and diagnostic metrics.

Which model performed best and how did it compare with PI-RADS?

In external validation, the logistic regression model using three radiomic features achieved higher predictive value than PI-RADS, with AUC 0.870 versus 0.658 and improved PPV and NPV.

Aug 5, 2026

Upload by Ezra