Machine and Deep Learning Prediction Of Prostate Cancer Aggressiveness Using Multiparametric MRI.

Elena Bertelli,Vittorio Miele,Luca Gherardini,Andrea Barucci,Michela Baccini,Maria Antonietta Pascali,Sara Colantonio,Simone Agostini,Lorenzo Lattavo,Chiara Marzi,Laura Mercatelli,Eva Pachetti

doi:10.3389/fonc.2021.802964

Abstract

Prostate cancer (PCa) is the most frequent male malignancy and the assessment of PCa aggressiveness, for which a biopsy is required, is fundamental for patient management. Currently, multiparametric (mp) MRI is strongly recommended before biopsy. Quantitative assessment of mpMRI might provide the radiologist with an objective and noninvasive tool for supporting the decision-making in clinical practice and decreasing intra- and inter-reader variability. In this view, high dimensional radiomics features and Machine Learning (ML) techniques, along with Deep Learning (DL) methods working on raw images directly, could assist the radiologist in the clinical workflow. The aim of this study was to develop and validate ML/DL frameworks on mpMRI data to characterize PCas according to their aggressiveness. We optimized several ML/DL frameworks on T2w, ADC and T2w+ADC data, using a patient-based nested validation scheme. The dataset was composed of 112 patients (132 peripheral lesions with Prostate Imaging Reporting and Data System (PI-RADS) score ≥ 3) acquired following both PI-RADS 2.0 and 2.1 guidelines. Firstly, ML/DL frameworks trained and validated on PI-RADS 2.0 data were tested on both PI-RADS 2.0 and 2.1 data. Then, we trained, validated and tested ML/DL frameworks on a multi PI-RADS dataset. We reported the performances in terms of Area Under the Receiver Operating curve (AUROC), specificity and sensitivity. The ML/DL frameworks trained on T2w data achieved the overall best performance. Notably, ML and DL frameworks trained and validated on PI-RADS 2.0 data obtained median AUROC values equal to 0.750 and 0.875, respectively, on unseen PI-RADS 2.0 test set. Similarly, ML/DL frameworks trained and validated on multi PI-RADS T2w data showed median AUROC values equal to 0.795 and 0.750, respectively, on unseen multi PI-RADS test set. Conversely, all the ML/DL frameworks trained and validated on PI-RADS 2.0 data, achieved AUROC values no better than the chance level when tested on PI-RADS 2.1 data. Both ML/DL techniques applied on mpMRI seem to be a valid aid in predicting PCa aggressiveness. In particular, ML/DL frameworks fed with T2w images data (objective, fast and non-invasive) show good performances and might support decision-making in patient diagnostic and therapeutic management, reducing intra- and inter-reader variability.

Highlights

Prostate cancer (PCa) is the most frequent male malignancy and the third cause of cancer death in European men with significant consequences for healthcare systems [1]
In this paper, we have investigated the potential role of several ML and DL frameworks in predicting PCa aggressiveness from mpMRI data, using a computational workflow that prevents the previously mentioned issues
The PI-RADS 2.1 cohort has been used in two different ways: firstly, we have considered the entire cohort as an independent test set, and, secondly, we have split it in development set 2.1, containing images of 19 PCa patients, and test set 2.1, with images of eight patients

Summary

Introduction

Prostate cancer (PCa) is the most frequent male malignancy and the third cause of cancer death in European men with significant consequences for healthcare systems [1]. Over the last decades, mpMRI has become increasingly valuable for the detection and staging of PCa, gaining a key role in the diagnostic pathway [4] and apparent advantages compared to the systematic transrectal ultrasonography-guidedbiopsy (TRUSGB) [5] It can rule out non clinically significant (cs) PCa, reducing the number of unnecessary prostate biopsies and overdiagnosis. Quantitative assessment of lesion aggressiveness on mpMRI might reinforce MRI importance, role, and value in PCa diagnostic, prognostic and monitoring pathway, providing the radiologist with an objective and noninvasive tool and decreasing intra- and inter-reader variability [19] This would permit the urologist to choose and/or modify the management approach, optimizing quality of life of many patients. In a protocol-mandated future perspective, together with PSA and clinical data, quantitative mpMRI and relative analyses could actively bring out lesion progression, maybe reducing the need of re-biopsies

Objectives

Methods

Results

Conclusion