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

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.

Granulomatous Prostatitis, the Great Mimicker of Prostate Cancer: Can Multiparametric MRI Features Help in This Challenging Differential Diagnosis?

“…This choice is due, nowadays, to the lack of certain and definite imaging findings and to the need to rule out concomitant prostate cancer. Studies on larger series or, maybe, the use of artificial intelligence methods [ 8 ] such as radiomic features [ 22 ] and, in particular, deep learning methods, could help the radiologist to unravel this challenging differential diagnosis.…”

“…The MRI PI-RADS v2.1 acquisition protocol included high-resolution T2w sequences in the axial (TR/TE = 4790/123 ms, voxel size = 0.3 × 0.3 × 3.0 mm 3 ), sagittal (TR/TE = 4470/101 ms, voxel size = 0.3 × 0.3 × 3.0 mm 3 ) and coronal (TR/TE = 3520/123 ms, voxel size = 0.3 × 0.3 × 3.0 mm 3 ) planes, automatically interpolated from a voxel size of 0.74 × 0.63 × 3.00 mm 3 by the MRI console; a T1-weighted sequence (TR/TE = 450/10 ms, voxel size = 0.6 × 0.6 × 3.0 mm 3 ) in the axial plane; a multi-b DWI (b-values = [50, 100, 800, 1000] s/mm 2 , voxel size = 1.0 × 1.0 × 3.0 mm 3 , three directions) EPI sequence, automatically interpolated from a voxel size of 2.60 × 2.08 × 3.00 mm 3 by the MRI console, whose corresponding ADC maps were automatically calculated using software on board of the MRI console; a high-b DWI (b-values: [1400, 1800] s/mm 2 , voxel size = 2.2 × 2.2 × 3.0 mm 3 , three directions) EPI sequence; a Dynamic Contrast Enhancement (DCE) assessment with time intensity curves evaluation, as previously described [ 8 ].…”

Radiomics in PI-RADS 3 Multiparametric MRI for Prostate Cancer Identification: Literature Models Re-Implementation and Proposal of a Clinical–Radiological Model

“…Additionally, our model is based only on the following three features: the PSA density (routinely obtained in the standard workup of these patients) and the two radiomic features obtained in two standard mpMRI studies. We are aware that, in machine learning, wrapped and embedded feature selection methods that optimally combine a broader number of features within model optimization or even deep learning models, as seen in Bertelli et al [ 30 ], are often used. However, we preferred to follow a different approach to try to obtain an explainable radiomic model, i.e., one able to explain lesion characteristics related to malignancy.…”

Diffusion-Weighted MRI in the Genitourinary System

“…The expert panel and studies suggest a minimum of 100 training cases before obtaining an AUC on par with more experimented readers [ 70 , 71 ]. However, training requirements may be drastically modified by the introduction of new machine learning algorithms to assist prostate MRI analysis [ 72 , 73 ].…”