Automatic segmentation of prostate zonal anatomy on MRI: a systematic review of the literature

Abstract

Objectives Accurate zonal segmentation of prostate boundaries on MRI is a critical prerequisite for automated prostate cancer detection based on PI-RADS. Many articles have been published describing deep learning methods offering great promise for fast and accurate segmentation of prostate zonal anatomy. The objective of this review was to provide a detailed analysis and comparison of applicability and efficiency of the published methods for automatic segmentation of prostate zonal anatomy by systematically reviewing the current literature. Methods A Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) was conducted until June 30, 2021, using PubMed, ScienceDirect, Web of Science and EMBase databases. Risk of bias and applicability based on Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) criteria adjusted with Checklist for Artificial Intelligence in Medical Imaging (CLAIM) were assessed. Results A total of 458 articles were identified, and 33 were included and reviewed. Only 2 articles had a low risk of bias for all four QUADAS-2 domains. In the remaining, insufficient details about database constitution and segmentation protocol provided sources of bias (inclusion criteria, MRI acquisition, ground truth). Eighteen different types of terminology for prostate zone segmentation were found, while 4 anatomic zones are described on MRI. Only 2 authors used a blinded reading, and 4 assessed inter-observer variability. Conclusions Our review identified numerous methodological flaws and underlined biases precluding us from performing quantitative analysis for this review. This implies low robustness and low applicability in clinical practice of the evaluated methods. Actually, there is not yet consensus on quality criteria for database constitution and zonal segmentation methodology.

Comparison of Early Contrast Enhancement Models in Ultrafast Dynamic Contrast-Enhanced Magnetic Resonance Imaging of Prostate Cancer

“…Moreover, the ROIs for normal-appearing prostate specifically excluded PI-RADS category 3 or higher lesions but included regions that would be considered PI-RADS category 1 or 2 and thus suspected of being subject to pathology but not cancer. Prostate zone and lesion delineation is a rapidly developing research area, with a growing number of machine learning algorithms being presented [34][35][36][37], but the adaptation and retraining required for specific scanners and protocols were considered unwarranted in this small cohort. In routine use for a large population, such automated solutions would be a valid alternative.…”

Geometric Anatomy Basis for Safe and Effective Focal Ablation of Prostate Cancer by Irreversible Electroporation (IRE)

“…Usually, in the axial plane, partial removal of prostate cancer involves a focal area that covers at least the tumour area and its safety margin of 5 mm and at most a lateral or transvesical (anterior or posterior to the urethra) half ablation. The different zones are recognisable on the anatomical T2 MRI sequence [12] (Figure 2). The anatomy and its structural aspects and the contours of its borders evolve with aging according to the increase in size related to the benign hyperplasia of the transition zone.…”

Revisiting prostate segmentation in magnetic resonance imaging (MRI): On model transferability, degradation and PI-RADS adherence

“…Over the last decade, multi-parametric MRI (mp-MRI) has evolved as a key component for Prostate Cancer (PCa) detection, staging and treatment planning [1]. Its recommended upfront role and increasing relevance for PCa is expected to substantially increase the radiologist workload [2].…”

“…Existing shortcomings of manual MRI segmentation of the prostate WG in the form of inter and intra-reader variability [6] coupled with the growing shortage of available specialists, has motivated the development of a considerable amount of automatic prostate WG segmentation tools [1,7].…”