Deep Learning Methodologies Applied to Digital Pathology in Prostate Cancer: A Systematic Review

Abstract

Deep learning (DL), often called artificial intelligence (AI), has been increasingly used in Pathology thanks to the use of scanners to digitize slides which allow us to visualize them on monitors and process them with AI algorithms. Many articles have focused on DL applied to prostate cancer (PCa). This systematic review explains the DL applications and their performances for PCa in digital pathology. Article research was performed using PubMed and Embase to collect relevant articles. A Risk of Bias (RoB) was assessed with an adaptation of the QUADAS-2 tool. Out of the 77 included studies, eight focused on pre-processing tasks such as quality assessment or staining normalization. Most articles (n = 53) focused on diagnosis tasks like cancer detection or Gleason grading. Fifteen articles focused on prediction tasks, such as recurrence prediction or genomic correlations. Best performances were reached for cancer detection with an Area Under the Curve (AUC) up to 0.99 with algorithms already available for routine diagnosis. A few biases outlined by the RoB analysis are often found in these articles, such as the lack of external validation. This review was registered on PROSPERO under CRD42023418661.

Keywords: Gleason grading; artificial intelligence; convolutional neural networks; deep learning; digital pathology; prostate cancer.

Figure 1

DL applied to WSIs. (A) Segmentation algorithm, (B) Classification algorithm. WSIs are divided into many tiles, and every tile is encoded into features. Tiling can be performed at different magnifications, but an identical number of tiles per WSIs is generally required. The encoding into features can be trained or performed with a pre-trained algorithm. Features are then used to train a classification algorithm.

Figure 2

Flowchart illustrating the selection process of articles in this review.

Figure 3

Proportion of RoB for all articles. * Sufficient amount was estimated at 200 WSIs for prediction and diagnosis tasks.

Figure 4

Number of articles on each topic, separated by year of publication.