Deep Learning for Skin Melanocytic Tumors in Whole-Slide Images: A Systematic Review

Abstract

The rise of Artificial Intelligence (AI) has shown promising performance as a support tool in clinical pathology workflows. In addition to the well-known interobserver variability between dermatopathologists, melanomas present a significant challenge in their histological interpretation. This study aims to analyze all previously published studies on whole-slide images of melanocytic tumors that rely on deep learning techniques for automatic image analysis. Embase, Pubmed, Web of Science, and Virtual Health Library were used to search for relevant studies for the systematic review, in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist. Articles from 2015 to July 2022 were included, with an emphasis placed on the used artificial intelligence methods. Twenty-eight studies that fulfilled the inclusion criteria were grouped into four groups based on their clinical objectives, including pathologists versus deep learning models (n = 10), diagnostic prediction (n = 7); prognosis (n = 5), and histological features (n = 6). These were then analyzed to draw conclusions on the general parameters and conditions of AI in pathology, as well as the necessary factors for better performance in real scenarios.

A Spitzoid Tumor dataset with clinical metadata and Whole Slide Images for Deep Learning models

“…Most studies on DL models for melanocytic tumors have utilized clinical and dermatoscopic images 16 – 19 . However, a small group of studies has focused on using WSIs and comparing the performance of pathologists vs. DL models regarding diagnostic accuracy, prognosis prediction, and histological feature detection 19 . Only three studies have focused on Spitz tumors but excluded STUMP 20 – 22 .…”

A study on the recognition of monkeypox infection based on deep convolutional neural networks

“…In this regard, we must also define the following metrics: TP, FP, TN, and FN, representing true positive, false positive, true negative, and false negative, respectively. Among them, Precision is the proportion of positive cases correctly predicted by the model to the actual positive cases, reflecting the checking accuracy of the model, and the mathematical expression is shown in equation (11):…”

“…In recent years, with the updated iteration of computer vision technology and hardware equipment, image recognition using deep learning algorithms has become the mainstream method, and it has also achieved good results in disease investigation (10)(11)(12)(13). The experimental sample is usually a two-dimensional picture of the diseased part taken by medical equipment, which avoids the close contact between medical staff and patients as much as possible and dramatically guarantees the safety of front-line workers.…”

Histological interpretation of spitzoid tumours: an extensive machine learning‐based concordance analysis for improving decision making

“…This is due mainly to the lack of generalisability of their methodologies as one of the most common problems. 13 In this study, we propose a machine learning (ML) model based on an analysis of variance (ANOVA) according to different clinicopathological variables commonly used for the diagnosis of STs to objectively characterise, in order of relevance, the most important features according to the algorithm tested. Therefore, we attempt to demonstrate that with the interpretation shown by the model, pathologists could improve the certainty of using some particular features with more diagnostic significance and reduce the relevance of other characteristics for a proper classification of STs, and evaluate the utility of these algorithms in predicting low or high grade within the AST category.…”

“…Although the results that AI shows are promising, only a small fraction of all these studies are approved for clinical purposes. This is due mainly to the lack of generalisability of their methodologies as one of the most common problems 13 …”