Constructing machine learning models with interpretable cellular and tissue level features for prognosing early-stage cutaneous melanoma from primary tumor whole slide images

Sep 17, 2026

Uploaded by Theodora

This research presents a novel machine-learning framework designed to predict progression-free survival (PFS) in patients with early-stage cutaneous melanoma. By utilizing high-resolution primary tumor whole slide images (WSIs), the authors extract interpretable cellular and tissue-level features, including cell morphology, texture, and collagen organization. The study demonstrates that these features effectively stratify patient risk, yielding performance metrics competitive with state-of-the-art models like CLAM. Furthermore, the analysis identifies a significant dose-dependent relationship between ulceration scores and survival outcomes. Validated against a cohort from the Melanoma Institute of Australia, the model offers a promising tool for clinical decision-making, enabling more precise risk stratification, tailored follow-up care, and optimized enrollment in oncology research studies. The research highlights the critical importance of sub-staging patients categorized as IUSM IIA and IIB, particularly for those presenting with bulky tumors in superficial spreading or nodular melanoma subtypes.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages1

Reading Time3 min

AccessFree to download

Abstract

Introduction

Materials & Methods

Study Cohorts

Pipeline

Results

Conclusions

FAQs

What is the primary objective of this research?

The research aims to develop and validate an interpretable machine-learning tool that utilizes histopathological features from primary melanoma tumors to predict patient progression-free survival.

How does the proposed model perform compared to existing methods?

The research findings indicate that the proposed method performs similarly to the state-of-the-art deep-learning model known as CLAM.

What is the clinical significance of the ulceration score developed in this study?

The study demonstrates that the novel ulceration score has a significant dose-dependent relationship with progression-free survival, aiding in the stratification of patient risk.

Sep 17, 2026

Uploaded by Theodora