Using Spatial Logic and Model Checking for Nevus Segmentation

TL;DR

This paper presents a novel approach using spatial logic and model checking techniques for automatic segmentation of nevi in 2D medical images, addressing challenges posed by their inhomogeneity.

Contribution

It introduces a new method combining texture similarity and spatial logic for improved nevi segmentation in medical imaging.

Findings

Effective segmentation on large public database

Comparison shows promising accuracy against ground truth

Method supports explainability and reproducibility

Abstract

Spatial and spatio-temporal model checking techniques have a wide range of application domains, among which large scale distributed systems and signal and image analysis. In the latter domain, automatic and semi-automatic contouring in Medical Imaging has shown to be a very promising and versatile application that can greatly facilitate the work of professionals in this domain, while supporting explainability, easy replicability and exchange of medical image analysis methods. In recent work we have applied this model-checking technique to the (3D) contouring of tumours and related oedema in magnetic resonance images of the brain. In the current work we address the contouring of (2D) images of nevi. One of the challenges of treating nevi images is their considerable inhomogeneity in shape, colour, texture and size. To deal with this challenge we use a texture similarity operator, in combination with spatial logic operators. We apply our technique on images of a large public database and compare the results with associated ground truth segmentation provided by domain experts.