Delineation of Skin Strata in Reflectance Confocal Microscopy Images using Recurrent Convolutional Networks with Toeplitz Attention

TL;DR

This paper presents a novel recurrent convolutional neural network with Toeplitz attention for delineating skin layers in RCM images, achieving state-of-the-art accuracy and interpretability for skin cancer screening.

Contribution

Introduces a new Toeplitz attention mechanism within a recurrent CNN framework for interpretable skin layer delineation in RCM images, improving accuracy over existing methods.

Findings

Achieved 88.17% image-wise classification accuracy

Developed a Toeplitz attention mechanism for constrained attention maps

Provided an interpretable model for skin strata delineation

Abstract

Reflectance confocal microscopy (RCM) is an effective, non-invasive pre-screening tool for skin cancer diagnosis, but it requires extensive training and experience to assess accurately. There are few quantitative tools available to standardize image acquisition and analysis, and the ones that are available are not interpretable. In this study, we use a recurrent neural network with attention on convolutional network features. We apply it to delineate skin strata in vertically-oriented stacks of transverse RCM image slices in an interpretable manner. We introduce a new attention mechanism called Toeplitz attention, which constrains the attention map to have a Toeplitz structure. Testing our model on an expert labeled dataset of 504 RCM stacks, we achieve 88.17% image-wise classification accuracy, which is the current state-of-art.