# Application of convolutional neural networks for automated segmentation and classification in esophageal diseases

## TL;DR

This paper introduces a deep learning framework for automatically identifying and classifying esophageal lesions in endoscopic images.

## Contribution

A novel dual-stream CNN architecture (ELNet) and an ensemble U-Net for segmentation is proposed, showing improved performance over existing methods.

## Key findings

- The dual-stream ELNet achieved 92.14% classification accuracy with high specificity and sensitivity.
- The U-Net-based segmentation module reached 95.54% overall accuracy and 82.89% lesion segmentation sensitivity.
- The integrated framework demonstrated enhanced adaptability across diverse lesion types.

## Abstract

To develop a convolutional neural network (CNN) framework for the automated segmentation and classification of esophageal lesions in endoscopic images.

(1) Lesion localization was performed using a Region-based Convolutional Neural Network (R-CNN). (2) A dual-stream Esophageal Lesion Network (ELNet) was developed to classify images into four diagnostic categories. (3) Lesion segmentation was carried out using an ensemble of three U-Net architectures.

The dual-stream ELNet achieved a classification accuracy of 92.14%, with 97.1% specificity and 88.74% sensitivity. The segmentation module based on U-Net attained an overall accuracy of 95.54% and a lesion segmentation sensitivity of 82.89%. The dual-stream ELNet consistently outperformed single-stream baseline networks, and the integrated segmentation-with-classification architecture demonstrated enhanced adaptability across diverse lesion types.

The proposed CNN framework enables accurate, robust, and simultaneous classification and segmentation of esophageal endoscopic lesions, exhibiting high performance and clinical potential.