Endoscopic Diagnosis of Eosinophilic Esophagitis Using a Multi-Task U-Net: A Pilot Study

Endoscopically identifying eosinophilic esophagitis (EoE) is difficult due to its rare incidence and subtle morphology. We aimed to develop a robust and accurate convolutional neural network (CNN) model for EoE identification and classification in endoscopic images.

We collected 548 endoscopic images from 81 patients with EoE and 297 images from 37 normal patients. These datasets were labeled according to the four eosinophilic esophagitis endoscopic reference score (EREFS) features: edema, rings, exudates, and furrows. A multi-task U-Net with an auxiliary classifier on various levels of skip connections (scaU-Net) was proposed. Then, scaU-Net was compared with VGG19, ResNet50, EfficientNet-B3, and a typical multi-task U-Net CNN. The performances of each model were evaluated quantitatively and qualitatively based on accuracy (ACC), area under the receiver operating characteristics (AUROC), and gradient-weighted class activation map (Grad-CAM), and were also compared with those of 25 human endoscopists.

Our sca4U-Net with 4th-level skip connection showed the best performances in ACC (86.9%), AUROC (0.93), and outstanding Grad-CAM results compared to other models, reflecting the importance of utilizing the deepest skip connection. Moreover, the sca4U-Net showed generally better performance when compared with endoscopists with various levels of experience.

Our method showed robust performance compared to expert endoscopists and could assist endoscopists of all experience levels in the early detection of EoE-a rare but clinically important condition.