Toward Robust Canine Cardiac Diagnosis: Deep Prototype Alignment Network-Based Few-Shot Segmentation in Veterinary Medicine

TL;DR

This work tackles canine cardiomegaly segmentation from chest X-rays under limited data by applying few-shot segmentation (FSS) in veterinary medicine. It introduces DPANet, a Deep Prototype Alignment Network that replaces PANet's encoder with a deeper VGG-19 and adds Prototype Alignment Regularization (PAR) to align support-query embeddings via class-specific prototypes, optimizing a combined loss $L = L_{seg} + L_{PAR}$. On a pet chest X-ray dataset, DPANet achieves state-of-the-art $IoU$ in both 2way-1shot ($IoU = 0.6966$) and 2way-5shot ($IoU = 0.7797$) settings and demonstrates faster training in the 2way-5shot scenario compared to baselines. This indicates DPANet's potential as a data-efficient, foundational model for veterinary AI, enabling accurate cardiomegaly segmentation with limited annotated data and setting the stage for broader canine disease segmentation research.

Abstract

In the cutting-edge domain of medical artificial intelligence (AI), remarkable advances have been achieved in areas such as diagnosis, prediction, and therapeutic interventions. Despite these advances, the technology for image segmentation faces the significant barrier of having to produce extensively annotated datasets. To address this challenge, few-shot segmentation (FSS) has been recognized as one of the innovative solutions. Although most of the FSS research has focused on human health care, its application in veterinary medicine, particularly for pet care, remains largely limited. This study has focused on accurate segmentation of the heart and left atrial enlargement on canine chest radiographs using the proposed deep prototype alignment network (DPANet). The PANet architecture is adopted as the backbone model, and experiments are conducted using various encoders based on VGG-19, ResNet-18, and ResNet-50 to extract features. Experimental results demonstrate that the proposed DPANet achieves the highest performance. In the 2way-1shot scenario, it achieves the highest intersection over union (IoU) value of 0.6966, and in the 2way-5shot scenario, it achieves the highest IoU value of 0.797. The DPANet not only signifies a performance improvement, but also shows an improved training speed in the 2way-5shot scenario. These results highlight our model's exceptional capability as a trailblazing solution for segmenting the heart and left atrial enlargement in veterinary applications through FSS, setting a new benchmark in veterinary AI research, and demonstrating its superior potential to veterinary medicine advances.

Figures (3)

• Figure 1: Overview of the proposed DPANet for segmenting canine heart and left atrial enlargement part. The basic flow is similar to PANet, but we replace the encoder part with VGG-19.
• Figure 2: Loss and align loss graphs for each scenario and method. Align loss means the loss value of PAR, and loss means the training loss of DPANet combining align loss and segmentation loss.
• Figure 3: Visualization of segmentation results of each method in 2way-1shot and 2way-5shot. Red is the part that segments the canine heart, and green is the result of segmenting the left atrial enlargement, which is one of the symptoms of cardiomegaly.