Analyzing Pulmonary Abnormality with Superpixel Based Graph Neural Network in Chest X-Ray

In recent years, the utilization of graph-based deep learning has gained prominence, yet its potential in the realm of medical diagnosis remains relatively unexplored. The challenge arises from the inherent irregular and unordered nature of physiological data, making it challenging to depict intricate patterns using conventional methods. Our research focuses on abnormality screening: classification of Chest X-Ray (CXR) as Tuberculosis positive or negative, using Graph Neural Networks (GNN) that uses Region Adjacency Graphs (RAGs) and each superpixel serves as a dedicated graph node. By integrating residual and concatenation structures, our approach effectively captures crucial features and relationships among superpixels, facilitating advancements in tuberculosis identification. Through the amalgamation of state-of-the-art neural network architectures and innovative graph-based representations, our work introduces a new perspective to medical image analysis.