Learning to Recognize Thoracic Disease in Chest X-Rays with Knowledge-Guided Deep Zoom Neural Networks

Research Question: How can we use CNN for Thorax Disease detection, and can we get results better than the regular methods?

Diagnosis of thorax disease automatically and accurately is important for clinical assist analysis. But the major challenge is noise and similarity between visual features. For this we used KGZNet, which is a knowledge-guided deep zoom neural network. KGZNet is a data driven model. So this model learns from prior knowledge and finer region-based feature represen tation. Our results show that lung and lesion regions can boost the performance of thoracic disease classification and reinforce each other.