A Deep Learning Approach for Multi-Disease Diagnosis with ResNet50 and Spatial Attention

Healthcare is an important factor in improving quality of life. Hence, early disease diagnosis is a must for proper treatment. The integration of artificial intelligence, machine learning, and deep learning has transformed the healthcare sector, enhancing the accuracy and timeliness of medical diagnoses, especially through medical imaging. This paper presents a Deep Learning-based approach for multi-disease diagnosis, where a single model classifies images of four diseases: eye, lung, skin, and brain conditions. The proposed model used ResNet-50 as a backbone with spatial attention. Spatial attention helps focus the model's attention on relevant regions in medical images, while transfer learning applies pre-trained knowledge to avoid extensive labeled data and accelerates training. The model achieved a diagnostic accuracy of 90% and thus could be applied to real-world medical applications. Additionally, a comparison has been made between the proposed model and a few state-of-the-art models, which established that the proposed method gives better results as compared to the rest. It also shows how AI and its allied techniques can carry the vast potential to help advance diagnostics in the healthcare sector.