Feature Fusion: Enhancing Middle Ear Infection Diagnosis with Vision Transformer and Symptom Encoding

Acute otitis media and chronic suppurative otitis media are two examples of middle ear infections that are exceptionally hard to diagnose because of their overlapping symptoms and the subjective nature of standard otoscopy. This research presents a novel framework driven by Vision Transformer (ViT) to produce a cohesive diagnostic tool that addresses complexity in the real world. Utilizing the Kaggle “Otoscope Data” dataset, the model uses feature fusion, taking both the otoscopic image and clinical symptoms into account, with sophisticated preprocessing to uncover subtle patterns that traditional approaches often overlook. The system overcomes the constraints of sparse annotations and image variability through rigorous augmentation and adaptive learning, providing a solid approach that gives doctors an AI-driven solution using transformer model. This method helps us increase the accuracy of the clinical diagnosis, and also helps us to overcome the major challenges for underprivileged section of the society by enabling a cost-effective access to healthcare.