Med_Ai

Medical imaging plays a vital role in detecting diseases such as pneumonia, tumours, and 
neurological disorders at an early stage. However, interpreting these images manually can be 
slow and susceptible to human error. With recent advances in Artificial Intelligence (AI) and 
Deep Learning, faster and more accurate automated diagnostic tools have become possible 
. 
In this research, we introduce MedAI, a hybrid AI-driven imaging platform that brings 
together Convolutional Neural Networks (CNNs), Vision Transformers (ViTs), Blockchain 
technology for secure metadata storage, and Cloud-based inference to enable scalable, real
time computation. 
Based on experiments using 15,000 medically annotated images, the ResNet50-based MedAI 
model achieves an accuracy of 97.80%, surpassing traditional machine learning algorithms 
such as Logistic Regression, Naive Bayes, KNN, and Random Forest. The system 
also integrates Explainable AI to generate heatmaps that make model decisions easier to 
understand and employs blockchain hashing to ensure metadata remains tamper-proof.
Overall, the platform shows strong potential as a scalable diagnostic decision-support tool for 
hospitals, clinics, and telemedicine settings.