Publication

Abstract : MediScan is a web-based application designed for people to get first-hand opinions on their medical scans and stress the importance of seeking professional advice on the basis of the severity of the conditions identified. The application uses datasets that contain X Ray images of the bones (including un fractured and simple fractures) and MRI scan of the human brain that includes healthy brains and tumors like glioma, meningioma, and pituitary tumors. The study compares different machine learning techniques such as Logistic Regression, Decision Tree, Random Forest, Gradient Boosting, K-Nearest Neighbors (KNN), Naïve Bayes, and Convolutional Neural Networks (CNN) for identification of the best model for such an estimation. Random Forest achieves 85.64% accuracy was selected as the best fit. The model was then integrated into a Streamlit-based interface to generate real-time predictions and expedite deployment. It is equipped with features that enable interactive data analysis and visualization that even a layman can easily use. MediScan demonstrates the opportunities and prospects of machine learning in medical imaging, it is an effective tool for the initial medical examination that can potentially save the time of specialists and contribute to timely medical interventions. This study demonstrates the important role that machine learning applications play in improving the ability to diagnose and access health services.