CIKITSA.AI - An AI Powered X-Ray Analysis And Medical Report Generation

Abstract

The necessity for sophisticated technologies to guarantee precise diagnosis, localisation, and reporting is highlighted by the rising incidence of cardiothoracic illnesses. As one of the most widely available radiological tests, chest X-rays are essential for the diagnosis of a number of lung conditions. It is still difficult to efficiently use these weakly labelled datasets for deep learning applications, even with the accumulation of enormous X-ray imaging studies kept in hospital Picture Archiving and Communication Systems (PACS). In this study, we present CIKITSA.AI, an AI-powered system that can diagnose and localise a variety of cardiothoracic diseases by analysing chest X-ray pictures. The method achieves a validation accuracy of 85% by using pretrained deep learning models like ResNet-50, which were trained on the “ChestX-ray14” dataset, which consists of 112,121 frontal chest X-rays annotated with fourteen thoracic illness categories. Using the Gemini API and natural language processing, pathological labels were retrieved and verified by medical professionals. CIKITSA.AI is available through a secure online application that allows users to upload X-rays for automated reporting and produces succinct, easily navigable medical summaries. Its ability to diagnose eight prevalent cardiothoracic diseases and its potential to transform healthcare delivery are demonstrated by extensive validation against expert clinical reports. By addressing important gaps in early diagnosis and treatment, the system opens the door to patient-centered, effective, and easily accessible healthcare. Keywords: CIKITSA.AI,Computer-Aided Detection (CAD), Automated Diagnosis, ResNet, Deep Learning, Chest XRay, Cardiothoracic Disorders.

Ähnliche Arbeiten