Integração de modelos de inteligência artificial para análise de imagens médicas usando Streamlit

Abstract

The advance of artificial intelligence (AI) technologies has culminated in the rapid development of applied models to solve real-life problems. One of the areas that has greatly benefited from this progress is healthcare (RAJPURKAR et al., 2022), due to the many challenges faced in medical environments. For patients, it is essential to obtain an accurate and rapid diagnosis. However, targeting both attributes simultaneously is a major challenge in healthcare, especially in times of hospital overcrowding. Regularly, only one of the measures is prioritized, which can expose patients to risks. One solution that is widely discussed in the literature is the integration of AI systems with medical procedures, in order to speed up, assist and provide greater precision to hospitals. That said, this work aims to ease the balance between the two metrics by creating a system designed to help healthcare professionals in their task of recognizing lesions and diseases in medical images. This system is made up of several pre-trained computer vision models, capable of analyzing and alerting professionals to diseases or injuries in the patient, available in a Streamlit application, which makes it easy for users to access and use the software. The system was tested by healthcare professionals and evaluated using a survey to measure its effectiveness. The feedback form didn’t get many responses, but the results were positive.