RAIAT (Radiology Artificial Intelligence Assistant Tool): Enhancing Radiologic Diagnosis Through Integrated Computer Vision and Large Language Models
The proposed project focuses on developing a novel pipeline integrating deep learning and language models to highlight overlooked lesions in radiology in real-time, prompting radiologists for a reconsideration. While the current error rate stands at 3%-5% in radiology reports, natural language processing (NLP) and computer vision can be combined to identify discrepancies between reports and images, alerting the physician promptly to missed findings. Ultimately, we aspire to establish a universally compatible pipeline that can be integrated across major clinical platforms and shared with various institutions, marking a pivotal step towards reducing radiologist human error and potentially saving lives.