(MSc Thesis) MMG-CLIP: Automated Mammography Reporting through Image-to-Text Translation

Recently medical image-text datasets have gained growing utilization in the development of deep learning applications, including automated radiology report generation models. In this work, we tackle the task of automated mammography report generation following Breast Imaging Reporting & Data System (BI-RADS) guidelines. We utilize an image-label and exam reports datasets, along with text prompting techniques, to generate a well-structured text report that supports training.