부산대학교 도서관

Report generation is a time-consuming process in the medical world. It asks for the experience and sound medical knowledge. To save time, an overview of the report can be generated using the technologies available in the current world. This work includes detecting disorders and generating radiology reports from MRI scans of the brain by utilizing image-processing techniques and deep-learning frameworks. The agenda is to automate the process of report generation from the MRI scans. Given an MRI scan as an input, the CNN models which are trained on similar data, are used to detect the abnormalities in the scan. A report would then be generated using a transformer-based model that would list the findings from the models. The developed models are then used to automate the process of report generation for the MRI scans which in turn aid physicians in efficiently generating reports with minimal errors.