Year: 2025
Type: dissertation
Abstract: Medical reports are essential for conveying conclusions and diagnoses from medical scans, such as X-ray scans, in order to proceed with the patient’s treatment. Especially lung X-rays, which are used in diagnosing and treating diseases such as pneumonia, pulmonary diseases, COVID-19, lung cancer, etc. The traditional way of writing medical reports for such X-ray images is to write them manually, by specialists. This is a time-consuming, hard-to-keep-up with the increasing demand, and error-prone process since it is done by the specialists themselves after taking the scans. This prompts to make the process more automated so as to save time and lessen the probability of a mistake in diagnosis. Therefore, automating the process of writing the medical reports from chest X-ray images has become a critical need in the field of radiology as nowadays there are a lot of detrimental and even fatal diseases of the lungs that need to be treated as soon as possible. For this purpose, the machine learning approach using encoder-decoder transformers and vision transformers is a potential candidate for getting numeric results of an effective model to generate X-ray reports on two datasets: IU-Xray and MIMIC-CXR
Author: Bauyrzhanova, Aigerim
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7198807364