Medical Visual Question Answering using Bootstrapping Language Image Pre-train model

Keecheril George Mathew, Nirmal (2024) Medical Visual Question Answering using Bootstrapping Language Image Pre-train model. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (2MB) | Preview
Preview	PDF (Configuration Manual) Download (2MB) | Preview

Abstract

Medical Visual Question Answering, or mVQA, is slowly revealing its applicability to the medical field, particularly to the enhancement of the prognosis and diagnostic features. As medical imaging is one of the diagnostic processes, it is necessary to consider how it is possible to decrease the time of analysis of images and give the results in a short time. Hence the current issue is on the particular processing of noisy medical data and the actual diagnostic output of the technology. Regarding these difficulties in mVQA, applied in this study will be the Bootstrapping Language Image Pre-Trained (BLIP) model. The study involved two key case studies: the first compared the ability of BLIP in identifying noisy medical data, for which the model achieved a validation accuracy of 51.68%. Stil moderate, this result shows that BLIP is quite proficient in dealing with complex data. The second case study was to enhance the training of the model by the track of loss values, and the validation loss decreased to 0. 0930 the final epoch. Each of the above periods can further be divided into smaller sub-periods based on general classifications of technological evolutions still used today, such as the following: Another such conclusion runs that BLIP could be beneficial, particularly in the context of medical diagnostics, for the next instances with the key channels of the image analysis enhanced as far as main steps, as well as with the greater general efficiency and accuracy of the final diagnostic conclusions. This work also shows the successful implementation of the proposed technique, BLIP, in mVQA and will be helpful for the future advancement of medical AI to contribute to the improvement of health care services.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Ain, Qurrat Ul UNSPECIFIED
Uncontrolled Keywords:	Medical Visual Question Answering; BLIP; accuracy; efficiency
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence P Language and Literature > P Philology. Linguistics > Computational linguistics. Natural language processing R Medicine > Healthcare Industry Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	20 Aug 2025 09:28
Last Modified:	20 Aug 2025 09:28
URI:	https://norma.ncirl.ie/id/eprint/8580

Actions (login required)

View Item