Media of University of Babylon | Software Department Discusses Ph.D. Dissertation on Developing a Medical Visual Question Answering System Using Deep Learning Techniques

Media of University of Babylon - كلية الطب

Software Department Discusses Ph.D. Dissertation on Developing a Medical Visual Question Answering System Using Deep Learning Techniques

by : Duhaa Fadill Abbas

Date : 14/9/2025 48:42:06 PM

Veiw : 85

Software Department Discusses Ph.D. Dissertation on Developing a Medical Visual Question Answering System Using Deep Learning Techniques

Duhaa Fadill Abbas
As part of its continuous academic activities, the Department of Software at the College of Information Technology – University of Babylon hosted the defense of a Ph.D. dissertation presented by Nada Fadel Mohammed Yaseen, entitled “Medical Visual Question Answering Based on Deep Learning Using Multi-Modal Attention and Grad-CAM Techniques.” The dissertation was supervised by Prof. Israa Hadi Ali and was examined on Sunday, September 14, 2025, in the College Conference Hall.

The dissertation aimed to design an intelligent Medical Visual Question Answering (VQA) system utilizing advanced deep learning approaches. The proposed system processes medical images alongside natural-language clinical questions to generate precise answers that support physicians in clinical decision-making, providing a reliable “second opinion” that enhances confidence and offers valuable assistance to end-users.

On the technical level, the researcher employed a CNN Denoising Autoencoder model to extract visual features from medical images by dividing them into overlapping patches, thereby improving the accuracy of detecting abnormalities. For textual feature extraction, a BiLSTM model enhanced with BioWordVec medical embeddings was implemented to process clinical questions, coupled with an Attention Mechanism to refine contextual understanding. Additionally, the study incorporated BioBERT, a specialized biomedical language model, to achieve more efficient question representation and to handle complex and rare medical terminology with high precision.

The dissertation concluded that it is possible to develop a comprehensive medical VQA system that integrates computer vision and natural language processing techniques to provide innovative and practical solutions. Such a system holds significant potential in improving diagnostic accuracy and strengthening clinical decision support within healthcare institutions.

ت	اسماء اعضاء لجنة المناقشة	اللقب العلمي	الاختصاص الدقيق	مكان العمل	المنصب
1	د. نبيل هاشم كاغد	استاذ متمرس	ذكاء اصطناعي	جامعة بابل / كلية تكنولوجيا المعلومات	رئيساً
2	د. هدى ناجي نواف	استاذ	انظمة توصية و شبكات التواصل الاجتماعي	جامعة بابل / كلية تكنولوجيا المعلومات	عضوا
3	د. اسعد نوري هاشم	استاذ	معالجة الصور و ذكاء اصطناعي	جامعة الكوفة / كلية علوم الحاسوب و الرياضيات	عضوا
4	د. علي هادي حسن	استاذ مساعد	ذكاء اصطناعي	جامعة بابل / كلية تكنولوجيا المعلومات	عضوا
5	د.مي عبد المنعم صالح	استاذ	امنية بيانات و وسائط متعددة	جامعة بابل / كلية تكنولوجيا المعلومات	عضوا
6	د.اسراء هادي علي	استاذ	وسائط متعددة و تنقيب بيانات	جامعة بابل / كلية تكنولوجيا المعلومات	عضوا و مشرفا