همایش بین المللی هوش مصنوعی

صفحه اصلی / دومین همایش بین المللی هوش مصنوعی

Fibroglandular Tissue Classification in Breast MRI: A Comparative Study of Automated Decision Strategies

نویسندگان :

Meysam Khalaj¹ Arvin Arian² Ala Torabi³ Nasrin Ahmadinejad⁴ Masoumeh Gity⁵ Seyedeh Nooshin Miratashi Yazdi⁶ Mohammad Pooya Afshari⁷ Melika Sadeghi Tabrizi⁸ Hamid Soltanian-Zadeh⁹

1- دانشگاه تهران 2- دانشگاه علوم پزشکی تهران 3- دانشگاه علوم پزشکی تهران 4- دانشگاه علوم پزشکی تهران 5- دانشگاه علوم پزشکی تهران 6- دانشگاه علوم پزشکی تهران 7- دانشگاه تهران 8- دانشگاه تهران 9- دانشگاه تهران

کلمات کلیدی :

Fibroglandular Tissue Classification،Breast MRI،BI-RADS Assessment،Deep Learning،Shannon Entropy

چکیده :

Fibroglandular tissue (FGT) assessment in breast magnetic resonance imaging (MRI) is a key factor in breast cancer risk evaluation and follows the BI-RADS lexicon standard. Most automated methods have focused on segmentation, while classification-based approaches remain limited. Previous studies have often analyzed both breasts together, overlooking BI-RADS recommendations for side-specific evaluation and alternative decision strategies. This study investigates three assessment methods for automated FGT classification: the conventional BI-RADS Maximum Rule, a Probability Averaging Rule for bilateral integration, and a Lower-Uncertainty Rule that uses Shannon entropy to prioritize more confident predictions. These strategies were tested using three deep learning architectures, including MobileNetV2, ResNeXt-26, and a hybrid ViT-ResNet model, on 654 pre-contrast T1-weighted breast MRI scans. Across ten independent runs, the ViT-ResNet model with the Probability Averaging Rule achieved the highest performance, with an accuracy of 0.85, an F1 score of 0.84, and a Cohen’s kappa of 0.78. Both proposed strategies surpassed the conventional rule, and the expert-annotated dataset is publicly released to enable reproducible research.