Related Experiment Video
Updated: Jan 11, 2026
Author Spotlight: Integrating High-Resolution Intravital Imaging and MRI to Enhance Stereotactic Body Radiation Therapy Planning
Published on: April 12, 2024
Hybrid ResNet-ViT Framework for Endometrial Lesion Analysis: A Comparative Study of MRI and CT in Endometrial Cancer
Omar H Abu-Azzam1, Amer Mahmoud Sindiani2, Salem Alhatamleh3
1Department of Obstetrics and Gynecology, Faculty of Medicine, Mutah University, Al-Karak, Jordan.
Aim:
This study aimed to evaluate and compare the diagnostic performance of computed tomography (CT) and magnetic resonance imaging (MRI) in the detection of endometrial cancer, using a deep learning approach.
Methods:
Two endometrial image sets were obtained from King Abdullah University Hospital: the KAUH Endometrial Cancer MRI dataset (KAUH-ECMD) and the KAUH Endometrial Cancer CT dataset (KAUH-ECCTD), collected from 300 patients aged between 22 and 85. A hybrid deep learning model combining ResNet50 and Vision Transformer (ViT) was applied to classify the images into three categories: benign, malignant, and normal.
Results:
The proposed ViTNet model achieved an accuracy of 90.24% in detecting endometrial cancer using MRI images and 86.99% using CT images. The MRI-based approach demonstrated superior diagnostic performance in detecting endometrial cancer compared to CT-based classification.
Conclusion:
Deep learning models utilizing MRI and CT images demonstrate high accuracy in classifying endometrial cases. MRI in particular shows promise in supporting diagnostic workflows. Future work will focus on further validating the model's ability to evaluate depth of invasion and other prognostic features.
