Enhancing Image Segmentation for ICH through Transfer Learning from Stroke MRI to ICH CT

DOI

10.2991/978-94-6463-540-9_56How to use a DOI?

Keywords

Computer Vision; Image Segmentation; Transfer Learning

Abstract

A serious brain condition with a high death rate, intracranial hemorrhage (ICH) requires a precise and timely diagnosis. While Computed Tomography (CT) is commonly used for its speed and accessibility, its diagnostic accuracy is limited compared to Magnetic Resonance Imaging (MRI). However, the latter is often less accessible and more time-consuming. This study addresses this challenge by leveraging transfer learning techniques to enhance ICH detection in CT images using the detailed features available in MRI scans. Specifically, this study employed the ATLAS v2.0 stroke MRI dataset as the source domain and adapt it to a target dataset of ICH CT images from Kaggle. Three models, including U-Net, Residual U-Net, and Attention U-Net, are trained and evaluated on both datasets to assess their performance and transfer learning capabilities. The models are pretrained on the MRI dataset and fine-tuned on the CT dataset, utilizing different learning rates for encoders and decoders to preserve and adapt features effectively. The results indicate that the Attention U-Net outperforms the other models, demonstrating superior performance in both training and testing metrics. This research demonstrates the potential of combining the rapid assessment capabilities of CT with the detailed visualization strengths of MRI, setting a new standard in the diagnosis of ICH. The findings suggest that transfer learning from stroke MRI to ICH CT can significantly improve segmentation accuracy, thereby enhancing diagnostic efficiency and patient outcomes.

Copyright

© 2024 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Lianghan Dong
PY  - 2024
DA  - 2024/10/16
TI  - Enhancing Image Segmentation for ICH through Transfer Learning from Stroke MRI to ICH CT
BT  - Proceedings of the 2024 2nd International Conference on Image, Algorithms and Artificial Intelligence (ICIAAI 2024)
PB  - Atlantis Press
SP  - 563
EP  - 572
SN  - 2352-538X
UR  - https://doi.org/10.2991/978-94-6463-540-9_56
DO  - 10.2991/978-94-6463-540-9_56
ID  - Dong2024
ER  -