Learnable weight initialization for volumetric medical image segmentation
Kunhimon, Shahina Shaker, Abdelrahman Naseer, Muzammal Khan, Salman Khan, Fahad Shahbaz
Kunhimon, Shahina
Shaker, Abdelrahman
Naseer, Muzammal
Khan, Salman
Khan, Fahad Shahbaz
Supervisor
Department
Computer Vision
Embargo End Date
Type
Journal article
Date
License
Language
English
Collections
Research Projects
Organizational Units
Journal Issue
Abstract
Hybrid volumetric medical image segmentation models, combining the advantages of local convolution and global attention, have recently received considerable attention. While mainly focusing on architectural modifications, most existing hybrid approaches still use conventional data-independent weight initialization schemes which restrict their performance due to ignoring the inherent volumetric nature of the medical data. To address this issue, we propose a learnable weight initialization approach that utilizes the available medical training data to effectively learn the contextual and structural cues via the proposed self-supervised objectives. Our approach is easy to integrate into any hybrid model and requires no external training data. Experiments on multi-organ and lung cancer segmentation tasks demonstrate the effectiveness of our approach, leading to state-of-the-art segmentation performance. Our proposed data-dependent initialization approach performs favorably as compared to the Swin-UNETR model pretrained using large-scale datasets on multi-organ segmentation task. Our source code and models are available at: https://github.com/ShahinaKK/LWI-VMS.
Citation
S. Kunhimon, A. Shaker, M. Naseer, S. Khan, F.S. Khan, "Learnable weight initialization for volumetric medical image segmentation," Artificial Intelligence in Medicine, vol. 151, pp. 102863-102863, 2024, https://doi.org/10.1016/j.artmed.2024.102863.
Source
Artificial Intelligence in Medicine
Conference
Keywords
46 Information and Computing Sciences, 4611 Machine Learning, Algorithms, Humans, Image Processing, Computer-Assisted, Lung Neoplasms
Subjects
Source
Publisher
Elsevier