This is a page that curates AI-related papers published worldwide. All content here is summarized using Google Gemini and operated on a non-profit basis. Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.
A Multimodal Deep Learning Approach for White Matter Shape Prediction in Diffusion MRI Tractography
Created by
Haebom
Author
Yui Lo, Yuqian Chen, Dongnan Liu, Leo Zekelman, Jarrett Rushmore, Yogesh Rathi, Nikos Makris, Alexandra J. Golby, Fan Zhang, Weidong Cai, Lauren J. O'Donnell
Outline
This paper proposes Tract2Shape, a novel multimodal deep learning framework for white matter fascicle morphometry. To address the computational cost of existing voxel-based methods, we leverage geometric (point cloud) and scalar (tabular) features to predict ten white matter fascicle morphometry measurements. We improve model efficiency by predicting five key morphometry components using a dimensionality reduction algorithm. We train and evaluate the model on two datasets: HCP-YA and PPMI, and compare its performance to state-of-the-art models on the HCP-YA dataset. Cross-dataset evaluation on the PPMI dataset demonstrates its generalization performance. Tract2Shape outperforms existing state-of-the-art deep learning models across all ten morphometry measurements, achieving high Pearson correlation coefficients and low nMSE. In conclusion, Tract2Shape enables fast, accurate, and generalizable prediction of white matter morphometry measurements, supporting the analysis of large-scale datasets.
Takeaways, Limitations
•
Takeaways:
◦
Providing a fast, accurate, and generalizable deep learning framework for measuring white matter fascicle morphology.
◦
Expanding the possibilities of analyzing large datasets.
◦
Verifying performance improvement through multi-mode input and PCA.
◦
Demonstrated high generalization performance in cross-dataset evaluation.
•
Limitations:
◦
This paper only addresses specific morphometric measurements (10). Further research is needed to determine their applicability to other morphometric measurements.
◦
Performance may vary depending on the characteristics of the dataset used. Additional validation on various datasets is required.
◦
Further research is needed to determine the interpretability of the model.
