A preliminary attempt to harmonize using physics-constrained deep neural networks for multisite and multiscanner MRI datasets (PhyCHarm)
In magnetic resonance imaging (MRI), variations in scan parameters and scanner specifications can result in differences in image appearance. To minimize these differences, harmonization in MRI has been suggested as a crucial image processing technique. In this study, we developed an MR physics-based...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-08-01
|
| Series: | NeuroImage |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S1053811925003647 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849718224814342144 |
|---|---|
| author | Gawon Lee Dong Hye Ye Se-Hong Oh |
| author_facet | Gawon Lee Dong Hye Ye Se-Hong Oh |
| author_sort | Gawon Lee |
| collection | DOAJ |
| description | In magnetic resonance imaging (MRI), variations in scan parameters and scanner specifications can result in differences in image appearance. To minimize these differences, harmonization in MRI has been suggested as a crucial image processing technique. In this study, we developed an MR physics-based harmonization framework, Physics-Constrained Deep Neural Network for multisite and multiscanner Harmonization (PhyCHarm). PhyCHarm includes two deep neural networks: (1) the Quantitative Maps Generator to generate T1- and M0-maps and (2) the Harmonization Network. We used an open dataset consisting of 3T MP2RAGE images from 50 healthy individuals for the Quantitative Maps Generator and a traveling dataset consisting of 3T T1w images from 9 healthy individuals for the Harmonization Network. PhyCHarm was evaluated using the structural similarity index measure (SSIM), peak signal-to-noise ratio (PSNR), and normalized-root-mean square error (NRMSE) for the Quantitative Maps Generator, and using SSIM, PSNR, and volumetric analysis for the Harmonization network, respectively. PhyCHarm demonstrated increased SSIM and PSNR, the highest Dice score in the FSL FAST segmentation results for gray and white matter compared to U-Net, Pix2Pix, CALAMITI, and HarmonizingFlows. PhyCHarm showed a greater reduction in volume differences after harmonization for gray and white matter than U-Net, Pix2Pix, CALAMITI, or HarmonizingFlows. As an initial step toward developing advanced harmonization techniques, we investigated the applicability of physics-based constraints within a supervised training strategy. The proposed physics constraints could be integrated with unsupervised methods, paving the way for more sophisticated harmonization qualities. |
| format | Article |
| id | doaj-art-6404044b6f3c43fa8eba68d9b15a19bd |
| institution | DOAJ |
| issn | 1095-9572 |
| language | English |
| publishDate | 2025-08-01 |
| publisher | Elsevier |
| record_format | Article |
| series | NeuroImage |
| spelling | doaj-art-6404044b6f3c43fa8eba68d9b15a19bd2025-08-20T03:12:26ZengElsevierNeuroImage1095-95722025-08-0131712136110.1016/j.neuroimage.2025.121361A preliminary attempt to harmonize using physics-constrained deep neural networks for multisite and multiscanner MRI datasets (PhyCHarm)Gawon Lee0Dong Hye Ye1Se-Hong Oh2Department of Biomedical Engineering, Hankuk University of Foreign Studies, Yongin, the Republic of KoreaDepartment of Computer Science, Georgia State University, Atlanta, GA, USA; Corresponding author at: Department of Computer Science, Georgia State University, 33 Gilmer Street SE, Atlanta, GA, 30303, USA.Department of Biomedical Engineering, Hankuk University of Foreign Studies, Yongin, the Republic of Korea; Department of Diagnostic Radiology, Diagnostics Institute, The Cleveland Clinic Foundation, Cleveland, OH, USA; Corresponding author at: Department of Biomedical Engineering, Hankuk University of Foreign Studies, 81, Oedae-ro, Yongin, 17035, the Republic of Korea.In magnetic resonance imaging (MRI), variations in scan parameters and scanner specifications can result in differences in image appearance. To minimize these differences, harmonization in MRI has been suggested as a crucial image processing technique. In this study, we developed an MR physics-based harmonization framework, Physics-Constrained Deep Neural Network for multisite and multiscanner Harmonization (PhyCHarm). PhyCHarm includes two deep neural networks: (1) the Quantitative Maps Generator to generate T1- and M0-maps and (2) the Harmonization Network. We used an open dataset consisting of 3T MP2RAGE images from 50 healthy individuals for the Quantitative Maps Generator and a traveling dataset consisting of 3T T1w images from 9 healthy individuals for the Harmonization Network. PhyCHarm was evaluated using the structural similarity index measure (SSIM), peak signal-to-noise ratio (PSNR), and normalized-root-mean square error (NRMSE) for the Quantitative Maps Generator, and using SSIM, PSNR, and volumetric analysis for the Harmonization network, respectively. PhyCHarm demonstrated increased SSIM and PSNR, the highest Dice score in the FSL FAST segmentation results for gray and white matter compared to U-Net, Pix2Pix, CALAMITI, and HarmonizingFlows. PhyCHarm showed a greater reduction in volume differences after harmonization for gray and white matter than U-Net, Pix2Pix, CALAMITI, or HarmonizingFlows. As an initial step toward developing advanced harmonization techniques, we investigated the applicability of physics-based constraints within a supervised training strategy. The proposed physics constraints could be integrated with unsupervised methods, paving the way for more sophisticated harmonization qualities.http://www.sciencedirect.com/science/article/pii/S1053811925003647MRIHarmonizationPhysics-constrained deep neural networksMultisite datasetBloch equation |
| spellingShingle | Gawon Lee Dong Hye Ye Se-Hong Oh A preliminary attempt to harmonize using physics-constrained deep neural networks for multisite and multiscanner MRI datasets (PhyCHarm) NeuroImage MRI Harmonization Physics-constrained deep neural networks Multisite dataset Bloch equation |
| title | A preliminary attempt to harmonize using physics-constrained deep neural networks for multisite and multiscanner MRI datasets (PhyCHarm) |
| title_full | A preliminary attempt to harmonize using physics-constrained deep neural networks for multisite and multiscanner MRI datasets (PhyCHarm) |
| title_fullStr | A preliminary attempt to harmonize using physics-constrained deep neural networks for multisite and multiscanner MRI datasets (PhyCHarm) |
| title_full_unstemmed | A preliminary attempt to harmonize using physics-constrained deep neural networks for multisite and multiscanner MRI datasets (PhyCHarm) |
| title_short | A preliminary attempt to harmonize using physics-constrained deep neural networks for multisite and multiscanner MRI datasets (PhyCHarm) |
| title_sort | preliminary attempt to harmonize using physics constrained deep neural networks for multisite and multiscanner mri datasets phycharm |
| topic | MRI Harmonization Physics-constrained deep neural networks Multisite dataset Bloch equation |
| url | http://www.sciencedirect.com/science/article/pii/S1053811925003647 |
| work_keys_str_mv | AT gawonlee apreliminaryattempttoharmonizeusingphysicsconstraineddeepneuralnetworksformultisiteandmultiscannermridatasetsphycharm AT donghyeye apreliminaryattempttoharmonizeusingphysicsconstraineddeepneuralnetworksformultisiteandmultiscannermridatasetsphycharm AT sehongoh apreliminaryattempttoharmonizeusingphysicsconstraineddeepneuralnetworksformultisiteandmultiscannermridatasetsphycharm AT gawonlee preliminaryattempttoharmonizeusingphysicsconstraineddeepneuralnetworksformultisiteandmultiscannermridatasetsphycharm AT donghyeye preliminaryattempttoharmonizeusingphysicsconstraineddeepneuralnetworksformultisiteandmultiscannermridatasetsphycharm AT sehongoh preliminaryattempttoharmonizeusingphysicsconstraineddeepneuralnetworksformultisiteandmultiscannermridatasetsphycharm |