Poster No:
1938
Submission Type:
Abstract Submission
Authors:
Mohsen Karami1, Jacob Matthews2, Jonathan Polimeni3, Avery Berman4, Kâmil Uludağ5, J. Jean Chen6
Institutions:
1University of Toronto, Toronto, Ontario, 2Rotman Research Institute, Toronto, Ontario, 3Massachusetts General Hospital and Harvard Medical School, Boston, MA, 4Carleton University, Carleton , Ontario, 5University Health Network, Toronto, ON, 6Rotman Research Institute, Baycrest Health Sciences, Toronto, MT
First Author:
Co-Author(s):
Jonathan Polimeni
Massachusetts General Hospital and Harvard Medical School
Boston, MA
J. Jean Chen
Rotman Research Institute, Baycrest Health Sciences
Toronto, MT
Introduction:
Advancements in magnetic resonance imaging (MRI) techniques have paved the way for comprehensive, non-invasive assessments of cerebral vascular health. This study presents an innovative approach to whole-brain simultaneous multi-slice estimation of key vascular parameters, including vessel size, blood oxygenation, blood volume, T1, T2, T2*, and B0. Utilizing a pseudo-random sequence, we employ deep-learning-based parameter estimation to achieve fast and accurate results while maintaining computational feasibility. This integration, validated by microsphere-based methods, aims to enhance the precision and reliability of vascular measurements, providing invaluable insights into cerebral physiology and pathophysiology.
Methods:
In this study, the optimized sequence was developed and implemented on a Siemens 3T Prisma scanner with a 64-channel head coil. Computational optimization techniques were employed to derive this sequence, which features multiple pseudo-random curve patterns capable of simultaneously estimating vessel size, blood oxygenation, blood volume, T1, T2, T2*, and B0. An SMS factor of 5 or 6 was implemented to achieve whole-brain coverage in less than 100 seconds, with a matrix size of 128x128x128, achieving isotropic 2x2x2 mm voxels.
The 3D voxel assumption was based on literature, and randomly distributed blood vessels were used to generate the dictionary. A fully connected 4-layer neural network was utilized for parameter estimation, with layer sizes of 1250 (number of samples) x 2500 x 500 x 7. This approach significantly reduces the dictionary size and estimation time by up to 100,000 times, demonstrating the efficiency and feasibility of this method.
Six phantoms consisting of different sphere sizes, oxygenation levels, and volume fractions were built with polystyrene microspheres with varying levels of aqueous dysprosium. Additionally, five healthy subjects were scanned to demonstrate the practical implementation of this approach.
Results:
The coefficient of determination (R²) was used to evaluate the performance of this approach. The R² values for all parameters were found to be as follows: T1: 0.97±0.02, T2: 0.93±0.06, T2*: 0.88±0.11, B0: 0.79±0.13, SO2: 0.70±0.22, CBV: 0.86±0.14, Radius: 0.69±0.28.
These results corroborate the highest degree of accuracy for T1 and CBV estimation, while showing lower accuracy for Radius and SO2 estimation. This consistency was observed across different phantoms and subjects, indicating the robustness of our approach.
Parameter estimation maps were derived from all healthy subjects and were also validated against standard relaxometry for T1, T2, T2*, and B0 estimation. This validation confirms the reliability and accuracy of our results, demonstrating that the deep-learning-based parameter estimation provides precise measurements that align with established techniques.
Conclusions:
his study demonstrates the effectiveness of our deep-learning-based approach for simultaneous multi-slice estimation of multiple vascular parameters in the brain. The method proves to be both fast and accurate, significantly reducing the dictionary size and estimation time by up to 100,000 times. This efficiency makes it a viable clinical approach for quick parameter estimation, allowing for detailed and reliable assessments of cerebral vascular health in a timely manner. The robustness and precision of this method, validated through both phantoms and healthy subjects, suggest its potential for broad applications in neurological research and improved diagnostic accuracy in clinical settings.
Modeling and Analysis Methods:
Methods Development 2
Novel Imaging Acquisition Methods:
Imaging Methods Other 1
Keywords:
Data analysis
Design and Analysis
Modeling
MRI PHYSICS
1|2Indicates the priority used for review
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
I accept
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information.
Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
I am submitting this abstract as an original work to be reproduced. I am available to be the “source party” in an upcoming team and consent to have this work listed on the OSSIG website. I agree to be contacted by OSSIG regarding the challenge and may share data used in this abstract with another team.
Please indicate below if your study was a "resting state" or "task-activation” study.
Other
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Healthy subjects
Was this research conducted in the United States?
No
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel?
NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Yes
Were any animal research approved by the relevant IACUC or other animal research panel?
NOTE: Any animal studies without IACUC approval will be automatically rejected.
Not applicable
Please indicate which methods were used in your research:
Structural MRI
Computational modeling
For human MRI, what field strength scanner do you use?
3.0T
Which processing packages did you use for your study?
SPM
Provide references using APA citation style.
Christen, T., Pannetier, N. A., Ni, W., Qiu, D., Moseley, M. E., Schuff, N., & Zaharchuk, G. (2014). MR vascular fingerprinting: A new approach to compute cerebral blood volume, mean vessel radius, and oxygenation maps in the human brain. NeuroImage, 89, 262-270. https://doi.org/10.1016/j.neuroimage.2013.11.052
Christen, T., Delphin, A., Boux, F., Brossard, C., Coudert, T., Warnking, J. M., Lemasson, B., Barbier, E. L., & Christen, T. (2024). Enhancing MR vascular Fingerprinting through realistic microvascular geometries. Imaging Neuroscience. https://doi.org/10.1162/imag_a_00377
Barrier, A., Coudert, T., Delphin, A., Lemasson, B., & Christen, T. (2024). MARVEL: MR Fingerprinting with Additional micRoVascular Estimates using bidirectional LSTMs. MICCAI 2024 - Open Access. https://papers.miccai.org/miccai-2024/paper/2226_paper.pdf
Monga, A., Singh, D., Moura, H. L. de, Zhang, X., Zibetti, M. V. W., & Regatte, R. R. (2024). Emerging trends in magnetic resonance fingerprinting for quantitative biomedical imaging applications: A review. Bioengineering, 11(3), 236. https://doi.org/10.3390/bioengineering11030236
McGee, K. P., Sui, Y., Witte, R., Panda, A., Campeau, N., Rodrigues-Mostardeiro, T., Sobl, N., Ravaioli, U., Zhang, S. L., Larson, N. B., Schwarz, C. G., & Gunter, J. L. (2024). Synthesis of MR fingerprinting information from magnitude-only MR imaging data using a parallelized, multi-network U-Net convolutional neural network. Frontiers in Radiology. https://doi.org/10.3389/fradi.2024.1498411
Yes
Please select the country that the first author on this abstract resides and works in from the UNESCO Institute of Statistics and World Bank List of Low and Middle Income Countries (based on gross national income per capita).
Iran, Islamic Rep.