Accurate medical imaging segmentation is crucial for effective medical interventions, but convolutional neural
networks (CNNs) still struggle with fine-scale features and image scale variations, particularly in complex tasks
like the BraTS brain tumor segmentation challenge. To address these challenges, we propose two architectures,
FMG-Net and W-Net, that incorporate geometric multigrid principles into CNNs. Our experiments on the BraTS 2020
dataset show that FMG-Net and W-Net outperform the commonly used U-Net in tumor subcomponent segmentation accuracy
and training efficiency. The figure above shows sketches of the U-Net, FMG-Net, and W-Net architectures
for different network detphs. More details and results are available
here.
Adrian Celaya
About Me
I am currently a Ph.D. student in the Department of Computational and Applied Mathematics (CAAM) at Rice University in Houston, TX. I am co-advised by Dr. Beatrice Riviere and Dr. David Fuentes. My research focuses on developing novel techniques for 3D medical image segmentation. Previously, I served as the Information System Security Manager onboard the U.S. Navy aircraft carrier USS Carl Vinson. After leaving the Navy in August 2020, I became a research assistant in the Department of Imaging Physics at The University of Texas MD Anderson Cancer Center.
Research Interests
Current Research Topics
FMG-Net and W-Net Architectures
Weighted Normalized Boundary Loss
The Weighted Normalized Boundary Loss (WNBL) is a boundary-based loss for minimizing the Hausdorff distance for deep
learning-based medical imaging segmentation. This novel loss function has more desirable numerical properties than
current methods and weighting terms for class imbalance. The WNBL outperforms other losses when tested on
the BraTS dataset using a standard 3D U-Net and the state-of-the-art nnUNet architecture.
The WNBL is given below:
\begin{align*}
\mathcal{L}_{wnbl} = 1 - \frac{\sum_{k = 1}^{C} w_k \sum_{i = 1}^{N} \left(D_{i}^{k}\left((1 - T_{i}^{k})^2 - P_{i}^{k}\right)\right)^2}{\sum_{k = 1}^{C} w_k \sum_{i = 1}^{N} \left(D_{i}^{k}\left((1 - T_{i}^{k})^2 - T_{i}^{k}\right)\right)^2},
\end{align*}
where $N$ denotes the total number of pixels (or voxels in the 3D case), $C$ denotes the number of segmentation classes,
$P_i^k \in [0, 1]$ is the $i$-th voxel in the $k$-th class of the predicted mask, $T_i^k \in \{0, 1\}$ is the $i$-th
voxel in the $k$-th class for the ground truth, and $D_i^k \in \mathbb{R}$ is the $i$-th voxel for the $k$-th class in
the ground truth distance transform map. More details and results are available
here.
Deep Learning-Based Inversion of Surface Gravity For CO$_2$ Monitoring
CO$_2$ geological storage is a carbon capture and storage technology that captures CO$_2$ emissions from fixed
sources like power plants and stores it in underground saline formations, making it a practical and near-term
solution. However, long-term monitoring is necessary to ensure volume storage integrity. 4D gravity monitoring
is an effective low-cost and environmentally friendly technique for monitoring geological storage sites. To
predict subsurface CO$_2$ plumes, a deep learning approach was developed that outperforms traditional
inversion methods, producing high-resolution 3D subsurface reconstructions in near real-time. The figure above
highlights the results of two of our proposed approaches. More details and results can be found
here.
Grants and Fellowships
National Defense Science and Engineering (NDSEG) Fellowship
Sponsor: Department of Defense (Air Force Office of Scientific Research)Dates: September 2022 to May 2025
Media:
Adrian Celaya, Dane Grundvig and Alejandro Diaz are all PhD students in the Department of Computational Applied Mathematics and Operations Research.
First-year PhD student earned his B.A. in CAAM from Rice in 2016.
Loewenstern Fellowship
Sponsor: Rice UniversityDates: August 2021 to October 2022
Publications and Presentations
A. Celaya, B. Riviere, and D. Fuentes. "FMG-Net and W-Net: Multigrid Inspired Deep Learning Architectures For Medical Imaging Segmentation," submitted to 37th Conference on Neural Information Processing Systems, under review, 2023. | |
A. Celaya, A. Diaz, A. Balsells, B. Riviere, and D. Fuentes. "A Weighted Normalized Boundary Loss for Reducing the Hausdorff Distance in Medical Imaging Segmentation," submitted to 26th International Conference on Medical Image Computing and Computer Assisted Intervention, under review, 2023. | |
A. Celaya, B. Denel, Y. Sun, M. Araya-Polo, and A. Price. "Inversion of Time-Lapse Surface Gravity Data for Detection of 3D CO$_2$ Plumes via Deep Learning," in IEEE Transactions on Geosciences and Remote Sensing, doi: 10.1109/TGRS.2023.3273149. | |
R. Muthusivarajan, A. Celaya, J. Yung, S. Viswanath, D. Marcus, C. Chung, and D. Fuentes. "Evaluating the relationship between magnetic resonance image quality metrics and deep learning-based segmentation accuracy of brain tumors," submitted to Medical Physics, under review, 2022. | |
A. Celaya, J. A. Actor, R. Muthusivarajan, E. Gates, C. Chung, D. Schellingerhout, B. Riviere, and D. Fuentes. "PocketNet: A Smaller Neural Network For Medical Image Analysis," in IEEE Transactions on Medical Imaging, doi: 10.1109/TMI.2022.3224873. | |
E. Gates, D. Suki, A. Celaya, J. Weinberg, S. Prabhu, R. Sawaya, J. Huse, J. Long, D. Fuentes, and D. Schellingerhout. "Cellular Density in Adult Glioma, Estimated with MR Imaging Data and a Machine Learning Algorithm, Has Prognostic Power Approaching World Health Organization Histologic Grading in a Cohort of 1181 Patients," in American Journal of Neuroradiology, doi: 10.3174/ajnr.A7620. | |
E. Gates, A. Celaya, D. Suki, D. Schellingerhout, and D. Fuentes. "Technical Note: An efficient MR image data quality screening dashboard," in Journal of Applied Clinical Medical Physics, doi: 10.1002/acm2.13557. |
A. Celaya. "Inversion of Time-Lapse Surface Gravity Data for Detection of 3D CO$_2$ Plumes via Deep Learning," in 16th Annual Energy High Performance Computing Conference. Technical Talk. Houston, TX. February 2023. | |
A. Celaya. "PocketNet: A Smaller Neural Network For Medical Image Analysis," in 5th Annual SIAM Texas-Louisiana Section Meeting. Invited Minisymposium Presentation. Houston, TX. November 2022. | |
A. Celaya. "Small Convolutional Neural Networks for Efficient 3D Medical Image Segmentation," in 63rd American Association of Physicists in Medicine Annual Meeting. Virtual. July 2021. |
Software
The software MIST is a simple, fully automated framework for deep learning-based medical image segmentation. The framework can seamlessly ingest various medical imaging data and is easily expandable to test new ideas (i.e., new architectures, loss functions, etc.). MIST is open source, written in Python for PyTorch and TensorFlow. Please star the MIST repository!
Contact Me