M
Mert R. Sabuncu
Researcher at Cornell University
Publications - 148
Citations - 12746
Mert R. Sabuncu is an academic researcher from Cornell University. The author has contributed to research in topics: Computer science & Convolutional neural network. The author has an hindex of 24, co-authored 118 publications receiving 9297 citations. Previous affiliations of Mert R. Sabuncu include Massachusetts Institute of Technology & Harvard University.
Papers
More filters
Proceedings ArticleDOI
A Convolutional Autoencoder Approach To Learn Volumetric Shape Representations For Brain Structures
Evan M. Yu,Mert R. Sabuncu +1 more
TL;DR: In this article, the learned shape descriptor is invariant to affine transformations, including shifts, rotations and scaling, and is automatically enhanced in the learned representation, while intra-subject variances are minimized.
Journal ArticleDOI
LARO: Learned acquisition and reconstruction optimization to accelerate quantitative susceptibility mapping
Jinwei Zhang,Pascal Spincemaille,Hong Zhang,Thanh D. Nguyen,Chao Li,Jiahao Li,Ilhami Kovanlikaya,Mert R. Sabuncu,Yian Wang +8 more
TL;DR: In this paper , a learned acquisition and reconstruction optimization (LARO) framework is proposed to accelerate the multi-echo gradient echo (mGRE) pulse sequence for QSM.
Posted Content
From Connectomic to Task-evoked Fingerprints: Individualized Prediction of Task Contrasts from Resting-state Functional Connectivity
TL;DR: A surface-based convolutional neural network (BrainSurfCNN) model is proposed to predict individual task contrasts from their resting-state fingerprints, and a reconstructive-contrastive loss is introduced that enforces subject-specificity of model outputs while minimizing predictive error.
Posted Content
Probabilistic Dipole Inversion for Adaptive Quantitative Susceptibility Mapping
TL;DR: PDI provides additional uncertainty estimation compared to the conventional MAP approach, meanwhile addressing the potential issue of the pre-trained CNN when test data deviates from training.
Journal ArticleDOI
Computing Multiple Image Reconstructions with a Single Hypernetwork
TL;DR: This work presents a hypernetwork-based approach, called HyperRecon, to train reconstruction models that are agnostic to hyperparameter settings, and demonstrates the method in compressed sensing, super-resolution and denoising tasks, using two large-scale and publicly-available MRI datasets.