Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Time-Embedded Algorithm Unrolling for Computational MRI
Authors: Junno Yun, Yasar Utku Alcalar, Mehmet Akcakaya
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our extensive experiments on the fast MRI dataset, spanning various acceleration rates and datasets, demonstrate that our method effectively reduces aliasing artifacts and mitigates noise amplification, achieving state-of-the-art performance. |
| Researcher Affiliation | Academia | Junno Yun University of Minnesota EMAIL Ya sar Utku Alçalar University of Minnesota EMAIL Mehmet Akçakaya University of Minnesota EMAIL |
| Pseudocode | Yes | Algorithm 1 Time-embedded Unrolling Algorithms |
| Open Source Code | Yes | Code available at https://github.com/JN-Yun/TE-Unrolling-MRI. |
| Open Datasets | Yes | These scans were obtained from the New York University (NYU) fast MRI database [39, 68], and were acquired with appropriate institutional review board approvals. |
| Dataset Splits | Yes | For knee datasets, model training was conducted using 300 slices from 10 subjects, while testing was carried out on 380 slices from a separate set of 10 subjects [28]. For the brain dataset, training and testing were performed using 300 slices each. |
| Hardware Specification | Yes | All training processes are conducted using one NVIDIA A100-SXM4-40GB GPU. |
| Software Dependencies | No | We train Res Net for 100 epochs and U-Net for 50 epochs, using a learning rate of 5 10 4 for coronal PD/PD-FS knee data and 2 10 4for Axial T2 brain data with the Adam optimizer. |
| Experiment Setup | Yes | We train Res Net for 100 epochs and U-Net for 50 epochs, using a learning rate of 5 10 4 for coronal PD/PD-FS knee data and 2 10 4for Axial T2 brain data with the Adam optimizer. For our proposed time-embedded unrolled networks... the data fidelity scalars µt were initialized to 1.5 10 2, and the time-dependent Onsager correction parameters ρt to 1 10 1. The scaling factor τ of Fi LM in Res Net is set to 0.1. |