Progressive Feature Reconstruction and Fusion to Accelerate MRI Imaging: Exploring Insights across Low, Mid, and High-Order Dimensions
Bibliographic record
Abstract
Magnetic resonance imaging (MRI) faces ongoing challenges associated with prolonged acquisition times and susceptibility to motion artifacts. Compressed Sensing (CS) principles have emerged as a significant advancement, addressing these issues by subsampling k-space data points and enabling rapid imaging. Nevertheless, the recovery of intricate details from under-sampled data remains a complex endeavor. In this study, we introduce an innovative deep learning approach tailored to the restoration of high-fidelity MRI images from under-sampled k-space data. Our method employs a cascaded reconstruction strategy that progressively restores hierarchical features and fuses them to achieve the final reconstruction. This cascade encompasses low, intermediate, and high orders of reconstruction, which is followed by a return through intermediate and low orders. At distinct reconstruction stages, we introduce a novel reconstruction block to recapture diverse frequency information crucial for image reconstruction. The other core innovation of our proposal lies in a fusion algorithm that harmonizes results from various reconstruction tiers into the final MRI image. Our methodology is validated using two distinct datasets. Notably, our algorithm achieves impressive PSNR values of 32.60 and 31.02 at acceleration factors of 4× and 8× in the FastMRI dataset along with SSIM scores of 0.818 and 0.771, outperforming current state-of-the-art algorithms. Similarly, on the Calgary–Campinas dataset, our algorithm achieves even higher PSNR values, reaching 37.68 and 33.44, which is accompanied by substantial SSIM scores of 0.954 and 0.901. It is essential to highlight that our algorithm achieves these remarkable results with a relatively lower parameter count, underscoring its efficiency. Comparative analyses against analogous methods further emphasize the superior performance of our approach, providing robust evidence of its effectiveness.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".