RGAN-Driven T1 Map Synthesis from OS-CMR: a step toward AI-assisted amyloidosis diagnosis
Bibliographic record
Abstract
Abstract Background T1-weighted MRI is crucial for detecting myocardial amyloidosis, where native T1 values are significantly elevated [1]. However, its clinical application is limited by motion artifacts, lengthy acquisition times, and variations in imaging protocols. OS-CMR provides functional insights with embedded T1 information, offering a potential non-invasive alternative[2]. This study introduces an advanced Residual Generative Adversarial Network (R-GAN) to synthesize high-fidelity T1 parametric maps from OS-CMR, enabling AI-assisted myocardial amyloidosis diagnosis. Purpose This study aims to develop and validate an R-GAN model for synthesizing T1 parametric maps from OS-CMR, assess its performance against Pix2Pix, and evaluate its agreement with ground-truth T1 maps Methods Two separate datasets of 1,481 matched OS-CMR and T1-weighted images was used. Preprocessing included normalization, augmentation, and image registration. The proposed R-GAN, enhanced with residual blocks, was trained and compared to Pix2Pix using PSNR, SSIM, and PCC metrics. Validation included signal intensity analysis and Extended Phase Graph (EPG) simulations, with T1 curve evaluations conducted using Python scripts and CVI42 software. Results R-GAN with augmentation outperformed other models, achieving the highest SSIM (0.810), PSNR (16.1), and PCC (0.748). In contrast, Pix2Pix yielded lower SSIM (0.254) and PCC (0.231), with the lowest performance observed in Pix2Pix without augmentation (SSIM = 0.158, PCC = 0.03)(Table1). EPG simulation display the best-matching original T1 curve with synthesis curve which had a Pearson correlation of 0.99, indicating agreement between the synthesized and ground-truth T1 maps (Figure1) Conclusion The proposed R-GAN demonstrates the feasibility of generating high-fidelity T1 maps from OS-CMR, offering a non-invasive approach for detecting myocardial amyloidosis. The model significantly outperforms Pix2Pix, particularly with augmentation, and shows strong agreement with original T1 maps. Future work will focus on validating the approach across diverse datasets and optimizing its clinical applicabilityFigure 1.R-GAN method Table1.Metrics
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 machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
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 source (direct Gemma or distilled Codex), 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".