Search:
Match:
2 results
Medical Imaging#AI in Medical Imaging🔬 ResearchAnalyzed: Jan 3, 2026 06:19

ProDM: AI for Motion Artifact Correction in Chest CT

Published:Dec 31, 2025 16:29
1 min read
ArXiv

Analysis

This paper presents a novel AI framework, ProDM, to address the problem of motion artifacts in non-gated chest CT scans, specifically for coronary artery calcium (CAC) scoring. The significance lies in its potential to improve the accuracy of CAC quantification, which is crucial for cardiovascular disease risk assessment, using readily available non-gated CT scans. The use of a synthetic data engine for training, a property-aware learning strategy, and a progressive correction scheme are key innovations. This could lead to more accessible and reliable CAC scoring, improving patient care and potentially reducing the need for more expensive and complex ECG-gated CT scans.
Reference

“ProDM significantly improves CAC scoring accuracy, spatial lesion fidelity, and risk stratification performance compared with several baselines.”

PermalinkArXiv
Research Paper#Deep Learning, Quantization, Mixed-Precision Training🔬 ResearchAnalyzed: Jan 3, 2026 19:34

MoR: Dynamic Mixed-Precision Training

Published:Dec 28, 2025 06:28
1 min read
ArXiv

Analysis

This paper introduces Mixture-of-Representations (MoR), a novel framework for mixed-precision training. It dynamically selects between different numerical representations (FP8 and BF16) at the tensor and sub-tensor level based on the tensor's properties. This approach aims to improve the robustness and efficiency of low-precision training, potentially enabling the use of even lower precision formats like NVFP4. The key contribution is the dynamic, property-aware quantization strategy.
Reference

“Achieved state-of-the-art results with 98.38% of tensors quantized to the FP8 format.”

PermalinkArXiv