Robust brain age estimation from structural MRI with contrastive learning

A new multimodal BrainAGE framework has been developed, integrating structural MRI and AI-synthesized cerebral blood volume measures to enhance brain age estimation. This approach utilizes two separate 3D VGG-based networks, trained on 2,851 scans from 13 datasets, to improve predictions related to neurobiological aging and disease risk, particularly in conditions like Alzheimer's disease and mild cognitive impairment.

A new study introduces CLOP, a semi-supervised loss function aimed at enhancing contrastive learning by preventing dimensional collapse in embeddings. This research identifies a critical learning-rate threshold that, if exceeded, leads to ineffective solutions in standard contrastive methods. Through experiments on various datasets, CLOP demonstrates improved performance in image classification and object detection tasks.

A new foundation model named NeuroFM has been developed to enhance AI-based analysis in neuropathology, specifically targeting the unique characteristics of neurological diseases. This model addresses the limitations of existing general-purpose models that primarily focus on surgical pathology data, which often overlooks critical morphological patterns in neurodegenerative diseases like Alzheimer's and Parkinson's.

The introduction of the Degradation Guided Generative Adversarial Network (DGGAN) marks a significant advancement in endoscopic video enhancement, addressing the challenges posed by poor image quality during surgical procedures. This framework utilizes degradation-aware techniques to enhance video quality in real-time, ensuring that critical anatomical details are visible to surgeons.