Felix Pinkston
May 23, 2025 10:32
AI technology has enabled the identification of atrial fibrillation in brain MRIs, potentially preventing strokes through early detection, according to new research.
Artificial intelligence (AI) is revolutionizing the medical field by offering new methods for stroke prevention. According to an article by NVIDIA, researchers have developed an AI model capable of analyzing routine brain MRIs to identify atrial fibrillation (AFib), a common precursor to strokes.
AI Model for Early Detection
In a study published in the journal Cerebrovascular Diseases, scientists from Royal Melbourne Hospital introduced a deep learning model that detects AFib by recognizing subtle patterns in MRI scans. This neural network, named ConvNeXt, was trained on MRIs from 235 stroke patients, achieving an impressive accuracy rate of 84% in distinguishing AFib-induced strokes from those caused by other factors.
Significance of AFib Detection
Atrial fibrillation is a major cause of ischaemic strokes, which account for nearly 90% of all stroke cases. The ability to detect AFib early could significantly reduce the incidence of these strokes. Bernard Yan, professor and neurologist at Royal Melbourne Hospital, emphasized the importance of efficient AFib detection in preventing strokes. The condition often goes undiagnosed due to its subtle signs, even in patients with existing brain scans.
Technological Advancements
The researchers utilized NVIDIA A100 TensorCore GPUs and other NVIDIA technologies, such as CUDA 12.1, cuDNN, and NVIDIA Apex, to train their AI model. These advancements enabled the model to process extensive data sets with high precision, facilitating the identification of AFib in a cost-effective and less invasive manner compared to traditional ECG and cardiac monitoring methods.
Future Implications
While the results are promising, the research team aims to validate their findings with a larger sample size and seek external validation for broader applicability. If successful, this AI-driven approach could transform stroke prevention strategies, making point-of-care MRI analysis a feasible alternative for early AFib detection.
For more detailed information, you can read the full study on the NVIDIA blog.
Image source: Shutterstock
Credit: Source link
