A global team of researchers led by the University of Manchester used Artificial Intelligence (AI) to investigate the role genetics play in forming the structure of the heart’s left ventricle. They utilized unsupervised deep learning to examine over 50,000 three-dimensional MRI images from the UK Biobank. The goal was to establish a better understanding of how the heart’s structure is linked to genetics, which could potentially aid research into genetically influenced congenital heart disease.
The study began with data collection and preparation. More than 50,000 3D MRI images of hearts were selected from the UK Biobank database, providing a vast base of data for analyzing the structure of the left ventricle.
To make sense of this vast data set, the researchers trained unsupervised deep learning models. These models discover patterns and attributes in data without prior labeling. The important part of this process was the extraction of geometric features from images symbolizing the left ventricle from the myocardial MRI data.
Subsequently, the research group carried out genome-wide and transcriptome-wide association studies (GWAS and TWAS) with the extracted features, to test for an association between genetics and the structure of the left ventricle.
Through this analytical process, they identified 49 new genetic locations that were strongly linked with the heart’s morphology, and found an additional 25 moderately associated locations.
Professor Alejandro F. Frangi, one of the lead researchers, proclaimed the achievement as a significant stride forward. He pointed out the benefits of utilizing AI to understand the genetic basis of the left ventricle’s structure. Frangi noted the breakthrough made possible by AI, in terms of revealing novel genetic locations associated with various cardiovascular phenotypes.
The research findings offer critical insights into the genetic factors influencing cardiovascular health. The application of AI and machine learning in big data analysis helps in addressing significant problems in cardiovascular research. It also opens up new possibilities for developing targeted therapies and precision medicine.
In a related development, AI models have been previously employed to generate detailed 3D maps of organs, including the human brain, in the EU’s wide-scale Human Brain Project (HBP). However, this research on the heart’s structure goes a step further in associating organ structure with genetics. It provides a deeper understanding of the morphological aspects of the heart and its genetic influences.
