Principal Investigator: Dr Zongyuan Ge
Monash UniversityTags: 51922, Machine Learning, medical imaging, opthimology
1a: Retinal fundus photograph is an effective media to observe retinal abnormality. That is where ophthalmic organs and diseases can be visually detected. A recent study shows that the information like gender, age, blood pressure, body mass index could also be measured from the photograph via deep learning to some extent . Obviously, this study demonstrates that there is rich information in the photograph and novel features of diseases could be discovered from the fundus image. The primary aspect of this project is to try to predict the vision-related measurements (such as visual acuity and spherical power) and health indicators (such as heart variability) from the fundus image via deep learning. This quantification of these measurements and indicators could further evaluate the risk of systemic disease.
1b: Our work aligns with UK Biobank’s stated purpose by improving the prevention and early detection of myopia, hypopsia and systemic disease.
1c: We will use the technology of machine learning and computer vision to train a model that could automatically predict the risks of myopia, arrhythmia and etc. by extract information from retinal fundus photographs.
We are requesting a complete dataset of left or right fundus images with paired measurements which are listed in our submission information.
Dataset size: 84,767 participants.
. Poplin, R., Varadarajan, A.V., Blumer, K., Liu, Y., McConnell, M.V., Corrado, G.S., Peng, L. and Webster, D.R., 2018. Prediction of cardiovascular risk factors from retinal fundus photographs via deep learning. Nature Biomedical Engineering, 2(3), p.158.