UCL-CNT Early Career Investigator Award in Neuroimaging Techniques 2023 presented to Yukun Zhou

by Stefanie Rudrich
Head of Marketing (Brain Products)

On July 11, 2024, the UCL-CNT Early Career Investigator Award in Neuroimaging Techniques 2023 was presented to Yukun Zhou.

Yukun is a final-year PhD student in the Department of Medical Physics at University College London (UCL) and working between the Centre for Medical Image Computing (CMIC) and Moorfields Eye Hospital (MEH). He received the award for his outstanding research in retinal image analysis which provides a new approach to detecting and observing neurological conditions.

Initiated in 2007, this annual award aims to acknowledge an exceptional contribution by a UCL student or staff member in the early stages of their career in the field of Neuroimaging. Brain Products GmbH and Brain Products UK are proud to sponsor the award, which includes a trophy, £1000 and a certificate – signed by the President and Provost of UCL, Professor Michael Spence and Professor Louis Lemieux (chair of the UCL-CNT).

UCL-CNT Early Career Investigator Award in Neuroimaging Techniques 2023 presented to Yukun Zhou

From left to right: Professor Daniel Alexander (director of the CMIC), Yukun Zhou (2023 CNT Early Career Award winner) and Professor Louis Lemieux (Chair of the UCL-CNT).

More about Yukun Zhou
(Excerpt from the nomination statement by Prof. Daniel Alexander)

… Yukun made two major innovations during his PhD: AutoMorph [1] and RETFound [2]. AutoMorph, the first publicly available tool for automated retinal image analysis, quantifies morphology indices such as the size of optic nerve head and optic cup which are informative for neurological conditions. AutoMorph now underpins the standard data processing pipeline at MEH; every retinal fundus photograph collected from clinical routine now goes through AutoMorph. This hugely benefits cross-sectional and longitudinal epidemiology research, supporting the modelling of e.g. neurodegenerative disease progression and the identification of early disease markers. It already supports large clinical studies related to neurological diseases, such as schizophrenia [3] and Alzheimer’s disease [4].

RETFound is the first foundation model in ophthalmology and was published in Nature in October 2023. RETFound automatically learns features of retinal scans, requiring no labels, and can be efficiently fine-tuned to a wide variety of downstream disease detection tasks including ocular disease, neurodegenerative diseases, such as Parkinson’s, and wider systemic conditions such as cardio-vascular disease. While there is substantial international interest in foundation models for medicine, RETFound is the first example of a truly generalisable and actionable model in this space. RETFound is publicly available to expedite the progress of AI in medicine and facilitate the exploration of downstream applications, including neurodegenerative disease detection.

Yukun’s work promotes non-invasive monitoring of neurology health via the retina, facilitating the early detection of wide-ranging conditions. The inception of the projects was a collaborative effort among Yukun and the co-authors of the papers. Yukun designed and implemented the model, identifying the opportunity to translate advanced AI techniques to healthcare applications. He ran all the experiments reported in the paper and led the drafting and revision. The work is a remarkable achievement, fully deserving the high-profile publication in Nature as well as the substantial attention and news coverage (AltMetric score > 900 for RETFound [2]) it has received since. It would not have been possible without Yukun’s insights, meticulous work, and dedication to completion …

On behalf of the whole Brain Products and Brain Products UK team, congratulations, Yukun, on winning this award! We will continue to follow your research and wish you all the best for your future.

Most recent articles

References

[1] Zhou, Y., Wagner, S.K., Chia, M.A., Zhao, A., Xu, M., Struyven, R., Alexander, D.C. and Keane, P.A., 2022. AutoMorph: automated retinal vascular morphology quantification via a deep learning pipeline. Translational vision science & technology, 11(7), pp.12-12.

[2] Zhou, Y., Chia, M.A., Wagner, S.K., Ayhan, M.S., Williamson, D.J., Struyven, R.R., Liu, T., Xu, M., Lozano, M.G., Woodward-Court, P. and Kihara, Y., 2023. A foundation model for generalizable disease detection from retinal images. Nature, pp.1-8.

[3] Wagner, S.K., Cortina-Borja, M., Silverstein, S.M., Zhou, Y., Romero-Bascones, D., Struyven, R.R., Trucco, E., Mookiah, M.R., MacGillivray, T., Hogg, S. and Liu, T., 2023. Association Between Retinal Features From Multimodal Imaging and Schizophrenia. JAMA psychiatry, 80(5), pp.478-487.

[4] Yousefzadeh, N., Tran, C., Ramirez-Zamora, A., Chen, J., Fang, R. and Thai, M.T., 2023. LAVA: Granular Neuron-Level Explainable AI for Alzheimer’s Disease Assessment from Fundus Images. arXiv preprint arXiv:2302.03008.