Retinal reflectance spectra measured using hyperspectral imaging are associated with prevalent dementia in a general older Japanese population

Advances in retinal hyperspectral imaging demonstrate its potential usefulness in neurodegenerative disorders; however, community-based evidence is lacking. We examined associations between retinal reflectance spectra and prevalent dementia in 857 Hisayama Study residents aged ≥ 65 years who underwent eye and dementia examinations in 2022–2023. Retinal relative reflectance spectra were measured in four retinal regions using hyperspectral imaging; ganglion cell-inner plexiform layer (GC-IPL) thickness was assessed using optical coherence tomography (OCT).

Principal component analysis (PCA) was performed on Z-standardized reflectance values for each retinal region and all four regions combined. PCA-derived scores were constructed as the linear predictor from logistic regression models including the first four principal components, with higher scores indicating lower overall reflectance. Higher PCA-derived scores were significantly associated with higher odds of prevalent dementia in all regions, strongest in inferotemporal region (odds ratio [OR], 2.21; 95% confidence interval [CI], 1.46–3.35) and combined regions (OR, 2.20; 95% CI, 1.42–3.40).

After internal validation, the shrinkage-adjusted ORs were 2.31 and 2.52, and C-statistics were 0.707 and 0.688, respectively. Coexisting high PCA-derived scores and GC-IPL thinning further had higher odds of prevalent dementia in combined regions (OR, 3.87) and inferotemporal region (OR, 6.98). Thus, hyperspectral imaging, particularly when considered alongside OCT, may capture retinal alterations associated with dementia in older adults.

We extend our gratitude to the residents of the town of Hisayama for their participation in the survey and the staff of the Division of Health of Hisayama for their cooperation with this study. Statistical analyses were conducted using the computer resources provided by the Research Institute for Information Technology, Kyushu University, offered under the General Projects category. No compensation was received from any funding sponsor.

This study was supported in part by the Ministry of Education, Culture, Sports, Science and Technology of Japan (JSPS KAKENHI Grant Number JP22K07421, JP23K09692, JP23K09717, JP23K16330, JP23K06787, JP23K09060, JP24H00659, and JP25K13589); the Health and Labour Sciences Research Grants of the Ministry of Health, Labour and Welfare of Japan (JPMH23FA1006, JPMH24GB1002, and JPMH24FA1005); the Japan Agency for Medical Research and Development (JP25dk0207053, JP25km0405209, JP25tm0524003, JP25he2202021, and JP25dk0307136); and by JST Grant Number (JPMJPF2210 and JPMJPF2153). Department of Ophthalmology, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan Emi Ueda, Kohta Fujiwara, Yuki Tsuda, Shun Nakamura & Koh-Hei Sonoda Center for Cohort Studies, Graduate School of Medical Sciences, Kyushu University, 3-1-1 Maidashi, Higashi-ku, Fukuoka, 812-8582, Japan Emi Ueda, Takanori Honda, Mao Shibata, Satoko Sakata, Emi Oishi, Jun Hata & Toshiharu Ninomiya Department of Electrical Engineering, Graduate School of Information Science and Electrical Engineering, Kyushu University, Fukuoka, Japan Department of Epidemiology and Public Health, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan Tomoyuki Ohara, Takanori Honda, Mao Shibata, Kohta Fujiwara, Yoshihiko Furuta, Satoko Sakata, Emi Oishi, Yuki Tsuda, Shun Nakamura & Toshiharu Ninomiya Department of Neuropsychiatry, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan Department of Psychosomatic Medicine, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan Department of Medicine and Clinical Science, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan Department of Health Care Administration and Management, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan The authors declare no competing interests. Authors only used generative AI and AI-assisted technologies to improve readability and language.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Below is the link to the electronic supplementary material. Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material.