Multimodal longitudinal in vivo retinal and cerebral imaging-based framework for neurodegenerative diseases using mouse model of Alzheimer's Disease

Abstract

Purpose : Neurodegenerative diseases in the brain such as Alzheimer's Diseases (AD) and other types of Dementia often also manifest in the retina in the early stage, making retinal imaging a candidate for early diagnosis. In this study, we present a multimodal in vivo imaging and analysis framework to understand the longitudinal progression and correlation between the retinal and cerebral pathology using an AD mouse model.

Methods : Animal: 12 APP/PS1 transgenic mice and 11 wildtype littermates were scanned at 7 timepoints: 5, 7, 9, 11, 14, 16 and 18 months.
Retinal imaging: Multi-modal retinal-imaging, which combines the fluorescent scanning laser ophthalmoscope (fSLO) and optical coherent tomography (OCT), were used to extract the longitudinal progression of neurodegeneration-related retinal morphological and pathological changes. The retinal sublayers were segmented from the OCT images using a gradient-base graph-cut algorithm, to extract the thicknesses of retinal layers. The deposition of amyloid beta (Ab) plaques, an early hallmark of AD, were segmented and quantified as the specks with high fluorescence index from the fSLO images.
Brain imaging: Longitudinal in vivo MRIs were acquired in vivo. The contrast between Ab and other tissues were enhanced by intravenous injecting of Gadolinium and transit opening of brain-blood-barrier using microbubble ultrasound. Brain structures were automatically segmented to extract the cerebral morphological and texture changes such as hippocampal atrophies.

Results : Sample results of the acquired and processed images on the APP/PS1 mouse model were shown in Figure 1. The (A) retina OCT images were segmented into: (B) the inner retinal complex and the outer retinal complex. The (C) fundus image reconstructed from the OCT and (D) the fSLO fundus image showing Ab specks. The (E) Ab contrast-enhanced MRI and (F) The parcellated brain structures.

Conclusions : We have developed an integrated multimodal in vivo retinal- and cerebral- image acquisition, processing and analysis framework. We demonstrated the feasibility of the framework to study longitudinal progression and correlation between the retinal and cerebral neurodegeneration using a mouse model of AD.

This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.