Dementia is a devastating age-related disease with important public health challenges. Worldwide cost of dementia exceeds US$ 800 billion. Every 3 seconds a new case of dementia is diagnosed and the number of patients will almost double to 82 million by 2030. Up to 75% of patients with dementia have Alzheimer’s disease (AD). Accumulation of amyloid beta (Aβ) is an essential marker for clinical AD diagnosis. Aβ concentrations can be determined using brain Positron Emission Tomography (PET) and cerebrospinal fluid (CSF) sampling . However, these methods suffer from large infrastructure costs, limited resolution and patient risk. Today, AD diagnosis is made too late and individuals are not sufficiently stratified in early stages of the disease. This is an important reason why clinical trials in search of new AD drugs have failed so far.

The need for better and more convenient AD identification and monitoring is urgent. Non-invasive hyperspectral imaging of the back of the eye, the retina, is a possible solution. Anatomically and developmentally, the retina is an extension of the brain. Unlike the rest of the central nervous system, the retina can be visualised directly at very high resolution. In this respect, pilot data show that aggregates of retinal Aβ can be seen based on a hyperspectral signature. Unfortunately, no data exist about the possible translation of these findings to humans. In addition, applicability of this type of imaging in humans is limited because no practical and affordable hyperspectral imaging system exists.

HERALD wants to force a breakthrough in the sensitive and convenient detection of retinal Aβ. The application of compact and fast hyperspectral sensors is one technology pillar. The second technology pillar is advanced image analysis for precise Aβ detection. The project has 3 Work Packages. Work Package 1 is the realisation of hyperspectral imaging systems with microscope for ex vivo Aβ detection and with fundus camera for in vivo imaging of patients. Work Package 2 collects ex vivo images from brain slices of post-mortem AD patients and from retinas of elderly patients who underwent enucleation or evisceration for medical reasons. Additionally, brain and retina images will be obtained from mouse models. This approach makes a systematic cross-species and cross-tissue comparison possible. Work Package 3 will deliver a proof-of-concept of in vivo retinal Aβ detection using a fundus camera in combination with a hyperspectral camera.

HERALD can pave the way for more convenient quantification of Aβ in three domains. Hyperspectral imaging of retinal Aβ can be supportive in preclinical AD research (e.g. detection and follow-up of disease development in mouse model). Retinal analysis can help in better selection and stratification of patients in clinical trials. In a much later stage, it can spur innovation in clinical practice through better prediction and monitoring of AD high-risk patients and population-wide screening.