Dementia is associated with a variety of neurovascular and neurometabolic abnormalities. Traditional imaging techniques used to investigate such abnormalities, such as Positron Emission Tomography and functional Magnetic Resonance Imaging, are not always well tolerated, have expensive start up and running costs, and are limited with regards to the types of experiments that can be performed as they can be highly sensitive to movement, are noisy, and have physical restrictions. Near-infrared spectroscopy (NIRS) is a non-invasive neuroimaging technique which uses light in the near-infrared spectrum to detect relative changes in concentration of oxygenated and deoxygenated haemoglobin, and the oxidation state of Cytochrome C Oxidase. As such, NIRS can provide measures of brain oxygenation and metabolism. NIRS is less sensitive to movement, is well tolerated and has few contraindications. It is thus a promising candidate for use in clinics or in peoples' homes for monitoring dementia. In the present study, the investigators aim to use both dual-wavelength and broadband NIRS in a range of dementia subtypes, including Alzheimer's Disease and Dementia with Lewy Bodies, and severities, including Mild Cognitive Impairment, to identify how brain oxygenation and metabolism is altered in dementia and across various clinical subgroups. The investigators also aim to determine the relationship between brain oxygenation and metabolism in dementia, and use machine learning approaches to identify optical biomarkers for dementia.