Unsupervised analysis of tumour vascular parameters in photoacoustic tomography

Researchers: John Aston, Sarah Bohndiek, Shahin Tavakoli (now at Warwick University), Michal Tomaszewski,

Photoacoustic tomography is an emerging medical imaging modality that combines the high contrast of optical imaging with the high spatial resolution of ultrasound. By acquiring acoustic signals excited using optical energy across a range of wavelengths, it is possible to resolve different chromophores in tissue, including those that are endogenous (e.g. haemoglobin) and those introduced into the body by intravenous injection (e.g. small molecule dyes). A significant challenge that arises in photoacoustic tomography is the ‘spectral colouring’ (equivalent to X-ray beam hardening) that occurs as light propagates into this tissue. We have recently been exploring the potential of unsupervised classification methods such as principal components analysis to extract biologically meaningful information from data sets that are both spatially and temporally resolved, avoiding the need for analytical compensation for spectral colouring. This approach has revealed interesting correlations between tumour perfusion measured using a label-free oxygen gas challenge, compared to the introduction of a vascular perfusion contrast agent. We have also begun to apply this methodology to study contrast at longer wavelengths, where absorption of water and lipids begin to dominate the spectral features.

Who's involved