MULTISPECTRAL QUANTITATIVE IMAGE RECONSTRUCTION METHODS FOR PHOTOACOUSTIC MOLECULAR IMAGING

Following the sequencing of the mouse genome, and the ability to genetically engineer small animal models for the purpose of studying disease processes, there is enormous interest in the life sciences, pre-clinical medicine and the pharmaceutical industry in developing new imaging tools that can characterise these models with high spatial resolution at a cellular or molecular level. Photoacoustic imaging, a new technique based upon the laser excitation of ultrasound, is rapidly becoming recognised as one of a new generation of optical molecular imaging modalities that is set to make a major impact on this field. Its promise is underscored by two factors: firstly, its demonstrated ability to provide high resolution anatomical images based on optical absorption in small animals such as mice and secondly, the availability of a wide range of targeted contrast agents developed for other optical molecular imaging modalities that can potentially be used to provide information at a cellular or molecular level. Termed molecular imaging, the latter can be achieved by introducing an optical absorbing contrast agent that selectively binds to a cellular or molecular receptor known to be associated with a specific disease process - eg tumour growth. By obtaining photoacoustic measurements at multiple wavelengths and, knowing the spectral signature of the contrast agent, it should be possible to identify where, and in what quantity, the contrast agent has accumulated. In this way, the location and expression levels of specific genes or proteins known to be involved in the disease process can be imaged thus providing an insight into the underlying biological processes. To achieve this, a major advance in the way photoacoustic images are reconstructed is required. Current methods provide an image of the internally absorbed optical energy distribution. It is commonly assumed that by obtaining a set of these images at different wavelengths, and matching the spectral characteristics of the contrast agent to those of the absorbed energy at each image pixel, it will be possible to detect and quantify the accumulation of the contrast agent. However, this hypothesis is highly questionable on account of the spatial-spectroscopic crosstalk in an absorbed energy map - in essence the significant spectral characteristics of tissue constituents such as haemoglobin corrupts those of the contrast agent compromising the ability to detect and quantify its presence. The aim of the proposed research is to overcome this by taking the image reconstruction process a stage further than has been previously attempted. This involves accounting for the light transport in the tissue and the known spectral characteristics of the contrast agent and tissue absorbers and scatterers. To achieve this, it is proposed to employ a mathematical model that describes the absorbed optical energy distribution, and its wavelength dependence, within the tissue. By repeatedly adjusting the input parameters of the model until its output matches that of the conventionally reconstructed absorbed energy images, a map of the absolute concentration of the contrast agent can be obtained. The project will require the development of novel computational methods to solve what is an inverse problem of significant scale, analysis of the accuracy and resolution with which the contrast agent concentration can be determined and evaluation using simulated and experimentally obtained phantom and in vivo data / the latter being obtained using a 3D photoacoustic small animal scanner which we are currently developing under a recently awarded EPSRC project. This related project, along with the proposed research and our existing expertise in photoacoustic imaging, modelling light transport and inverse problems provides a timely opportunity to make a significant contribution to the development of photoacoustic imaging as a sensitive, specific and quantitative molecular imaging tool.