Time-resolved optical tomography of functional activation in the adult brain using a supercontinuum laser source

Diffuse optical imaging (DOI) is finding widespread application in the study of human brain activation. However, the process of wavelength selection is unclear and may lead to choices that negatively affect the accuracy and reliability of the data. This thesis describes a method to empirically determine the combination of four nearinfrared (NIR) wavelengths that yields the best images of functional activity in the brain. An experiment was designed to quantify changes of chromophores oxyhaemoglobin and deoxyhaemoglobin in the motor cortex of a healthy adult subject using UCL’s second generation optical imaging device, MONSTIR. This devices allows us to use four wavelengths simultaneously. A set of twelve was tested and recombined into 495 combinations of four wavelengths. This data was analysed with three methods that ranked the four-wavelength combinations according to different conditions. Method A looked for the largest change resulting from a modified Beer-Lambert law algorithm. Method B favoured the combinations whose output of the same algorithm was closest to a theoretical model. Method C compared the 495 four-wavelength images to a twelve-wavelength image (a gold-standard) and ranked the wavelength combinations by looking for the minimum root-mean-square error between each pair of images. We concluded that the most adequate method is method C because it provides with the best images. The best set of wavelengths obtained was 690, 790, 810 and 840 nm. These results are specific to the imaging device used, however our general method can be applied to other optical imaging systems.