Photo-acoustic tomography is a hybrid imaging modality that combines the advantages of optical as well as ultrasound imaging techniques to produce images with high resolution and good contrast at high penetration depths. Choice of reconstruction algorithm as well as experimental and computational parameters plays a major role in governing the accuracy of a tomographic technique. Therefore error estimates with the variation of these parameters have extreme importance. Due to the finite support, that photo-acoustic source has, the pressure signals are not band-limited, but in practice, our detection system is. Hence the reconstructed image from ideal, noiseless band-limited forward data (for future references we will call this band-limited reconstruction) is the best approximation that we have for the unknown object. In the present study, we report the error that arises in the universal back-projection (UBP) based photo-acoustic reconstruction for planer detection geometry due to sampling and filtering of forward data (pressure signals).Computational validation of the error estimates have been carried out for synthetic phantoms. Validation with noisy forward data has also been carried out, to study the effect of noise on the error estimates derived in our work. Although here we have derived the estimates for planar detection geometry, the derivations for spherical and cylindrical geometries follow accordingly.
Under the assumption of high scattering and weak absorbing media, diffusion approximation holds in the radiative transport equation to model propagation of light. Diffusion approximation is valid deep inside the medium, not near the boundary. So, we need to implement accurate boundary conditions. Diffuse reflectance close to the source, majorly, depends on the source model inside the medium and boundary conditions used to derive the analytical solution. We have implemented partial current boundary condition and extrapolated boundary condition with extended isotropic source (exponentially decaying) model. Our model predicts diffuse reflectance close to the source at distance less than one mean free path is more accurate than the other methods. Monte-carlo simulation is the standard model to provide diffuse reflectance close to source most accurately. In this report, partial current, extrapolated boundary condition and a unified boundary condition have been compared for accuracy at different regions from the source. It is found that different boundary conditions work in different regimes and the relative error is less with extended source compared to point source.
Mueller matrix has a vast application regarding information of any scattering (turbid) media such as fog, sea water, and biological tissues. It can extract information from scattering properties of the medium. Recently, information from Mueller images and their interpretation are being used for diagnostic purposes in biological tissues. Polar decomposition of Mueller matrices for scattering medium have also been developed, which could be a very powerful and sensitive tool for mapping the morphology of human tissue sections. On the basis of such decomposition, we report here the variation of diattenuation, depolarization and retardance from normal to dysplasia state in cervix tissue.