Statistically principled use of in-line measurements in intensity diffraction tomography

Intensity diffraction tomography (I-DT) reconstruction theory provides a mathematical mapping between two in-line intensity measurements acquired at a given tomographic view angle and Fourier components of the object function. Poles in this mapping will cause certain Fourier components to contain greatly amplified noise levels when applied to noisy measurement data, which can result in noisy and distorted images in practice. We investigate the statistically principled use of multiple in-line intensity measurements in I-DT. Reconstruction methods are developed that exploit the statistical structure of the in-line measurements to minimize the variance of the estimated Fourier components of the object function.