Target depth-regularized reconstruction in diffuse optical tomography using ultrasound segmentation as prior information

Ultrasound (US)-guided diffuse optical tomography (DOT) is a promising non-invasive functional imaging technique for diagnosing breast cancer and monitoring breast cancer treatment response. However, because larger lesions are highly absorbing, reconstructions of these lesions using reflection geometry may exhibit light shadowing, which leads to inaccurate quantification of their deeper portions. Here we propose a depth-regularized reconstruction algorithm combined with a semi-automated interactive neural network (CNN) for depth-dependent reconstruction of absorption distribution. CNN segments co-registered US to extract both spatial and depth priors, and the depth-regularized algorithm incorporates these parameters into the reconstruction. Through simulation and phantom data, the proposed algorithm is shown to significantly improve the depth distribution of reconstructed absorption maps of large targets. Evaluated with 26 patients with larger breast lesions, the algorithm shows 2.4 to 3 times improvement in the top-to-bottom reconstructed homogeneity of the absorption maps for these lesions.