ABSTRACT
Once the basic framework of reconstruction algorithm is established, there are a number of ways to incorporate image enhancement schemes into the forward and inverse solution procedure for improved performance of reconstruction. Our rst try on this started in the summer of 1995. I recall that Keith Paulsen visited some microwave and electrical impedance imaging laboratories in Europe that summer. After his return from the trip, he told me that the researchers in those laboratories were all talking about total variation minimization (TVM) that could signicantly improve the image reconstruction in microwave and electrical impedance imaging. We quickly envisioned that it could also do some good in diffuse optical tomography (DOT), as it would add an interesting constraint as an additional penalty term in the minimization function to strengthen the search for an optimal inverse solution. I spent just a few days deriving the formula and implemented the TVM constrained code. I tried it using both simulated and phantom data, and the results were really promising (see Section 5.2).
