A novel approach to assess reliability in computation of perfusion CT indexes

Dynamic contrast-enhanced CT imaging is widely used in the current clinical practice to provide functional indexes of activity of organs. Perfusion CT has imposed itself as one of the most important techniques to assess the effectiveness of treatments for liver and lung cancer. However, respiratory movements and the calculation of perfusion indexes may introduce artifacts and errors, ultimately affecting the examination reliability. In this work, we present the approach devised to achieve robust perfusion indexes, also proposing our strategy to help selecting reliable values only. Two temporal signal models have been compared as well as the perfusion index values computed accordingly. Numerical analysis and visual evaluation confirm the effectiveness of our approach to detect unreliable values even when they come from robust models.