One of the most powerful tools medical diagnostics has today, is radiology. Radiology has gone through major improvements over the past decades, which is partially thanks to the improvements in digital image processing. Through many years only radiological experts could differentiate harmful mutations from background noise or the shadows of bronchial or vessels. Today this process is helped by computer programs. A new and promising procedure is diagnostics based on chest tomosintesis (DTS), which has many advantages compared to CT or MRI. It also has many problems to be solved. One of these problems is that round-shadow searching algorithms have many false positive outcomes. This is because they are unable to make a difference between the shadows of harmful mutations, and vessels. In this paper I try to find a method to map the vessel network of the lungs, thus removing the false positive outcomes.
In this paper I will study different noise reduction techniques, apply them to the DTS images and compare them to each other. Then I will study vessel enhancement methods used with CT images, apply one of them to my images and evaluate the outcome. Finally I propose a method that can connect vessels through multiple generated images that I have generated.