Detecting abnormalities in chest tomosynthesis images

OData support
Dr. Horváth Gábor
Department of Measurement and Information Systems

Digital chest tomosynthesis (C-DTS) reconstructs coronal slices of chest from a set of X-ray projections acquired over different geometrical arrangement of the X-ray source and the flat panel detector. This type of chest imaging is an active research subject, the C-DTS devices are not as refined as the P-A chest X-rays and the chest CT devices. Due to the fact that the planned main application of the C-DTS is detecting early stage tumors in population lung cancer screening tests, engineering and implementing computer aided diagnostic decision support system (CAD) is important. According to recent publications, pulmonologists are able to detect 51% of the early stage lung tumors in C-DTS reconstructions, while after interpreting the CT scans of the examined patients, 91% of these tumors was localized in C-DTS scans. The main goal of the computer aided diagnostic decision support system (CAD) presented in this diploma work is to increase the sensitivity of detection early stage tumors, without increasing the number of false detections.

Based on my recent knowledge, there are not any available CAD systems for C-DTS scans, and the Department-independent scientific literature of this topic contains only one patent. Thus it can be stated that the topic of my diploma work is important, but there aren’t any working or publicized solution, therefore engineering such a C-DTS CAD can be considered as a research challenge.

In the diploma work I engineered, implemented and analyzed a C-DTS CAD system in MATLAB programming environment. 4 real C-DTS scans were available in order to construct and validate the CAD system. 3 scans were acquired from real patient, and one scan was acquired from anthropomorphic chest phantom, which contained physical simulated tumors. Due to the number of the available C-DTS scans, the results of the validation of the implemented CAD system should be treated with cautiously.


Please sign in to download the files of this thesis.