Removal of anatomical shadows appearing in X-ray pictures of chest reconstructed from CT-pictures

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

In the ever-improving area of computer-assisted diagnostics algorithms, the problem of detecting irregularities on projected images is quite a significant one. The development of these algorithms requires processes that are able to output thoracic X-ray images, which are devoid of the shadows cast by bones and organs. The purpose of this is making the analysis of critical areas, such as the lung interior, easier. In the following paper I introduce methods for removing bones and selecting the lung interior on CT images, as well as evaluating the results of methods which perform the same bone-removal on PA images. There are also suggestions for the detection of the vessel structure of the lung.

On the subject of bone removal, there is a description of the technology known as dual energy radiography. My solution shows a method which uses CT images as input, and using image processing methods outputs simulated PA images which are 'de-boned'. This method was tested on 6 different sets of thoracic CT images, the output quality was satisfactory.

In another chapter I describe a method for creating PA images containing only the lung interior. The solution of this problem also contains the manipulation of CT images.

There is also a section devoted to evaluating the output of PA bone-removal algorithms. For this purpose, the previously created boneless images are used as reference. The section is based on the results of the single available set of output images.

I also provide an outlook on the interpretation and analysis of the lung interior, primarily the vessel structure. A research of the current literature is included, as well as my own thoughts on the subject.


Please sign in to download the files of this thesis.