The acute otitis media and otitis media with effusion are one of the most common causes why patients have to visit their doctors during the childhood. Due to the middle ear’s closed structure it is difficult to collect data about the various physiological processes in the middle ear, therefore, there is a need for mathematical models that can simulate the processes taking place in the middle ear, which is the basis of the development of appropriate diagnostic and therapeutic methods. To create patient specific models the volume, the surface and the volume/surface ratio of the middle ear have to be identified.
The topic of my thesis was to develop and implement a semi-automated method (and get the necessary knowledge to it), which can identify the volume and surface of the middle ear.
During my work I got to know the CT imaging and several image processing algorithms. To make easier the automated image processing I suggested to identify so-called significant points of the middle ear, which can be used to determine the middle ear area. Based on these significant points and using my programming knowledge I worked out a method and implemented a program, which can determine the volume, surface and surface/volume ratio of the middle ear. In my thesis I expounded all the image processing algorithms I used step by step: image filtering, making masks, thresholding and the method of the volume and surface calculations.
The developed application has two working modes: One for the doctors, where they can determine the necessary calculations with minimal user intervention. The other mode is for experts, where they can develop algorithms, test and compare results, process images and record all these steps in a log file.
The CT pictures I used to test and validate my algorithms are provided by Dr. Zsuzsanna Csákányi, Pál Heim Children's Hospital. I compared the results of my middle ear identification method to the results processed manually by Dr. Zsuzsanna Csákányi.