“Processing and Analysis of Transcranial Ultrasound Images” Doctoral Thesis

Author, institution: Andrius Sakalauskas, Kaunas University of Technology

Science area, field: Technological Sciences, Electrical and Electronic Engineering

The Doctoral Dissertation is available at the library of Kaunas University of Technology (K. Donelaičio St. 20, Kaunas).

Scientific Supervisor: Prof. Dr. Habil. Arūnas LUKOŠEVIČIUS (Kaunas University of Technology, Technological Sciences, Electrical and Electronic Engineering – 01T)

Dissertation defence board of Electrical and Electronic Engineering Science Field

Prof. Dr. Arminas RAGAUSKAS (Kaunas University of Technology, Technical Sciences,  Electrical and Electronic Engineering 01T) – chairman;
Prof. Dr. Algidas BASEVIČIUS (Lithuanian University of Health Sciences, Biomedical Sciences, Medicine 06B);
Prof. Dr. Algimantas KRIŠČIUKAITIS (Lithuanian University of Health Sciences, Biomedical Sciences, Biophysics 02B);
Prof. Dr. Liudas MAŽEIKA (Kaunas University of Technology, Technical Sciences,  Electrical and Electronic Engineering 01T);
Prof. Dr. Dalius NAVAKAUSKAS (Vilnius Gediminas Technical University, Electrical and Electronic Engineering 01T).

Annotation:

The thesis presents developed automated transcranial sonography (TCS) image analysis system for quantitative Parkinson’s disease (PD) assessment. TCS is relatively new neuroimaging modality proposed for the early diagnostics of PD. The main PD indicator is a hyperechogenicity found in the substantia nigra (SN) region in the middle of the brainstem. The main limitation of TCS image based diagnostics is a certain subjectivity which is caused by a low spatial resolution of the images achievable, resulting into difficulties to evaluate the images manually. The developed system consists of 5 main components: (1) algorithm for the mesencephalon region segmentation based on statistical shape model and local phase congruence, (2) algorithm for the extraction of intra-mesencephalic region, (3) classifier based SN segmentation algorithm, (4) parameters used for echogenicity assessment in the extracted informative regions (1-3), and (5) Mahalanobis distance based classifier. The system tested using collected clinical and image analysis data. In total the images of 341 subjects were analysed. It was demonstrated that proposed novel quantitative image feature set for hyperechogenicity assessment outperforms the previously presented by authors and. The results are comparable to manual TCS approach. The proposed system could be used as a supplementary tool for the automated assessment of the parameters for the decision support in the diagnostics of PD.