Master thesis defense by Chrysoula Stathaki

Segmentation of micro-CT scans of biomaterials and steps towards their joint reconstruction and segmentation

The delivery of vaccines through oral administration is becoming popular, due to the low cost, the convenient distribution as well as the unnecessary contact between the health provider and the patient. Toxoids are used as vaccines, but here, we note that after vaccination, an immunological response is formed against the molecular markers of the toxoid without resulting in toxin-induced illness. Such preparation is also known as an anatoxin. In this study, we worked with Diphtheria anatoxin (DT) and Tetanus anatoxin (TT). Due to the harsh environment of the gastrointestinal tract, the immuno- logical response is difficult to be achieved from the anatoxins. A promising adjuvant, i.e toxoid carrier, for the oral delivery of vaccines is the Santa Barbara Amorphous-15 (SBA-15). The key feature of SBA-15 effectiveness is the determination of the anatoxin agglomeration in the carrier, which constitutes the central interest of this work. X-ray micro-CT data were acquired to determine the way that different morphologies of SBA- 15 affect the incorporation. The analysis of the data was performed using Analytical Reconstruction Techniques (ANRT) and Iterative Segmentation Techniques (IST). The thesis is separated into two parts. The first half is centered on a series of implementa- tions for computing solutions to discretized inverse problems. The area of application is the territory of Computed Tomography (CT). Iterative Reconstruction Techniques (IRT) were studied, derived from the Kaczmarz iterative solver. IRT are responsible for the restoration and reconstruction of the obtained signals, and often provide better results compared to ANRT. Segmentation techniques, which are the cornerstone of the analysis of the samples, will be discussed subsequently. The chosen IST were built from spa- tial regularisation and K-means clustering. The second part focuses on IST that were employed to extract information about the agglomeration of the anatoxins inside the hi- erarchical structure of SBA-15.Three different morphologies of SBA-15 were examined. For this case, only ANRT were considered as a means of restoration and reconstruction. The benchmark that shaped this study was the detection of an unexpected inclusion in the samples, resulting from the calcination procedure used during the samples’ prepara- tion. Segmentation techniques drawn from already existing libraries were not adaptive to the samples’ special traits, i.e could not detect and classify successfully this inclusion. Therefore, a segmentation technique was built particularly to deal with this specific chara- cteristic. For the study of both reconstruction and segmentation, proximal mappings were considered and spatial regularization techniques. These methods
enable the visual- ization of the anatoxins, and have implications for the immunological response. The first important remark of the analysis is that the distribution of the anatoxins’ in the samples is not uniform. The anatoxin is found agglomerated in locations with high concentration of Phosphate Buffered Saline (PBS), a buffer solution commonly used in biological research. The quantification of the samples’ revealed a significantly higher 4 percentage of the inclu- sion in one of the three examined carriers, denoted as SBA-15a. The agglomeration ratios are proved to be slightly greater in the DT than in the TT samples. Among the three dif- ferent carriers, SBA-15a, SBA-15b and SBA-15c, the first one presents the highest loading capacity in terms of mass of anatoxin incorporation, but also the highest agglomeration percentages compared to the others. This sample is followed by SBA-15b and SBA-15c.