Διερεύνηση των χαρακτηριστικών πήξης του αίματος χρησιμοποιώντας τεχνικές επεξεργασίας εικόνας

Abstract

This thesis deals with the characterization of the blood coagulation process using image processing techniques for the development of a diagnostic biomedical device. Blood coagulation is a process that is triggered in cases of injury or other diseases, preventing blood loss. In some pathological conditions, the clotting mechanism is activated and clots are formed that impede blood circulation, even causing blood flow to be stopped. Through this study, the structural properties of blood affected by the fluid coagulation process, were characterized by image processing techniques through MATLAB® software. A total of 9 experiments were performed with samples from 2 donors. The specimens were placed on a microscope slide in a 5mm channel with adhesive tape, covered up with another slide. The experimental setup was placed under a microscope and with the use of a camera, each of the 9 samples was recorded for 30 minutes and a total of 1850 images were taken for each sample. The 1850 images of each sample via a loop in the algorithm were processed by specific procedures and the curves of the Ac spline regions were extracted, from which the curve characterization indices, clotting start time CT, maximum amplitude MCA and area under the curve AUCC were then extracted. For further analysis of Ac curves, modelling was performed using a double exponential function to fully characterize them. The Ac curves exported have an exponential function behaviour since they follow an upward path and then the curve stays fixed at the y-axis. The Ac curves of the present study agree with those of the existing TEG and ROTEM devices since they follow the same behaviour and have the same quality characteristics. From the calculation of the indices it appeared that the values for each sample differ from each other and this is due to factors that affect the experimental setup such as illumination, gap between the slides, sample volume, etc. With better control of these characteristics the curves are expected to match much more with each other.