Dispozitive electronice mobile și software integrat pentru aplicații biomedicale / Mobile electronic devices and integrated software for biomedical applications

This thesis details the results of the author's contributions in several European and National Research projects in which he participated, all projects falling in the biomedical field, and being compiled under the title of the work. Considering the indicated direction, in the present work we present several methods of identification of specific biomarkers, methods and implementations of image and signal processing algorithms, integration of those algorithms into mobile or modern desktop applications, 3D printing machines, and devices developed for certain purposes, such as handling biological cells material and complementary hardware integrate in point-of-care type of devices. The purpose has been to acquire and improve the obtained signals, to perform measurements and implements optimized and adapted AI classification algorithms for rapid diagnosis on-site with high sensitivity and specificity. The present work demonstrates the application of state-of-the-art methods for image and signal processing, together with specific optimizations, in completely new use cases targeting ambitious biomedical applications for rapid diagnosis of bacterial meningitis and tropical diseases, ocular reconstruction and cell manipulation with microgrippers. As such, a series of practical prototypes involving hardware and software developments have been included in the present work. The author developed software that integrates methods available in the literature and managed to achieve the targets set in each specific project, and did specific optimizations to select the right algorithms and parametrization combination, measuring sensitivity, specificity, accuracy and other properties of interest for the proposed POCs. All in all, the results were compared with those obtained by other authors or partners in the projects' consortiums, highlighting results at least as good or even better for the metrics of interes.