Institutional Repository
Technical University of Crete
Technical University of Crete
EN
|
EL
| URI: | http://purl.tuc.gr/dl/dias/E2ACD74E-5BA2-464A-A1A2-313B89C0AD09 | ||
| Year | 2022 | ||
| Type of Item | Diploma Work | ||
| License |
|
||
| Bibliographic Citation | Christos Tsiaousis, "A novel multi modal high throughput screening and material identification imaging system", Diploma Work, School of Electrical and Computer Engineering (ECE), Technical University of Crete, Chania, Greece, 2022 https://doi.org/10.26233/heallink.tuc.92947 | ||
| Appears in Collections |
The curiosity-driven impulses of humanity have directed technologies to broaden biological senses, by enhancing the observation capabilities. Since the beginning of scientific evolution, researchers rely on such technologies to visualize new features of diagnostic importance. Demanding tasks with critical time constraints which portray the evolving society, brought the necessity of automation in research spotlight. Today's High Throughput Screening (HTS) technologies are prone to monolithic design, are deficient of a wide palette of imaging modalities and are fast approaching their identification capacity limits. Here, we present a novel Multi Modal (MM) HTS imaging system that overtakes the modality limits of traditional HTS apparatus, by introducing four different Hyper Spectral (HS) imaging modalities that uniquely fingerprint materials. The morphological and molecular structure of specimens is depicted using the Transmission, Reflection and Fluorescence modalities, while a contemporary Polarization imaging technique is used to depict the crystalline structure. Essential high throughput requirements for the latter are real-time and rotation independent measurements, which are achieved by an innovative Polarization State Generator (PSG) -- Detector (PSD) design. This expands the identification capacity by generalizing data representation into a high dimensional Multi Modal Feature Space (MMFS). In this space there almost definitely exists a differentiation hyperplane that can separate seemingly similar materials. We leverage this property to demonstrate a new potential supervised classification method that relies on the multi modal feature space and indicate some examples that necessitate it.
Copyright © DIAS 2013
