Virtual reality assisted fluorescence microscopy data visualisation and analysis for improved understanding of molecular structures implicated in neurodegenerative diseases

Theart, Rensu (2020-03)

Thesis (PhD)--Stellenbosch University, 2020.

Thesis

ENGLISH ABSTRACT: Confocal microscopy is one of the major imaging tools used in molecular life sciences. It delivers detailed three-dimensional data sets and is instrumental in biological analysis and research where structures of interest are labelled using fluorescent probes. Usually, this three-dimensional data is rendered as a projection onto a two-dimensional display. This can however lead to ambiguity in the visual interpretation of the structures of interest in the sample. Furthermore, analysis and region of interest (ROI) selection are also most commonly performed two-dimensionally. This may inadvertently lead to either the exclusion of relevant or the inclusion of irrelevant data points, consequently affecting the accuracy of the analysis. We present a virtual reality (VR) based system that allows firstly, precision region of interest selection and colocalisation analysis, secondly, the spatial visualisation of the correlation of colocalised fluorescence channels and, thirdly, an analysis tool to automatically determine the localisation and presence of mitochondrial fission, fusion and depolarisation. The VR system allows the three-dimensional reconstructed sample data set to be interrogated and analysed in a highly controlled and precise manner, using either fullyimmersive hand-tracking or a conventional handheld controller. We apply this system to the specific task of colocalisation analysis, an important tool in fluorescence microscopy. We evaluate our system interface by means of a set of user trials and show that, despite inaccuracies associated with the hand tracking, it is the most productive and intuitive interface compared to the handheld controller. Applying the VR system to biological sample analysis, we subsequently calculate several key colocalisation metrics using both two-dimensionally and three-dimensionally derived super-resolved structured illumination-based data sets. Using a neuronal injury model, we investigate the change in colocalisation between two proteins of interest, Tau and acetylated -tubulin, under control conditions as well as after 6 hours and again after 24 hours of neuronal injury. Applying the VR based system, we demonstrate the ability to perform precise ROI selections of 3D structures for subsequent colocalisation analysis. We demonstrate that performing colocalisation analysis in three dimensions enhances its sensitivity, leading to a greater number of statistically significant differences than could be established when using two-dimensionally based methods. Next, we propose a novel biological visual analysis method for the qualitative analysis of colocalisation. This method visualises the spatial distribution of the correlation between the underlying fluorescence channel intensities by using a colourmap. This method is evaluated using both synthetic data and biological fluorescence micrographs, demonstrating enhancement of the visualisation in a robust manner by indicating only truly colocalised regions. Mitochondrial fission, fusion and depolarisation events are important in cellular function and viability. However, the quantitative analysis linked to the localisation of each event in the three-dimensional context has not been accomplished. We extend the VR system to analyse fluorescence-based time-lapse sequences of mitochondria and propose a new method to automatically determine the location and quantity of the mitochondrial events. The detected mitochondrial event locations can then be superimposed on the fluorescence z-stacks. We apply this method both to control samples as well as cells that were treated with hydroxychloroquine sulphate (HCQ) and demonstrate how a subsequent quantitative description of the fission/fusion equilibrium as well as the extent of depolarisation can be determined. We conclude that virtual reality offers an attractive and powerful means to extend fluorescence-based microscopy sample navigation, visualisation and analysis. Three-dimensional VR-assisted ROI selection enable samples to be interrogated and assessed with greater precision, thereby exploiting the potential of fluorescence-based image analysis, such as colocalisation, in biomedical research. The automatic localisation and quantification of mitochondrial events can support research of mitochondrial function in healthy and diseased cells, where quantitative analysis of fission, fusion and depolarisation is of importance.

AFRIKAANSE OPSOMMING: Konfokale mikroskopie is een van die belangrikste beeldinstrumente wat in die molekulêre lewenswetenskappe gebruik word. Dit lewer gedetailleerde drie-dimensionele datastelle wat instrumenteel is vir biologiese analise en navorsing waar strukture van belang gemerk word met behulp van fluoresserende probes. Gewoonlik word hierdie drie-dimensionele data as ’n projeksie op ’n twee-dimensionele skerm weergegee. Dit kan egter lei tot dubbelsinnigheid in die visuele interpretasie van die struktuur van belang in die biologiese monster. Verder word analise en streek van belang (SVB) seleksie ook meestal tweedimensioneel uitgevoer. Dit kan onbedoeld lei tot die uitsluiting van relevante of die insluiting van irrelevante datapunte, wat die akkuraatheid van die analise kan beïnvloed. Ons bied ’n virtuele realiteit (VR)-gebaseerde stelsel aan wat eerstens presisie SVB seleksie en kolokaliseringsanalise in staat stel, tweedens, die ruimtelike visualisering weergee van die korrelasie van gekolokaliseerde fluoressensie-kanale, en derdens ’n analise-metode vir die outomaties lokalisering van mitochondriale splitsing, samesmelting en depolarisasie. Die VR-stelsel laat toe dat die drie-dimensionele gerekonstrueerde monsterdatastel op ’n hoogs gekontroleerde en akkurate wyse ondersoek en geanaliseer kan word deur gebruik te maak van ’n volledig immersiewe hand-volg toestel óf ’n konvensionele spelbeheerder. Ons pas hierdie stelsel toe op die spesifieke taak van kolokaliseringsanalise, ’n belangrike instrument in fluoressensie-mikroskopie. Ons evalueer die twee stelsel-koppelvlakke aan die hand van ’n stel gebruikersproewe en wys dat die hand-volg toesetel, ondanks onakkuraathede wat met die dit verband hou, die mees produktiewe en intuïtiewe koppelvlak is in vergelyking met die spelbeheerder. Met die toepassing van die VR-stelsel op biologiese monsteranalise, bereken ons daarna verskeie sleutel kolokalisasiemates met behulp van beide twee-dimensionele en drie-dimensionele “super-resolved structured illumination” gebaseerde datastelle. Met behulp van ’n neuronale beseringsmodel ondersoek ons die verandering in kolokalisasie tussen twee proteïene van belang, Tau en geasetileerde -tubulien, onder beheerstoestande sowel as na 6 uur en weer na 24 uur na neuronale besering. Met gebruik van die VR-gebaseerde stelsel, demonstreer ons die vermoë om presiese SVB-seleksies van 3D-strukture uit te voer vir latere kolokaliseringsanalise. Ons demonstreer dat die uitvoering van kolokaliseringsanalise in drie dimensies die sensitiwiteit daarvan verhoog, wat lei tot ’n groter aantal statisties beduidende verskille as wat vasgestel kan word by die gebruik van twee-dimensionele metodes. Vervolgens stel ons ’n nuwe biologiese visuele analise-metode voor vir die kwalitatiewe analise van kolokalisering. Hierdie metode visualiseer die ruimtelike verdeling van die korrelasie tussen die onderliggende fluoressensie-kanaal-intensiteite met behulp van ’n kleurkaart. Hierdie metode word geëvalueer deur gebruik te maak van sintetiese data sowel as biologiese fluoressensiemikrograwe, wat die verbetering van die visualisering op ’n robuuste manier demonstreer deur slegs waarlik gekolokaliseerde streke aan te dui. Mitochondriale splitsing, samesmelting en depolarisasie gebeurtenisse is belangrik vir sellulêre funksie en lewensvatbaarheid. Die kwantitatiewe ontleding gekoppel aan die lokalisering van elke gebeurtenis in die drie-dimensionele konteks is egter nog nie gedoen nie. Ons brei die VR-stelsel uit om fluoressensie-gebaseerde tydsverloop-sekwensies van mitochondria te ontleed en stel ’n nuwe metode voor om outomaties die ligging en hoeveelheid van die mitochondriale gebeure te bepaal. Die waargenome liggings van mitochondriale gebeurtenisse kan dan op die fluoressensie-z-stapels aangebring word. Ons pas hierdie metode toe op beide kontrole monsters sowel as selle wat met hydroxychloroquine sulfaat (HCQ) behandel is en demonstreer hoe ’n daaropvolgende kwantitatiewe beskrywing van die splitsing/samesmelting-ewewig sowel as die omvang van depolarisasie bepaal kan word. Ons kom tot die gevolgtrekking dat virtuele werklikheid ’n aantreklike en kragtige manier bied om fluoressensie-gebaseerde mikroskopie-monsternavigasie, -visualisering en -analise uit te voer. Drie-dimensionele VR-ondersteunde SVB-seleksie laat toe dat monsters met groter noukeurigheid ondersoek en beoordeel kan word en sodoende die potensiaal van fluoressensie-gebaseerde beeldanalise, soos kolokalisering, in biomediese navorsing te benut. Die outomatiese lokalisering en kwantifisering van mitochondriale gebeure kan die navorsing van mitochondriale funksie in gesonde en siek selle ondersteun, waar kwantitatiewe analise van splitsing, samesmelting en depolarisasie van belang is.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/107804
This item appears in the following collections: