master's thesis
Marina Kovač (Author), Ciril Bohak (Mentor), David Murphy (Co-mentor)

Abstract

In this work, we were dealing with how to generate new models of cellular structures from real volumetric images. From the database of volumetric images with labeled cellular structures, we extracted two kinds of cellular structures (endolysosomes and fusiform vesicles), and prepared new volumetric images with only one object in each to later find points on their surfaces. With those points, we created a shape model used in the generation of new cellular structures. We defined a process to generate new models of cellular structures based on the three input parameters. Output is the same if none of those three parameters change. Thus, the user has the ability to generate countless different models. After we created a few new models with the proposed generator, we evaluated those new models with two approaches. First, calculating the root mean square error (RMSE) and then calculating and comparing norms of average vectors and vectors for standard deviations. A quick evaluation with an eye may reveal some of the generated models are nicer than existing ones. Nevertheless, new models could be even better if we would also modify the sizes of the learning and testing objects.

Keywords

cellular structures;endolysosome;fusiform vesicle;visualization;computer graphics;object generation;multimedia;computer science;master's degree;

Data

Language: English
Year of publishing:
Typology: 2.09 - Master's Thesis
Organization: UL FE - Faculty of Electrical Engineering
Publisher: [M. Kovač]
UDC: 004.92(043.2)
COBISS: 83640323 Link will open in a new window
Views: 223
Downloads: 25
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: Slovenian
Secondary title: Vizualizacija proceduralno generiranih celičnih struktur
Secondary abstract: V tem magistrskem delu smo se ukvarjali z nalogo kako generirati nove modele celičnih struktur na podlagi realnih volumetričnih slik. Iz baze z volumetričnimi slikami z označenimi celičnimi strukturami smo pridobili dve vrsti celičnih struktur (endolizosome in fuziformne vezikle), ter jih zapisali v nove volumetrične slike, kjer vsaka slika vsebuje samo en objekt. Te nove slike smo uporabili pri izračunavanju površinskih točk pridobljenih celičnih struktur. Z njimi smo naredili model oblike, ki smo ga uporabili v procesu generiranja novih objektov. Zastavili smo proces, kako generirati nove modele celičnih struktur na podlagi treh vhodnih parametrov. Končni model bo enak, če se vhodni parametri ne spremenijo. Uporabnik ima možnost generiranja neskončno različnih končnih modelov. Predlagan postopek smo ovrednotili z uporabo dveh metrik, in sicer z izračunom povprečnih korenjenih srednjih kvadratnih napak, ter z izračunom in primerjavo norm povprečnih vektorjev in vektorjev standardnih odklonov. Pogled na nove generirane modele lahko da občutek, da so nekateri lepši od objektov iz testne množice, vendar se nam zdi, da bi bili še boljši, če bi v predlaganem postopku poizkusili poenotiti tudi velikosti objektov.
Secondary keywords: celične strukture;endolizosomi;fusiformni vezikli;vizualizacija;generiranje objektov;multimedija;magisteriji;Računalniška grafika;Računalništvo;Univerzitetna in visokošolska dela;
Type (COBISS): Master's thesis/paper
Study programme: 1001017
Thesis comment: Univ. v Ljubljani, Fak. za računalništvo in informatiko
Pages: VIII, 79 str.
ID: 13763214