Modelling Uncertainty of Electrical Properties Tomography

BELE, SIMON (author), Sadikov, Aleksander (mentor)

This work uses estimation of error as a way to estimate per voxel data certainty, through which, to the best of our knowledge, we achieve the first known localisation of pathologies in noisy Helmholtz-based EPT reconstructions of brain tissue conductivity. We estimate data certainty by comparing the same voxels across two different conductivity reconstructions (H-EPT and CR-EPT) as well through estimating boundary error (as the simplified Helmholtz equation used in our data assumed piece-wise constant conductivity) and its propagation through the computation of the spatial derivative required to evaluate the Helmholtz equation.

• Language: English
• Keywords: EPT, electrical conductivity, H-EPT, CR-EPT, localisation of pathologies, modelling error, noise.
• Work type: Bachelor thesis/paper
• Typology: 2.11 - Undergraduate Thesis
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2022
• PID: 20.500.12556/RUL-135402
• COBISS.SI-ID: 100691715
• Publication date in RUL: 11.03.2022

Secondary language

• Language: Slovenian
• Title: Modeliranje negotovosti procesa tomografije električnih lastnosti

Abstract

Diplomska naloga uporabi pristope modeliranja gotovosti podatkov preko katerih doseže prvo nam znano instanco lokalizacije patologij v šumnih Helmholtz-osnovanih EPT rekonstrukcijah prevodnosti možganskega tkiva. Gotovost podatkov ocenjujemo s primerjavo istih vokslov v dveh različnih rekonstrukcijah električne prevodnosti (H-EPT in CR-EPT) ter z oceno mejne napake (ker poenostavljena Helmholtzova enačba, uporabljena v naših podatkih, predpostavlja kosovno konstantno prevodnost) in njenega širjenja z izračunom drugega odvoda, ki je potreben za evalvacijo Helmholtzove enačbe.

• Keywords: EPT, prevodnost, H-EPT, CR-EPT, lokalizacija patologij, modeliranje napake, šum.