Atlas based image reconstruction for diffuse optical imaging of the human brain

Wu, Xue (2016). Atlas based image reconstruction for diffuse optical imaging of the human brain. University of Birmingham. Ph.D.

Preview

Wu16PhD.pdf
PDF - Accepted Version
Download (14MB)

Abstract

Diffuse Optical Tomography (DOT) has been applied to imaging functional activations in the adult brain. Registered-atlas models are acceptable alternative forward models for the subject-specific models. In this work, different landmark-based rigid registration methods are quantitatively evaluated and compared in geometrical accuracy of the registration result, accuracy of light propagation approximation and recovery accuracy of the brain activations based on the whole head and localized head regions. The most suitable registration methods are selected based on accuracy and efficiency and they vary based on region of interest. For example, the most suitable registration method for recovery of whole cortex activation is the registration method based on 19 landmarks from the EEG 10/20 system and non-iterative Point to Point algorithms (EEG19nP2P registration).
Efficiency of the recovery process is another popular research area in DOT. In this work, a modified generation approach of the light propagation approximation is designed based on a reduced sensitivity matrix and parallelisation process. It improves the storage efficiency by >1000% and time efficiency by ~400%. Based on this approach, the brain activation recovery of DOT can be processed on a normal laptop without large memory requirements within 45 minutes which is more suitable for a portable system.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):