A fast webcam photogrammetric system to support optical imaging of brain activity.
Doctoral thesis, UCL (University College London).
Optical topography (OT) is an emerging neuroimaging technique utilising the tight coupling between neural activity and regional cerebral blood flow to monitor relative regional changes of haemoglobin concentration. OT systems are compact, low cost, easily portable, and relatively tolerant of body movements enabling clinical diagnosis, psychological experiments and even monitoring brain activities during daily living. There is a requirement for such systems to present their output functional data in a brain model based coordinate space in order to map to the signal source with brain anatomy. However OT data are obtained from a network of OT sensing devices (optodes) placed in contact with the head surface and cannot capture structural information of the underlying brain which might otherwise be used for registration. An appropriate registration method, widely used in Electroencephalography (EEG), is the 10-20 system which utilises bony landmarks as common points to co-register locations on the scalp with a brain model to a repeatability of a few millimetres in clinical applications to an MRI set of reference points. Inheriting the low cost and portability of OT, this thesis develops and validates a novel registration approach utilising off-the-shelf webcam technology in combination with photogrammetric bundle adjustment techniques in order to reliably coordinate targets on optodes and bony landmarks within the 10:20 reference frame to an accuracy of better than 1mm. Initial research includes an assessment of the 3D coordination accuracy, precision and stability of a series of low cost webcams in order to prove their suitability for clinical applications. Results demonstrate the capability of a system based on these cameras to reliably coordinate 3D target locations to the order of 0.5mm and better. Difficulties in automated clinical target image extraction due to poor image quality are circumvented through the development of new target image detection methods. Incremental improvements in image quality from successive webcam generations, up to and including the latest HD systems, are shown to increase coordination accuracy by one order of magnitude. The result is a novel webcam photogrammetric system that is able to rapidly and consistently coordinate targets on optodes and bony landmarks to better than 1mm in OT studies and is able to take advantage of the rapid advances being made in consumer webcam technology. The system is proven in pre-clinical studies to evaluate its coordination accuracy and in simulated clinical OT studies with a head-sized phantom conducted in collaboration with Department of Medical Physics and Bioengineering. Clinical OT studies with human subjects, demonstrate the capability of the system to continuously coordinate targets on optodes and scalp and detect differential movement between optodes and scalp which would invalidate a static registration procedure.
|Title:||A fast webcam photogrammetric system to support optical imaging of brain activity|
|Open access status:||An open access version is available from UCL Discovery|
|UCL classification:||UCL > School of BEAMS > Faculty of Engineering Science > Civil, Environmental and Geomatic Engineering|
Archive Staff Only