[Legacy Report] Data processing in Spectral Optical Coherence Tomography on graphics processing units
Optical Coherence Tomography (OCT) is a method of non-contact and non-invasive imaging of the internal structure of the objects semi-transparent to the infrared light. This method originates from medicine where it has been successfully used as a diagnostic tool in ophthalmology since 1991. OCT has been also used in conservation science for about ten years.
The application of Fourier domain methods led to a significant increase in the data acquisition speed. However, relatively complicated data processing is necessary to generate OCT tomograms in this case and thus the data visualization and analysis usually take more time than the data acquisition. This is a considerable disadvantage, especially in applications in which an immediate evaluation of results is crucial. Recent progress in graphic cards technology gives a promising solution to this problem – the newest graphics processing units (GPU) allow not only for high speed three dimensional (3D) rendering, but also for very fast general purpose numerical calculations with efficiency higher than provided by the CPU.
In this presentation the application of general-purpose computing on graphics processing units (GPGPU) for optical coherence tomography data processing and imaging will be presented. Utilisation of GPU and reformulation of algorithms towards higher efficiency of data processing allowed for real-time imaging of two dimensional tomograms with a best available quality: our software for the OCT data processing is capable of visualization of 2D data (2000 A-scans, 2048 pixels per spectrum) with an image refresh rate higher than 120 Hz.
Data processing on GPU allows also for 3D imaging in real-time which has not been possible so far: the processing and rendering of the 3D (volume) data comprising 100 2D tomograms built of 100 A-scans each is performed at a rate of about 11 volumes per second.
In recent years, several extensions to the OCT method have been successfully applied. One of them is Doppler OCT which allows, among others, for examination of blood flow velocity in capillary vessels. The other variant is used for reduction of speckle noise in the resulting cross sectional image. Both of them are dependent on the scanning protocol parameters which have to be adjusted to the examined object properties and examination conditions. Until now, the parameters of the scanning protocol have to be regulated by the user. But if the OCT data processing is fast enough for real-time analysis of the image (as it is possible with use of the GPGPU), the process of adaptation of the scanning protocol may be done automatically. In this report, the algorithms for automatic adaptation of the scanning protocols in the Doppler OCT and for the speckle noise reduction method will be presented.
Date and Time
Location
Hosts
Registration
Add Event to Calendar
Speakers
Nicolaus Copernicus University Torun
OCT Medical Systems
Biography:
Email:
Address:Poznan, Wielkopolskie, Poland