Algorithmic techniques for manual-scanning and multi-sensory representation of optical coherence tomography data

Optical coherence tomography (OCT) is an emerging biomedical imaging modality that interferometrically measures the depth-resolved back-scattered light from a sample. The high resolution subsurface imaging capabilities of OCT can potentially provide valuable diagnostic feedback about tissue morphology during intra-operative applications of OCT. However, the intra-operative use of OCT is hindered by the lack of suitable sample scanning mechanisms and the difficulty in real-time interpretation of OCT data. Screening or diagnostic procedures typically require high resolution imaging over a large field-of-view. The current scanning mechanisms, which are predominantly based on mechanically scanning the OCT beam over the specimen, have a limited scan range and lack the required flexibility for large field-of-view imaging. Moreover, OCT, being a high resolution imaging modality, requires a very high data acquisition rate and would generate huge amounts of data if used to image over a large surgical field. These extremely high data rates would make the real-time interpretation of OCT data a challenge, especially in the highly demanding operating room environment where the surgeon has to simultaneously keep track of a number of parameters. In this thesis, algorithmic techniques are proposed that may be helpful in overcoming some of these limitations. Image acquisition over a large field-of-view with flexible scan geometry would require a manually scanned hand-held probe. This thesis describes a cross-correlation based image acquisition technique that may be used for image formation by compensating for the variable scan velocity encountered while manually scanning the probe. This thesis also describes an approach for multi-sensory representation of OCT data by converting OCT data and images into sound. Audio rendering of OCT data when used in conjunction with the visual display may facilitate rapid interpretation of OCT data as the human auditory sense can detect more rapid transitions in the data than the visual sense and hence may be used for identifying suspected regions in real-time, which can subsequently be monitored in high resolution using the visual display