Multidimensional OCT Imaging Enabled by Compressed Sensing

Summary Recent advances in the temporal speed and resolution of optical coherence tomography (OCT) imaging systems have opened the door for new possibilities in time-lapse imaging of dynamic biological processes and imaging samples with large surface area with micron scale resolution. Both spectral domain OCT (SD-OCT) and swept source OCT (SS-OCT) systems frequently boast a-line rates well above 100 kHz and as high as 400 kHz. The consequence of higher scan-rates is reduced motion artifacts and also higher data through-put. Modern endoscopic OCT system are proposing scan-rates that result in data rates upwards of 1GB/s. This approaches the data transfer rate and write speeds of modern PCs. Compressing this data at acquisition would have a massive impact on these specific challenges. The kinds of tasks being tackled in OCT imaging increasingly involve working with high-resolution datasets which could be multi-dimensional, where dimensions include space (x,y,z), time, and contrast (polarization, elastography, phase sensitive, doppler). Compressed Sensing (CS) provides tools for dealing with these massive datasets by sampling data far below the Nyquist limit to reduce data through-put. Because of the potential of compressed sensing to reduce the required amount of data to extract the necessary information within an image, compressed sensing has revolutionized magnetic resonance imaging (MRI), accelerating data acquisition which results in shorted scan times and thus cost savings. There have been initial demonstrations on the use of compressed sensing in OCT. It is our goal to develop a comprehensive approach for CS-OCT that will allow for imaging samples with a variety of architecture. Recent advances in OCT hardware have resulted in better image quality. However, combined with compressed sensing, we believe the full power of OCT can be utilized within applications that require large field of view imaging or imaging over long periods of time to assess therapeutic effects, dynamics, and perturbations to organ systems. Our goal within this project is to utilize predictive coding to showcase that compressed sensing can be utilized within cardiac samples that have diversity of tissue types, while yielding >75% compression rate. Within this small grant R03 proposal, we aim to develop, test, and demonstrate a new approach for compressed sensing OCT that can be utilized within images with cardiac imaging datasets. Our two aims include: 1) Compressed sensing to enable large field of view volumetric 3D OCT imaging and 2) Impact of compressed sensing on extracting tissue features and architecture.