Feature Of The Week 5/12/13: University of Illinois at Urbana-Champaign Reports on Real-Time In Vivo Computed Optical Interferometric Tomography
Real-time, high-resolution tomography over an extended field-of-view in scattering tissues remains an elusive goal for biomedical imaging. The apparent trade-off between the depth-of-field and lateral resolution in optical imaging degrades the image resolution outside the confocal region in OCT. Although a number of approaches have been utilized to overcome these limitations, many of these techniques are plagued by either high computational complexity or complicated optical setups. ISAM is a computational technique that can yield spatially invariant transverse resolution by resampling the Fourier space of the data. However, a full 3D ISAM reconstruction requires extensive manipulation of volumetric datasets, imposing significant constraints on processing hardware and software.
This article demonstrates a much simpler and memory efficient implementation of 3D ISAM by decomposing the 3D ISAM resampling into two simpler 2D sub-problems. This not only dramatically simplifies the implementation in memory limited devices such as GPUs, but also enables the processing of full-3D high resolution tomograms in real-time. In this work, a memory efficient implementation of 3D ISAM is demonstrated by two successive orthogonal 2D ISAM reconstructions by harnessing the parallel processing capabilities of the GPU. High acquisition rates coupled with real-time processing and feedback enabled the acquisition of phase stable data making it possible to demonstrate the first real-time in vivo 3D ISAM tomograms. Real-time feedback was used to strategically place the focus inside the tissue by optimizing the signal collection and reconstruction quality. These computational and processing approaches enabled the visualization of tissue structures in vivo in real-time at high resolution throughout the imaging volume. This has the potential to provide important diagnostic information in time-sensitive clinical applications such as image-guided surgery.
For more information see recent Article. Courtesy of Nathan Shemonski and Adeel Ahmad from University of Illinois at Urbana-Champaign.