High-speed, Quantitative Tissue Spectral Imaging in the Spatial Frequency Domain
- Author(s): Nadeau, Kyle Patrick
- Advisor(s): Tromberg, Bruce J
- et al.
The goal of this work is to enable high-speed, video-rate (>30 Hz) spatial frequency domain imaging (SFDI). SFDI is a non-contact, quantitative spectral imaging approach for mapping in vivo tissue optical properties, allowing for the determination of chromophore concentrations and structural parameters. Conventional SFDI employs spatially-modulated sinusoidal projection patterns of light to decouple absorption from scattering in the spatial frequency domain. In particular, three projected patterns per wavelength per spatial frequency are used for each timepoint. Typical SFDI imaging rates are currently on the order of seconds, preventing use in cases involving real-time intraoperative guidance and visualization of physiological signals such as heart rate and respiration.
To increase SFDI imaging speed, a 2D Hilbert transform demodulation technique and the use of binary, square wave projection patterns are demonstrated, mitigating bottlenecks related to frame count and pattern projection rate, respectively. The Hilbert demodulation technique allows for optical property and structural orientation mapping using a single frame of data for each spatial frequency, increasing data acquisition speed by threefold. Square wave patterns are projected one to two orders of magnitude faster than sinusoids using standard SFDI hardware, eliminating the pattern refresh rate bottleneck. This approach is adapted to a real-time, multi-spectral SFDI instrument, using hardware triggering and a digital micromirror device (DMD) to generate square wave patterns having refresh rates faster than the camera exposure time. This system is capable of acquiring oxy/deoxy-hemoglobin, tissue oxygen saturation, and reduced scattering map data at 33 Hz.
Performance of the Hilbert demodulation and square wave pattern techniques are compared to conventional SFDI pattern schemes on tissue-simulating phantoms and an in vivo forearm model, showing agreement in absorption and reduced scattering maps to within 1% of conventional SFDI. The real-time SFDI instrument is applied to a pressure cuff occlusion and paced breathing model on an in vivo forearm, allowing for the acquisition of sub-second physiological signals and hemodynamics.