Quantifying the optical properties (absorption and scattering) of strongly turbid media is crucial for characterizing biological tissues and fluid fields. Current methods, such as photoacoustic imaging and hyperspectral imaging, offer limitations in providing wide-field quantification and high-speed capabilities. Diffuse optical spectroscopy (DOS) can separate absorption from scattering but is slow and requires contact. The need exists for a high-speed, wide-field, label-free, and non-contact method to quantify optical properties. This paper addresses this need by developing halftone-SFDI, a novel technique designed to overcome the limitations of existing methods and provide quantitative optical property imaging at kilohertz speeds. The ability to achieve such high speeds opens up possibilities for studying dynamic processes in various fields, including brain science and fluid dynamics, where rapid changes in optical properties need to be captured in real-time. The inherent challenges in quantifying optical properties in strongly turbid media stem from the convolution of absorption and scattering effects which hinder direct measurement. Existing techniques like frequency-domain and time-domain diffuse optical techniques utilize modulated illumination and model-based analysis (often employing the diffusion approximation or Monte Carlo simulations) to separate these effects. However, these techniques are typically slow, especially for wide-field imaging, limiting their applicability to dynamic systems. This paper introduces halftone-SFDI as a solution to address these limitations by using a halftone strategy to significantly accelerate the measurement process.

The paper reviews existing techniques for quantifying optical properties in turbid media. Photoacoustic imaging (PA) is mentioned as a technique that can probe absorbing contents in tissue, but it cannot obtain quantitative absorption and scattering values. Hyperspectral imaging is noted as a more accessible technique, but it cannot reliably quantify specific tissue chromophores due to the confounding effects of optical absorption and scattering. Diffuse optical spectroscopy (DOS) is presented as a technique that can separate absorption and scattering but is limited to point measurements, requires mechanical contact, and is slow, with imaging speeds on the order of 1e-3 Hz. The limitations of existing techniques highlight the need for a faster, wide-field, non-contact method for quantifying optical properties in strongly turbid media, motivating the development of halftone-SFDI.

Halftone-SFDI employs a halftone strategy to significantly increase the speed of spatial frequency domain imaging (SFDI). A digital micromirror device (DMD) projects spatially modulated light patterns onto the sample. In contrast to conventional SFDI using continuous-tone 8-bit grayscale patterns (limited to ~290 Hz), halftone-SFDI uses 1-bit binary patterns, achieving a maximum projection rate of 23 kHz. Five projection patterns (planar illumination, black image, and three patterns with different phases) are used for each wavelength. The reflected light is captured by a camera synchronized with the illumination. Cross-polarizers minimize specular reflection. Optical properties (absorption (µa) and reduced scattering (µs')) are extracted using a pre-computed look-up table (LUT) generated from Monte Carlo simulations. The LUT maps diffuse reflectance values at two spatial frequencies (e.g., 0 and 0.1 mm⁻¹) to µa and µs'. The process involves demodulation and calibration of raw reflectance images to obtain diffuse reflectance maps, followed by pixel-by-pixel calculation of optical properties using the LUT. The halftone approach achieves a significant speed improvement (~80x) without hardware modifications. Phantom studies and in vivo measurements on human tissue are used to validate the accuracy of halftone-SFDI by comparing its results to those of conventional continuous-tone SFDI.

The phantom validation study demonstrates that halftone-SFDI and continuous-tone SFDI provide nearly identical measurements of optical properties, with percent differences within the noise level (0.7 ± 0.9% for diffuse reflectance, -1.6 ± 1.8% for absorption, and 0.6 ± 0.6% for reduced scattering). In vivo human tissue measurements further confirm the equivalence of the two methods, showing negligible differences in optical property and chromophore concentration maps. The in vivo rat brain cortex study successfully demonstrates longitudinal monitoring of absolute chromophore concentration and spatial distribution at 15 Hz using 685 nm and 850 nm wavelengths. Furthermore, the study showcases kHz high-speed dual-wavelength monitoring of a highly dynamic flow field's optical properties, demonstrating the potential of halftone-SFDI for high-speed, quantitative imaging of dynamic systems. The noise levels for optical absorption and reduced scattering measurements were determined to be 2.5% and 1.3%, respectively, based on repeated measurements. For chromophore concentration measurements, the noise levels were 4.2% for oxy-hemoglobin and 7.6% for deoxy-hemoglobin. The agreement between halftone-SFDI and continuous-tone SFDI across various metrics validates the efficacy and accuracy of the proposed method.

The results demonstrate that halftone-SFDI provides accurate and high-speed quantitative optical property imaging in strongly turbid media. The method's ability to achieve kilohertz speeds opens new possibilities for studying dynamic processes in various fields. The in vivo experiments showcase the potential of halftone-SFDI for applications in brain science and fluid dynamics. The significant speed improvement compared to existing techniques makes halftone-SFDI suitable for longitudinal studies and real-time monitoring of dynamic phenomena. The accuracy and reliability demonstrated through the validation studies establish halftone-SFDI as a valuable tool for quantitative optical property imaging.

Halftone-SFDI offers a significant advancement in optical property imaging. Its high speed, label-free, non-contact, and wide-field capabilities overcome limitations of existing techniques. The validation studies confirm the accuracy and reliability of the method. Future research could explore applications in diverse fields, such as advanced medical imaging and dynamic fluid studies. Further optimization of the system and exploration of broader ranges of spatial frequencies could also enhance the technique's capabilities.

While halftone-SFDI demonstrates significant speed improvements, the current implementation is limited by the camera's frame rate, preventing the full utilization of the DMD's 23 kHz projection rate. The accuracy of the method relies on the accuracy of the Monte Carlo LUT, and improvements to the LUT could further enhance the precision of the measurements. The study focuses on specific wavelengths (685nm and 850nm) and spatial frequencies (0 and 0.1 mm⁻¹), and further investigations are necessary to explore the method's performance with other wavelengths and frequencies.