Current Research Directions

We focus on the development of novel optical imaging and computational technologies to probe functional and structural information for diffusive media such as biological tissues. Below is a summary of our current research areas and recent works.

Spatial Frequency Domain Imaging

The Spatial Frequency Domain Imaging (SFDI) is an emerging label-free imaging technique that can provide quantitative chromophore concentrations in tissue. It has been applied to various biomedical scenarios for the mapping of oxy-hemoglobin, deoxy-hemoglobin, water and lipids, including burn wound monitoring, tumor monitoring, clinical tissue flap monitoring, inflammation monitoring, brown fat identification, as well as blood lipids monitoring. 

Hyperspectral Shortwave Infrared Meso-Patterned Imaging (SWIR-MPI)

Shortwave Infrared (SWIR) wavelengths have been widely used to increase imaging depth for natural scenes. The SWIR can also provide increased penetration depth for biological tissues. More importantly, it gives access to the absorption features of water and lipids which could not be probed by the near-infrared. We are developing a new imaging modality termed shortwave-infrared meso-patterned imaging (SWIR-MPI) that is able to provide label-free quantitative wide-field mapping of tissue optical properties in animal models and humans. The spatio-temporal concentrations of those species can be subsequently obtained with Beer's law. While water and lipids are key participants in many fundamental biological processes, SWIR-MPI would enable new capabilities in fundamental studies and clinical monitoring of obesity, thermoregulation, cancer, point-of-care monitoring, and cardiovascular disease.

Deep Learning Algorithms

Spatial frequency domain imaging (SFDI) is emerging as an important new method in biomedical imaging due to its ability to provide label-free, wide-field tissue optical absorption and scattering maps. Most prior SFDI studies have utilized 2 spatial frequencies (2-fx) for optical property extractions. The use of more than 2 frequencies (multi-fx) can vastly improve the accuracy and reduce uncertainties in optical property estimates for some tissue types, but has been limited in practice due to the slow speed of available inversion algorithms. We are developing novel deep learning algorithms that eliminates the speed bottleneck in SFDI, which would enable real-time, highly-accurate measurements for optical properties, chromophore concentrations, and tissue oxygenation.

Deep learning for ultrafast multi-fx SFDI

Multi-fx SFDI has the advantage of accuracy on tissue measurements, but is severely limited by slow optical property (OP) inversion speed. The conventional method needs to solve the inverse problem with iterative optimization at each pixel location, which is time-consuming and takes over 10 hours to process a 696x520 image. The proposed deep learning model dramatically reduces the processing time from 10 hours down to 0.2 seconds, speeding up the OP inversion over 100,000x faster.

Deep learning model for direct chromophore mapping

Existing inversion algorithms have to first convert the multi-fx diffuse reflectance to optical absorptions, and then solve a set of linear equations to estimate chromophore concentrations. We present a deep learning framework, noted as a deep residual network (DRN), that is able to directly map from diffuse reflectance to chromophore concentrations. The proposed DRN is over 10x faster than the state-of-the-art method for chromophore inversion and enables 25x improvement on the frame rate for in vivo real-time oxygenation mapping.

Wavefront Shaping

Focusing light inside scattering media by feedback-based wavefront shaping has been intensively investigated due to its potential applications such as deep tissue imaging. The state-of-the-art method, i.e. genetic algorithm, conducts phase optimization in a "half-blind" manner that relies on random variations of the population such as crossover and mutation, which is known to be time-consuming and can only achieve relatively low enhancement of the focus. We establish a gradient method to guide the focusing process with gradient information during the whole process of the phase optimization in the presence of scattering. The method is verified with both simulations and experiments for focusing light through scattering media. By incorporating the gradient information, this method enables us to achieve approximately 1000 peak-to-background ratio (PBR). Additionally, it is 60x faster compared to the state-of-the-art. It also improves the PBR with a linear time cost, whereas the genetic algorithm requires exponentially increased time cost.

Gradient-assisted wavefront shaping

The figure above shows an illustration of wavefront shaping and the flowchart of the gradient-assisted phase optimization. Compared to the state-of-the-art genetic algorithm method which has been widely used in the wavefront front shaping field, the gradient-assisted method is able to improved the speed by 60x and can achieve 1000 PBR.

Fast wavefront shaping with multi-pixel encoded digital-micromirror device

The figure above shows experimental setup of multi-pixel-encoding fast wavefront shaping system. Based on the gradient method, we further proposed a multi-pixel encoded DMD based wavefront shaping method by combining multiple micromirrors into a single modulation unit to overcome the drawbacks of traditional binary modulation. Compared with the state-of-the-art DMD based wavefront shaping method, the proposed method increased the speed of optimization and enhancement of focus by a factor of 179 and 16, respectively. The experimental results suggest that the proposed method will pave a new avenue for WS in the applications of biomedical imaging, photon therapy, optogenetics, dynamic holographic display, etc.