Ruoyang Yao

Hyperspectral time-resolved wide-field fluorescence molecular tomography based on structured light and single-pixel detection.

We present a time-resolved fluorescence diffuse optical tomography platform that is based on wide-field structured illumination, single-pixel detection, and hyperspectral acquisition. Two spatial light modulators (digital micro-mirror devices) are employed to generate independently wide-field illumination and detection patterns, coupled with a 16-channel spectrophotometer detection module to capture hyperspectral time-resolved tomographic data sets. The main system characteristics are reported, and we demonstrate the feasibility of acquiring dense 4D tomographic data sets (space, time, spectra) for time domain 3D quantitative multiplexed fluorophore concentration mapping in turbid media.

Generalized mesh-based Monte Carlo for wide-field illumination and detection via mesh retessellation.

Monte Carlo methods are commonly used as the gold standard in modeling photon transport through turbid media. With the rapid development of structured light applications, an accurate and efficient method capable of simulating arbitrary illumination patterns and complex detection schemes over large surface area is in great need. Here we report a generalized mesh-based Monte Carlo algorithm to support a variety of wide-field illumination methods, including spatial-frequency-domain imaging (SFDI) patterns and arbitrary 2-D patterns. The extended algorithm can also model wide-field detectors such as a free-space CCD camera. The significantly enhanced flexibility of source and detector modeling is achieved via a fast mesh retessellation process that combines the target domain and the source/detector space in a single tetrahedral mesh. Both simulations of complex domains and comparisons with phantom measurements are included to demonstrate the flexibility, efficiency and accuracy of the extended algorithm. Our updated open-source software is provided at http://mcx.space/mmc.

Wide-field fluorescence molecular tomography with compressive sensing based preconditioning

Wide-field optical tomography based on structured light illumination and detection strategies enables efficient tomographic imaging of large tissues at very fast acquisition speeds. However, the optical inverse problem based on such instrumental approach is still ill-conditioned. Herein, we investigate the benefit of employing compressive sensing-based preconditioning to wide-field structured illumination and detection approaches. We assess the performances of Fluorescence Molecular Tomography (FMT) when using such preconditioning methods both in silico and with experimental data. Additionally, we demonstrate that such methodology could be used to select the subset of patterns that provides optimal reconstruction performances. Lastly, we compare preconditioning data collected using a normal base that offers good experimental SNR against that directly acquired with optimal designed base. An experimental phantom study is provided to validate the proposed technique.

Improving mesoscopic fluorescence molecular tomography through data reduction

Mesoscopic fluorescence molecular tomography (MFMT) is a novel imaging technique that aims at obtaining the 3-D distribution of molecular probes inside biological tissues at depths of a few millimeters. To achieve high resolution, around 100-150μm scale in turbid samples, dense spatial sampling strategies are required. However, a large number of optodes leads to sizable forward and inverse problems that can be challenging to compute efficiently. In this work, we propose a two-step data reduction strategy to accelerate the inverse problem and improve robustness. First, data selection is performed via signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) criteria. Then principal component analysis (PCA) is applied to further reduce the size of the sensitivity matrix. We perform numerical simulations and phantom experiments to validate the effectiveness of the proposed strategy. In both in silico and in vitro cases, we are able to significantly improve the quality of MFMT reconstructions while reducing the computation times by close to a factor of two.

Time-resolved hyperspectral single-pixel camera implementation for compressive wide-field fluorescence lifetime imaging

Single-pixel imaging based on compressive sensing theory has been a highlighted technique in the biomedical imaging field for many years. This interest has been driven by the possibility of performing microscopic or macroscopic imaging based on low-cost detector arrays, increased SNR (signal-to-noise ratio) in the acquired data sets and the ability to perform high quality image reconstruction with compressed data sets by exploiting signal sparsity. In this work, we present our recent work in implementing this technique to perform time domain fluorescence-labeled investigations in preclinical settings. More precisely, we report on our time-resolved hyperspectral single-pixel camera for fast, wide-field mapping of molecular labels and lifetime-based quantification. The hyperspectral single-pixel camera implements a DMD (Digital micro-mirror device) to generate optical masks for modulating the illumination field before it is delivered onto the sample and focuses the emission light signals into a multi-anode hyperspectral time-resolved PMT (Photomultiplier tube) to acquire spatial, temporal and spectral information enriched 4-D data sets. Fluorescence dyes with lifetime and spectral contrast are embedded in well plates and thin tissues. L-1 norm based regularization or the least square method, is applied to solve the underdetermined inverse problem during image reconstruction. These experimental results prove the possibility of fast, wide-field mapping of fluorescent labels with lifetime and spectral contrast in thin media.

Sparse temporal sampling for fast time-domain wide-field fluorescence molecular tomography

Fluorescence Molecular Tomography (FMT) is a powerful optical imaging tool for preclinical research. Especially, its implementation with time-domain (TD) techniques allows lifetime multiplexing for simultaneously imaging multiple biomarkers and provides enhanced data sets for improved resolution and quantification compared to continuous wave (CW) and frequency domain (FD) methodologies. When performing time-domain reconstructions, one important aspect is the selection of a temporal sub-data set. Typically, such selection is performed a posteriori after dense temporal sampling during the acquisition. In this work, we investigate the potential to collect a priori sparse data sets for fast experimental acquisition without compromising FMT performances.

Enabling wide-field illumination and detection in mesh-based Monte Carlo simulations

We propose a generalized mesh-based Monte Carlo approach that supports various wide-field sources and free-space detectors. Simulations and phantom studies are performed to demonstrate the flexibility, efficiency and accuracy of our algorithm.

Molecular Fluorescence Tomography with Structured Light and Compressive Sensing

We apply compressive sensing based preconditioning techniques to improve time-resolved wide-field FMT reconstructions. By designing masks to illumination and detection experimental basis, the coherence of the sensitivity matrix is reduced and optical reconstructions are improved.

Hyperspectral Single-Pixel Wide-Field Time Domain Diffuse Optical Tomography

We report on the instrumentation design and experimental validation of a hyperspectral single-pixel wide-field time-resolved diffuse optical tomography (DOT) system. Reconstruction results show quantification and cross-talk improvements in fluorophore concentration mapping in turbid media.

Re-tessellated mesh-based Monte Carlo for wide-field illumination sources

Monte Carlo methods are the gold standard in modeling light propagation through complex turbid tissues. However, with the advent of structured light illumination applications, it is becoming crucial to implement fast and efficient methods to simulate arbitrary wide-field sources over large surface areas. Herein, we improve upon our previous mesh-based Monte Carlo method by enabling the simulation of an extended range of wide-field illumination strategies using a computationally efficient mesh re-tessellation technique.