This PDF file contains the front matter associated with SPIE Proceedings Volume 8799, including the Title Page, Copyright information, Table of Contents, Introduction (if any), and Conference Committee listing.

Presently time-resolved optical spectroscopy is applied with increasing success for non-invasive medical diagnostics mainly up to 1100 nm. We extended the investigation range beyond this limit, employing a supercontinuum fiber laser source and a Single-Photon Avalanche Diode in InGaAs/InP operated in gated mode. First in-vivo measurements were performed on the forearm and the breast of two healthy volunteers, reaching up to 1360 nm.

We present experimental results of time-resolved reflectance diffuse optical tomography performed with fast-gated single-photon avalanche diodes (SPADs) and show an increased imaged depth range for a given acquisition time compared to the non gated mode.

Elliptical polarization is used to explore the possibility of probing diffuse tissues at selective depths. The results of a recently published Monte Carlo simulations study are exposed. Experimental tests will be presented.

Three recipes for tissue constituent-equivalent phantoms of water and lipids are presented. Nature phantoms are made using no emulsifying agent, but just a professional disperser, instead Agar and Triton phantoms are made using agar or Triton X-100, respectively, as agents to emulsify water and lipids. Different water-to-lipid ratios ranging from 30 to 70 percent by mass are proposed and tested. Optical characterization by time-resolved spectroscopy was performed in terms of optical properties, homogeneity, reproducibility and composition retrieval.

Time-resolved diffuse optical spectroscopy measurements of phantoms at small source-detector separations yield good results for the retrieved coefficients of reduced scattering and absorption when a hybrid Green’s function of the radiative transfer equation for semi-infinite media is used.

The coupled radiative transfer - diffusion model can be used as light transport model in turbid media with non-diffusive regions. In the coupled radiative transfer - diffusion model, light propagation is modelled with the radiative transfer equation in sub-domains in which the approximations of the diffusion equation are not valid and the diffusion approximation is used elsewhere in the domain. In this work, the image reconstruction problem of diffuse optical tomography utilising the coupled radiative transfer - diffusion model is considered. Absorption and scattering distributions are estimated using the coupled radiative transfer - diffusion model as a forward model for light propagation. The results are compared to reconstructions obtained using other light transport models. The results show that the coupled radiative transfer - diffusion model can produce as good estimates for absorption and scattering as the full radiative transfer equation also in situations in which the approximations of the diffusion equation are not valid.

Fokker-Planck-Eddington approximation can be used to approximate the radiative transport equation when scattering is forward-peaked. In the approach, forward-peaked scattering probability is approximated using delta functions and smoothly varying Legendre polynomials. In this work, the approximation is used to model light propagation in turbid media with low-scattering regions. The proposed model is tested using simulations, and compared with the radiative transport equation, the diffusion approximation, and the coupled radiative transport - diffusion model. The results show that the Fokker-Planck-Eddington approximation describes light propagation with good accuracy.

Practical imaging constraints restrict the number of wavelengths that can be measured in a single Biolumines- cence Tomography imaging session, but it is unclear which set of measurement wavelengths is optimal, in the sense of providing the most information about the bioluminescent source. Mutual Information was used to integrate knowledge of the type of bioluminescent source likely to be present, the optical properties of tissue and physics of light propagation, and the noise characteristics of the imaging system, in order to quantify the information contained in measurements at different sets of wavelengths. The approach was applied to a two-dimensional sim- ulation of Bioluminescence Tomography imaging of a mouse, and the results indicate that different wavelengths and sets of wavelengths contain different amounts of information. When imaging at a single wavelength, 580nm was found to be optimal, and when imaging at two wavelengths, 570nm and 580nm were found to be optimal. Examination of the dispersion of the posterior distributions for single wavelengths suggests that information regarding the location of the centre of the bioluminescence distribution is relatively independent of wavelength, whilst information regarding the width of the bioluminescence distribution is relatively wavelength specific.

We develop a time-resolved system coupled to a new analysis method based on Mellin Laplace Transforms to reconstruct absorption and diffusion map in depth for neonate brain imaging. Phantoms and ex-vivo result are presented.

This paper emphasizes on the stability of the diffused signal acquired by a time domain diffuse optical tomography system. The robustness of the system is enhanced to make it perform consistently over a longer period of time.

Multiple artificial neural networks, which were trained with spatially resolved reflectance curves calculated by Monte Carlo simulations, were used to determine the optical properties of semi-infinite media. In comparison to one artificial neural network better results were achieved.

We consider four different Monte Carlo methods, widely used in tissue optics, based on four different ways to build photons’ trajectories. By means of numerical results we compare the temporal point spread functions calculated by the four methods for a wide range of the optical properties in the slab and semi-infinite medium geometry. Therefore, we show the statistical equivalence of the four methods and some of their convergence characteristics.

The hemodynamic change related to the brain activation can be located by the diffuse optical tomography (DOT) using the near-infrared spectroscopy (NIRS) signals and the spatial sensitivity profiles (SSP). Monte Carlo (MC) method and finite element method (FEM) have been used to predict the SSPs. The computation time for MC method is much longer than that for the FEM, however, the accurate solution in the region close to the light source cannot be obtained by FEM solutions of the diffusion equation. In this study, a hybrid MC-FEM model is proposed for fast and accurate simulation of light propagation in a highly scattering medium. In the hybrid model, the solution in the region close to the light source is calculated by the MC method whereas that in the region far from the light source is calculated by the FEM. The solutions by the FEM in hemispherical models were compared with thoseby the MC method to determine the region in which diffusion approximation does not hold and the number of photons for the MC method for the hybrid model. The results demonstratethat theproposed hybrid model can calculatethe accurate solutionswithin reasonable computation time for a multi-layered model.

In Diffuse Optical Tomography (DOT), an atlas-based model can be used as an alternative to a subject-specific anatomical model for recovery of brain activity. The main step of the generation of atlas-based subject model is the registration of atlas model to the subject head. The accuracy of the DOT then relies on the accuracy of registration method. In this work, 11 registration methods are quantitatively evaluated. The registration method with EEG 10/20 systems with 19 landmarks and non-iterative point to point algorithm provides approximately 1.4 mm surface error and is considered as the most efficient registration method.

A study is presented that demonstrates that bioluminescence tomography can reconstruct accurate 3D images of internal light sources placed at a range of depths within a physical phantom and that it provides more reliable quantitative data than standard bioluminescence imaging. Specifically, it is shown that when imaging sources at depths ranging from 5 to 15mm, estimates of total source strength are stable to within ±11% using tomography whilst values deduced by traditional methods vary 10-fold. Additionally, the tomographic approach correctly localises sources to within 1.5mm error in all cases considered.

Current uorescence di use optical tomography (fDOT) systems can provide large data sets and, in addition, the unknown parameters to be estimated are so numerous that the sensitivity matrix is too large to store. Alternatively, iterative methods can be used, but they can be extremely slow at converging when dealing with large matrices. A few approaches suitable for the reconstruction of images from very large data sets have been developed. However, they either require explicit construction of the sensitivity matrix, su er from slow computation times or can only be applied to restricted geometries. We introduce a method for fast reconstruction in fDOT with large data and solution spaces, which preserves the resolution of the forward operator whilst compressing its representation. The method does not require construction of the full matrix, and thus, allows storage and direct inversion of the explicitly constructed compressed system matrix. The method is tested using simulated data. Results show that the fDOT image reconstruction problem can be e ectively compressed, without sigini cant loss of information and with the added advantage of reducing image noise.

Image-guidance in fluorescence tomography is used to more accurately recover contrast at the cost of performing image segmentation. We propose a method of regularization which uses anatomical priors directly and does not require image segmentation

We show how a random matrix can be used to reduce the dimensionality of the bioluminescence tomography reconstruction problem. A randomised low-rank approximation for the sensitivity matrix is computed, and we show how this can be used to reconstruct the bioluminescence source distribution on a randomised basis for the mesh nodes. The distribution on the original mesh can be found easily via a simple matrix multiplication. The majority of the computation required can be performed in advance of the reconstruction, and the reconstruction time itself is of the order milliseconds. This could allow for high frame rate real-time reconstructions to be performed.

Multi-source operation in time-domain optical brain imaging often relies on the use of piezomechanical fiber switches which limit the speed when recording dynamic processes. The concept presented in this work overcomes this limitation by multiplexing on the nanosecond and microsecond time scales. In particular, the source positions were encoded by different delays on the nanosecond time scale. Multiplexing of wavelengths on the microsecond time scale (e.g. within 100 μs) was achieved by burst-mode operation of picosecond diode lasers in combination with addressing of different memory blocks in time-correlated single photon counting by means of routing inputs. This concept was implemented for 4 detectors and 5 source optodes yielding 12 measurement channels per hemisphere. In order to largely equalize the count rates for all source-detector pairs with minimal overall losses, a setup was developed that enabled the freely adjustable distribution of laser power to the various source optodes. It was based on polarization splitters and motorized broadband polarization rotators. The method was successfully demonstrated in an in vivo experiment employing two different types of motor activation of the brain.

We present results of first in-vivo tests of an optical non-contact scanning imaging system, intended to study oxidative metabolism related processes in biological tissue by means of time-resolved near-infrared spectroscopy. Our method is a novel realization of the short source-detector separation approach and based on a fast-gated single-photon avalanche diode to detect late photons only. The scanning system is built in quasi-confocal configuration and utilizes polarizationsensitive detection. It scans an area of 4×4 cm², recording images with 32×32 pixels, thus creating a high density of source-detector pairs. To test the system we performed a range of in vivo measurements of hemodynamic changes in several types of biological tissues, i.e. skin (Valsalva maneuver), muscle (venous and arterial occlusions) and brain (motor and cognitive tasks). Task-related changes in hemoglobin concentrations were clearly detected in skin and muscle. The brain activation shows weaker, but yet detectable changes. These changes were localized in pixels near the motor cortex area (C3). However, it was found that even very short hair substantially impairs the measurement. Thus the applicability of the scanner is limited to hairless parts of body. The results of our first in-vivo tests prove the feasibility of non-contact scanning imaging as a first step towards development of a prototype for biological tissue imaging for various medical applications.

Optical imaging of hemoglobin concentration changes in the exposed cortex has been used to investigate the functional brain activation. The concentration changes in oxygenated and deoxygenated hemoglobin can be independently obtained from the dual- or multi-wavelength measurements of the change in reflectance of the exposed cortex and wavelengthdependent optical path length in the cortical tissues. In the previous studies, the partial optical path length were generally estimated by homogeneous and layered models. In reality, the concentration changes in the hemoglobin only occurs in the blood vessels. In this study, the partial optical path lengths in the blood vessels were estimated by the heterogeneous model including the blood vessel structure based upon the image acquired by two-photon microscopy. Light propagation in the exposed-cortex model is simulated to estimate the wavelength dependence of the partial optical path length in the blood vessels. The wavelength dependence of the partial optical path length for the heterogeneous model was different from that for the homogeneous model. In the wavelength range from 500 to 580 nm, the partial optical path length in the blood vessels was mainly affected by the structure of the blood vessels in the region shallower than 50 μm.

We demonstrate the optimization of wavelengths for the imaging of cortical haemoglobin oxygenation with broadband RGB reflectometry. Wavelengths were chosen to minimize the crosstalk and to optimize the signal to noise ratio by simulating the effects of different combinations of wavelengths on the condition number of the resulting extinction coefficients matrices. The results obtained were evaluated for four combinations of commercial available LEDs and the condition number analysis is compared with data from literature.

Subjects at high risk for developing breast cancer for their high breast density can be effectively identified fitting a logistic regression model to time-resolved multi-wavelength optical data.

We have developed a diffuse optical tomography imaging system to track breast tumor progression in patients undergoing neoadjuvant chemotherapy. Preliminary results have shown that tumor response can be predicted by the second week of treatment.

Monte Carlo simulations and preliminary time-resolved spectroscopy measurements were performed to investigate the feasibility of the in vivo optical diagnostics of lung conditions and diseases. Absorption and reduced scattering properties of the chest, arising from in vivo spectral measurements on volunteers are presented.

Fluorescence molecular tomography is a novel approach that allows small animal imaging for early detection of cancer. However, challenges still remain in the improvement of the FMT software and hardware. In this work, a multislice FMT imaging system was developed and a finite element software with the modified calibration method was adapted to the geometry system. Phantom experiments are performed to verify the feasibility of the proposed method. The results demonstrate that the spatial resolution of the imaging system is less than 0.5mm and acquisition data time 30 minutes. In conclusion, The FMT system is capable to quantify 3D dye distribution using the well documented finite element software. OCIS codes: 110.6960, 170.3660

We analyzed the propagation of light in inhomogeneous scattering medium using voxel-based Monte-Carlo simulation with a new algorithm. We demonstrated the path-length distributions and three dimensional image of scattering properties reconstructed form simulated path-length distributions.

We propose a negative refraction for DPDW in an amplifying random medium, studied for two configurations – semiinfinite and slab, juxtaposed with absorbing ones. The negative phase introduced by the sign of the square root of the wave-vector induces unconventional modes at the interface between the two media.

The radiative transfer equation describes propagation of light in scattering media. It is widely used model, with applications in medical imaging, astronomy and atmospheric sciences to name a few. Simulating the radiative transfer equation in time-domain is, however, time consuming. In this work truncated Fourier series approximation is used to approximate the solution of the time-domain radiative transfer equation. Method is validated by comparison to direct temporal integration of the radiative transfer equation and time-domain Monte Carlo. Computational speedup is observed using the truncated Fourier series approximation in comparison to the direct temporal integration approach. Different simulation errors associated with the truncated Fourier series approximation and direct time-domain integration approaches are briefly demonstrated.

We propose a novel method for solving a system of linear equations based on non-negativity condition. This method was applied for reconstruction in fluorescence diffuse tomography and was compared with other well-known methods.

In study of the brain, oxygenation changes in the cerebral cortex are increasingly monitored using optical methods based on near-infrared spectroscopy (NIRS). When monitoring blood oxygenation in the cerebral cortex, at depth of approximately 15 mm - 20 mm from the skin surface, separation distance between source and detector becomes significant. Many studies show that by increasing the source-detector distance, illuminating light penetrates deeper into tissue. In this work, we use optical phantoms to determine experimentally the minimum source-detector distance between that allows sensing of the cerebral cortex, particularly the grey matter of the brain. A multilayered forehead phantom was fabricated and a silicon tube was added inside the phantom at depths of 15 mm and 19 mm, measured from the surface of the skin mimicking layer. This depth corresponds to the grey matter layer of the brain. The phantom’s optical properties were specifically designed to mimic the optical properties of tissue layers of the forehead and to facilitate near-infrared sensing. Optical sensing of liquid movement within the tube was measured by varying the distance between the near-infrared light source and the detector. Based on our measurements, we can conclude that it is possible to sense pulsations from a grey matter mimicking layer of the brain using near-infrared spectroscopy at a source-detector distance of 3 - 4 cm.

We report an application of Mesoscopic Fluorescence Molecular Tomography to 3D tissue engineering construct. Engineered thick tissue was hosting two 3D printed vasculatures. The channels were formed by live cells, expressing GFP and mCherry reporter genes, embedded in 3mm turbid media. Tissue and cells kept in a 3mm thick perfusion chamber during the entire imaging process which took less than 5 minutes.