Time-domain scanning optical mammography: II. Optical properties and tissue parameters of 87 carcinomas

Within a clinical trial on scanning time-domain optical mammography reported on in a companion publication (part I), craniocaudal and mediolateral projection optical mammograms were recorded from 154 patients, suspected of having breast cancer. Here we report on in vivo optical properties of the subset of 87 histologically validated carcinomas which were visible in optical mammograms recorded at two or three near-infrared wavelengths. Tumour absorption and reduced scattering coefficients were derived from distributions of times of flight of photons recorded at the tumour site employing the model of diffraction of photon density waves by a spherical inhomogeneity, located in an otherwise homogeneous tissue slab. Effective tumour radii, taken from pathology, and tumour location along the compression direction, deduced from off-axis optical scans of the tumour region, were included in the analysis as prior knowledge, if available. On average, tumour absorption coefficients exceeded those of surrounding healthy breast tissue by a factor of about 2.5 (670 nm), whereas tumour reduced scattering coefficients were larger by about 20% (670 nm). From absorption coefficients at 670 nm and 785 nm total haemoglobin concentration and blood oxygen saturation were deduced for tumours and surrounding healthy breast tissue. Apart from a few outliers total haemoglobin concentration was observed to be systematically larger in tumours compared to healthy breast tissue. In contrast, blood oxygen saturation was found to be a poor discriminator for tumours and healthy breast tissue; both median values of blood oxygen saturation are the same within their statistical uncertainties. However, the ratio of total haemoglobin concentration over blood oxygen saturation further improves discrimination between tumours and healthy breast tissue. For 29 tumours detected in optical mammograms recorded at three wavelengths (670 nm, 785 nm, 843 nm or 884 nm), scatter power was derived from transport scattering coefficients. Scatter power of tumours tends to be larger than that of surrounding healthy breast tissue, yet the 95% confidence intervals of both medians overlap.     

Multispectral breast imaging using a ten-wavelength, 64 x 64 source/detector channels silicon photodiode-based diffuse optical tomography system

We describe a compact diffuse optical tomography system specifically designed for breast imaging. The system consists of 64 silicon photodiode detectors, 64 excitation points, and 10 diode lasers in the near-infrared region, allowing multispectral, three-dimensional optical imaging of breast tissue. We also detail the system performance and optimization through a calibration procedure. The system is evaluated using tissue-like phantom experiments and an in vivo clinic experiment. Quantitative two-dimensional (2D) and three-dimensional (3D) images of absorption and reduced scattering coefficients are obtained from these experiments. The ten-wavelength spectra of the extracted reduced scattering coefficient enable quantitative morphological images to be reconstructed with this system. From the in vivo clinic experiment, functional images including deoxyhemoglobin, oxyhemoglobin, and water concentration are recovered and tumors are detected with correct size and position compared with the mammography.     

Reduction of image artifacts induced by change in the optode coupling in time-resolved diffuse optical tomography

Tomographic images of the optical properties can be reconstructed using inversion algorithms for diffuse optical tomography (DOT); however, changes in the optode coupling that occurs while obtaining an object's measurements may often lead to the presence of artifacts in the reconstructed images. To reduce the number of artifacts induced by optode coupling, previous studies have introduced (unknown) coupling coefficients in reconstruction algorithms, which were found to be effective for continuous wave- and frequency-domain DOT. This study aims to investigate the effects of optode calibration on the reconstructed images of time-domain DOT. Here, coupling coefficients are incorporated into the time-domain DOT algorithm based on a modified generalized pulse spectrum technique. The images of the absorption coefficient are reconstructed in various numerical simulations, phantom experiments, and in vivo experiments of time-domain DOT. As a result, the artifacts resulting from changes in optode coupling are reduced in the reconstructed images of the absorption coefficient, thereby demonstrating that introduction of coupling coefficients is effective in time-domain DOT. Moreover, numerical simulations, phantom experiments, and in vivo studies have validated this algorithm.     

Quantification of optical properties of a breast tumor using random walk theory

For the first time we use a random walk methodology based on time-dependent contrast functions to quantify the optical properties of breast tumors (invasive ductal carcinoma) of two patients. Previously this theoretical approach was successfully applied for analysis of embedded objects in several phantoms. Data analysis was performed on distributions of times of flight for photons transmitted through the breast which were recorded in vivo using a time-domain scanning mammograph at 670 and 785 nm. The size of the tumors, their optical properties, and those of the surrounding tissue were reconstructed at both wavelengths. The tumors showed increased absorption and scattering. From the absorption coefficients at both wavelengths blood oxygen saturation was estimated for the tumors and the surrounding tissue.     

Poly(vinyl alcohol) gels as photoacoustic breast phantoms revisited

A popular phantom in photoacoustic imaging is poly(vinyl alcohol) (PVA) hydrogel fabricated by freezing and thawing (F-T) aqueous solutions of PVA. The material possesses acoustic and optical properties similar to those of tissue. Earlier work characterized PVA gels in small test specimens where temperature distributions during F-T are relatively homogeneous. In this work, in breast-sized samples we observed substantial temperature differences between the shallow regions and the interior during the F-T procedure. We investigated whether spatial variations were also present in the acoustic and optical properties. The speed of sound, acoustic attenuation, and optical reduced scattering coefficients were measured on specimens sampled at various locations in a large phantom. In general, the properties matched values quoted for breast tissue. But while acoustic properties were relatively homogeneous, the reduced scattering was substantially different at the surface compared with the interior. We correlated these variations with gel microstructure inspected using scanning electron microscopy. Interestingly, the phantom's reduced scattering spatial distribution matches the optical properties of the standard two-layer breast model used in x ray dosimetry. We conclude that large PVA samples prepared using the standard recipe make excellent breast tissue phantoms.     

Time-domain scanning optical mammography: I. Recording and assessment of mammograms of 154 patients

Using a triple wavelength (670 nm, 785 nm, 843/884 nm) scanning laser-pulse mammograph we recorded craniocaudal and mediolateral projection optical mammograms of 154 patients, suspected of having breast cancer. From distributions of times of flight of photons recorded at typically 1000-2000 scan positions, optical mammograms were derived displaying (inverse) photon counts in selected time windows, absorption and reduced scattering coefficients or total haemoglobin concentration and blood oxygen saturation. Optical mammograms were analysed by comparing them with x-ray and MR mammograms, including results of histopathology, attributing a subjective visibility score to each tumour assessed. Out of 102 histologically confirmed tumours, 72 tumours were detected retrospectively in both optical projection mammograms, in addition 20 cases in one projection only, whereas 10 tumours were not detectable in any projection. Tumour contrast and contrast-to-noise ratios of mammograms of the same breast, but derived from measured DTOFs by various methods were quantitatively compared. On average, inverse photon counts in selected time windows, including total photon counts, provide highest tumour contrast and contrast-to-noise ratios. Based on the results of the present study we developed a multi-wavelength, multi-projection scanning time-domain optical mammograph with improved spectral and spatial (angular) sampling, that allows us to record entire mammograms simultaneously at various offsets between the transmitting fibre and receiving fibre bundle and provides first results for illustration.     

Quantitative photoacoustic imaging: correcting for heterogeneous light fluence distributions using diffuse optical tomography

The specificity of molecular and functional photoacoustic (PA) images depends on the accuracy of the photoacoustic absorption spectroscopy. The PA signal is proportional to the product of the optical absorption coefficient and local light fluence; quantitative PA measurements of the optical absorption coefficient therefore require an accurate estimation of optical fluence. Light-modeling aided by diffuse optical tomography (DOT) can be used to map the required fluence and to reduce errors in traditional PA spectroscopic analysis. As a proof-of-concept, we designed a tissue-mimicking phantom to demonstrate how fluence-related artifacts in PA images can lead to misrepresentations of tissue properties. To correct for these inaccuracies, the internal fluence in the tissue phantom was estimated by using DOT to reconstruct spatial distributions of the absorption and reduced scattering coefficients of multiple targets within the phantom. The derived fluence map, which only consisted of low spatial frequency components, was used to correct PA images of the phantom. Once calibrated to a known absorber, this method reduced errors in estimated absorption coefficients from 33% to 6%. These results experimentally demonstrate that combining DOT with PA imaging can significantly reduce fluence-related errors in PA images, while producing quantitatively accurate, high-resolution images of the optical absorption coefficient.     

Design, fabrication, and characterization of a tissue-equivalent phantom for optical elastography

We suitably adapt the design of a tissue-equivalent phantom used for photoacoustic imaging to construct phantoms for optical elastography. The elastography phantom we consider should have optical properties such as scattering coefficient, scattering anisotropy factor, and refractive index; mechanical properties such as storage and loss modulus; and acoustic properties such as ultrasound velocity, attenuation coefficient, and acoustic impedance to match healthy and diseased tissues. The phantom is made of poly (vinyl alcohol) (PVA) and its mechanical, optical, and acoustic properties are tailored by physical cross-linking effected through subjecting a suitable mix of PVA stock and water to a number of freeze-thaw cycles and by varying the degree of hydrolysis in the PVA stock. The optical, mechanical, and acoustic properties of the samples prepared are measured by employing different techniques. The measured variations in the values of optical scattering coefficient, scattering anisotropy factor, and refractive index and storage modulus are found to be comparable to those in normal and diseased breast tissues. The acoustic properties such as sound speed, acoustic attenuation coefficient, and density are found to be close to the average values reported in the literature for normal breast tissue.     

Changes in tissue optical properties due to radio-frequency ablation of myocardium

The optical properties of pig heart tissue were measured after in vivo ablation therapy had been performed during open-heart surgery. In vitro samples of normal and ablated tissue were subjected to measurements with an optically integrating sphere set-up in the region 470–900 nm. Three independent measurements were made: total transmittance, total reflectance and collimated transmittance, which made it possible to extract the absorption and scattering coefficients and the scattering anisotropy factor g, using an inverse Monte Carlo model. Between 470 and 700 nm, only the reduced scattering coefficient and absorption could be evaluated. The absorption spectra were fitted to known tissue chromophore spectra, so that the concentrations of haemoglobin and myoglobin could be estimated. The reduced scattering coefficient was compared with Mie computations to provide Mie equivalent average radii. Most of the absorption was from myoglobin, whereas haemoglobin absorption was negligible. Metmyoglobin was formed in the ablated tissue, which could yield a spectral signature to distinguish the ablated tissue with a simple optical probe to monitor the ablation therapy. The reduced scattering coefficient increased by, on average, 50% in the ablated tissue, which corresponded to a slight decrease in the Mie equivalent radius.     

Determination of optical properties in semi-infinite turbid media using imaging measurements of frequency-domain photon migration obtained with an intensified charge-coupled device

Frequency-domain photon migration measurements across the surface of a tissue-mimicking, semi-infinite phantom are acquired via an intensified charge-coupled device (ICCD) detection system and used in conjunction with the diffusion approximation to determine the optical properties. The absorption and reduced scattering coefficients are determined least accurately when relative measurements of average light intensity I(rel)dc are employed either alone or in a combination with relative modulation amplitude data I(rel)ac and/or relative phase shift data theta(rel). The absorption and reduced scattering coefficients are found accurate to within 15 and 11%, respectively, of the values obtained from standard single-pixel measurements when theta(rel) measurements are employed alone or in combination with I(rel)ac data.     

Experimental validation of Monte Carlo modeling of fluorescence in tissues in the UV-visible spectrum

The goal of the work is to experimentally verify Monte Carlo modeling of fluorescence and diffuse reflectance measurements in turbid, tissue phantom models. In particular, two series of simulations and experiments, in which one optical parameter (absorption or scattering coefficient) is varied while the other is fixed, are carried out to assess the effect of the absorption coefficient (mu(a)) and scattering coefficient (mu(s)) on the fluorescence and diffuse reflectance measured from a turbid medium. Moreover, simulations and experiments are carried out for several fiber optic probe geometries that are designed to sample small tissue volumes. Additionally, a group of conversion expressions are derived to convert the optical properties and fluorescence quantum yield measured from tissue phantoms for use in Monte Carlo simulations. The conversions account for the differences between the definitions of the absorption coefficient and fluorescence quantum yield of fluorophores in a tissue phantom model and those in a Monte Carlo simulation. The results indicate that there is good agreement between the simulated and experimentally measured results in most cases. This dataset can serve as a systematic validation of Monte Carlo modeling of fluorescent light propagation in tissues. The simulations are carried out for a wide range of absorption and scattering coefficients as well as ratios of scattering coefficient to absorption coefficient, and thus would be applicable to tissue optical properties over a wide wavelength range (UV-visible/near infrared). The fiber optic probe geometries that are modeled in this study include those commonly used for measuring fluorescence from tissues in practice.     

Fiber dispersion in time domain measurements compromising the accuracy of determination of optical properties of strongly scattering media

Limitations in the accuracy of measurements of optical properties (absorption and scattering coefficients) of strongly scattering media that are due to temporal dispersion inside the detection fibers of a time-of-flight setup were investigated for a tissue-like phantom. It is shown that the absorption and reduced scattering coefficients may be overestimated by up to 90% (depending on length and numerical aperture of the fibers) if the instrumental response measurements for the setup are performed by direct illumination of the tip of the detection fibers with collimated short laser pulses. However, the accuracy can be improved significantly if a thin layer of scattering media is used in front of the detection fibers during the response measurements. The relevance of the investigated dispersion effects is discussed with respect to frequency-domain measurements as well.     

Development of tissue-simulating optical phantoms: poly-N-isopropylacrylamide solution entrapped inside a hydrogel

The average turbid optical properties of the N-isopropylacrylamide (NIPA) polymer solution entrapped inside a polyacrylamide hydrogel (called an NIPA/PAAM gel system) were studied using a multiwavelength oblique-incidence reflectometer. The turbidity of such a system can be drastically changed by simply switching the temperature from below the low critical solution temperature of the NIPA, around 33 degrees C, to above. The absorption coefficient and the reduced scattering coefficient were obtained as a function of wavelength for samples with selected NIPA and blue dextran concentrations. It is found that the scattering of the optical phantom comes from the NIPA polymer chains and the absorption from the blue dextran. The turbid optical properties of an NIPA/PAAM gel system can be tuned to simulate biological tissues at a specific wavelength by varying compositions of NIPA and blue dextran and further modified by controlling the temperature.     

Broadband absorption spectroscopy of turbid media using a dual step steady-state method

We present a method for the determination of the absorption coefficient of turbid media in a broad wavelength range with high spectral resolution using a dual step method. First, the reduced scattering coefficient is determined for a few wavelengths with spatially resolved reflectance measurements. The reduced scattering coefficient for the intermediate wavelengths is interpolated by fitting a power law. Second, the absorption coefficient is obtained from measurements of the total reflectance using the a priori knowledge of the reduced scattering coefficient. By applying a white light source and a spectrometer to measure the total reflectance, the absorption coefficient is determined with a high spectral resolution. The methodology is verified by comparing the absorption coefficients determined by the spatially resolved reflectance measurements with those obtained by the dual step method. The influence of an unknown refractive index and phase function on the determination of the optical properties is investigated. In addition, the optical properties of Intralipid/ink phantoms and the fat layer of porcine rind were determined. The absorption coefficient of the investigated phantoms varying by four orders of magnitude could be determined with an average error of less than 10%.     

The determination of in vivo human tissue optical properties and absolute chromophore concentrations using spatially resolved steady-state diffuse reflectance spectroscopy

A method is described for measuring optical properties and deriving chromophore concentrations from diffuse reflection measurements at the surface of a turbid medium. The method uses a diffusion approximation model for the diffuse reflectance, in combination with models for the absorption and scattering coefficients. An optical fibre-based set-up, capable of measuring nine spectra from 400 to 1050 nm simultaneously, is used to test the method experimentally. Results of the analyses of phantom and in vivo measurements are presented. These demonstrate that in the wavelength range from 600 to 900 nm, tissue scattering can be described as a simple power dependence of the wavelength and that the tissue absorption can be accurately described by the addition of water, oxy- and deoxyhaemoglobin absorption.     

Optical properties of normal and diseased human breast tissues in the visible and near infrared

The optical absorption and scattering coefficients have been determined for specimens of normal and diseased human breast tissues over the range of wavelengths from 500 to 1100 nm. Total attenuation coefficients were measured for thin slices of tissue cut on a microtome. The diffuse reflectance and transmittance were measured for 1.0 mm thick samples of these tissues, using standard integrating sphere techniques. Monte Carlo simulations were performed to derive the scattering and absorption coefficients, as well as the mean cosine of the scattering angle. The results indicate that scatter exceeds absorption by at least two orders of magnitude. Absorption is most significant at wavelengths below 600 nm. The scattering coefficients lie in the range 30-90 mm-1 at 500 nm, and fall smoothly with increasing wavelength to between 10 and 50 mm-1 at 1100 nm. The scattering coefficient for adipose tissue differs, in that it is invariant with wavelength over this spectral range. For all tissues examined, the scattered light is highly forward peaked, with the mean cosine of the scattering angle in the range 0.945-0.985. Systematic differences between the optical properties of some tissue types are demonstrated.     

Intraoperatively assessed optical properties of malignant and healthy breast tissue used to determine the optimum wavelength of contrast for optical mammography

We use spatially resolved diffuse remittance spectroscopy (DRS) for the measurement of absorption (mu(a)) and reduced scattering (mu(s)') coefficients of normal and malignant breast tissue in vivo during surgery. Prior to these measurements, the linearity of the measurement technique was evaluated on liquid optical phantoms. In addition, the reproducibility of in-vivo tissue measurements was determined on a healthy volunteer. We present results of the in-vivo measurement of optical properties in the wavelength range from 600 to 1100 nm performed during radical mastectomy. A total of 24 patients were included in the study. Both the absorption and reduced scattering properties show large variations. Significant differences in optical properties between normal (glandular plus lipid rich tissue) and tumor tissues are present in 74% of all patients. However, in some cases the tumor showed lower values than normal tissue, and in other cases this was the other way around. Thus, a general trend in optical properties is not observed. However, the average absorption contrast of all patients as a function of wavelength reveals an optimal contrast peak at 650 nm. We believe that this relates to a difference in vascular saturation between tumor and adjacent normal tissue.     

Poly(vinyl alcohol) gels for use as tissue phantoms in photoacoustic mammography

Materials for solid photoacoustic breast phantoms, based on poly(vinyl alcohol) hydrogels, are presented. Phantoms intended for use in photoacoustics must possess both optical and acoustic properties of tissue. To realize the optical properties of tissue, one approach was to optimize the number of freezing and thawing cycles of aqueous poly(vinyl alcohol) solutions, a procedure which increases the turbidity of the gel while rigidifying it. The second approach concentrated on forming a clear matrix of the rigid poly(vinyl alcohol) gel without any scattering, so that appropriate amounts of optical scatterers could be added at the time of formation, to tune the optical properties as per requirement. The relevant optical and acoustic properties of such samples were measured to be close to the average properties of human breast tissue. Tumour simulating gel samples of suitable absorption coefficient were created by adding appropriate quantities of dye at the time of formation; the samples were then cut into spheres. A breast phantom embedded with such 'tumours' was developed for studying the applicability of photoacoustics in mammography.     

Reflectance-based determination of optical properties in highly attenuating tissue

Accurate data on in vivo tissue optical properties in the ultraviolet A (UVA) to visible (VIS) range are needed to elucidate light propagation effects and to aid in identifying safe exposure limits for biomedical optical spectroscopy. We have performed a preliminary study toward the development of a diffuse reflectance system with maximum fiber separation distance of less than 2.5 mm. The ultimate objective is to perform endoscopic measurement of optical properties in the UVA to VIS. Optical property sets with uniformly and randomly distributed values were developed within the range of interest: absorption coefficients from 1 to 25 cm(-1) and reduced scattering coefficients from 5 to 25 cm(-1). Reflectance datasets were generated by direct measurement of Intralipid-dye tissue phantoms at lambda=675 nm and Monte Carlo simulation of light propagation. Multivariate calibration models were generated using feed-forward artificial neural network or partial least squares algorithms. Models were calibrated and evaluated using simulated or measured reflectance datasets. The most accurate models developed-those based on a neural network and uniform optical property intervals-were able to determine absorption and reduced scattering coefficients with root mean square errors of +/-2 and +/-3 cm(-1), respectively. Measurements of ex vivo bovine liver at 543 and 633 nm were within 5 to 30% of values reported in the literature. While our technique for determination of optical properties appears feasible and moderately accurate, enhanced accuracy may be achieved through modification of the experimental system and processing algorithms.     

Reduction of Poisson noise in measured time-resolved data for time-domain diffuse optical tomography

A method to reduce noise for time-domain diffuse optical tomography (DOT) is proposed. Poisson noise which contaminates time-resolved photon counting data is reduced by use of maximum a posteriori estimation. The noise-free data are modeled as a Markov random process, and the measured time-resolved data are assumed as Poisson distributed random variables. The posterior probability of the occurrence of the noise-free data is formulated. By maximizing the probability, the noise-free data are estimated, and the Poisson noise is reduced as a result. The performances of the Poisson noise reduction are demonstrated in some experiments of the image reconstruction of time-domain DOT. In simulations, the proposed method reduces the relative error between the noise-free and noisy data to about one thirtieth, and the reconstructed DOT image was smoothed by the proposed noise reduction. The variance of the reconstructed absorption coefficients decreased by 22% in a phantom experiment. The quality of DOT, which can be applied to breast cancer screening etc., is improved by the proposed noise reduction.