Anant Agrawal

A review of consensus test methods for established medical imaging modalities and their implications for optical coherence tomography

In recent years there has been increasing interest in development of consensus, tissue-phantom-based approaches for assessment of biophotonic imaging systems, with the primary goal of facilitating clinical translation of novel optical technologies. Well-characterized test methods based on tissue phantoms can provide useful tools for performance assessment, thus enabling standardization and device inter-comparison during preclinical development as well as quality assurance and re-calibration in the clinical setting. In this review, we study the role of phantom-based test methods as described in consensus documents such as international standards for established imaging modalities including X-ray CT, MRI and ultrasound. Specifically, we focus on three image quality characteristics - spatial resolution, spatial measurement accuracy and image uniformity - and summarize the terminology, metrics, phantom design/construction approaches and measurement/analysis procedures used to assess these characteristics. Phantom approaches described are those in routine clinical use and tend to have simplified morphology and biologically-relevant physical parameters. Finally, we discuss the potential for applying knowledge gained from existing consensus documents in the development of standardized, phantom-based test methods for optical coherence tomography.

Regulatory perspectives and research activities at the FDA on the use of phantoms with in vivo diagnostic devices

For a number of years, phantoms have been used to optimize device parameters and validate performance in the primary medical imaging modalities (CT, MRI, PET/SPECT, ultrasound). Furthermore, the FDA under the Mammography Quality Standards Act (MQSA) requires image quality evaluation of mammography systems using FDA-approved phantoms. The oldest quantitative optical diagnostic technology, pulse oximetry, also benefits from the use of active phantoms known as patient simulators to validate certain performance characteristics under different clinically-relevant conditions. As such, guidance provided by the FDA to its staff and to industry on the contents of pre-market notification and approval submissions includes suggestions on how to incorporate the appropriate phantoms in establishing device effectiveness. Research at the FDA supports regulatory statements on the use of phantoms by investigating how phantoms can be designed, characterized, and utilized to determine critical device performance characteristics. These examples provide a model for how novel techniques in the rapidly growing field of optical diagnostics can use phantoms during pre- and post-market regulatory testing.

Multi-system comparison of optical coherence tomography performance with point spread function phantoms

Point spread function (PSF) phantoms based on unstructured distributions of sub-resolution particles in a transparent matrix have proven effective for evaluating resolution and its spatial variation in optical coherence tomography (OCT) systems. Measurements based on PSF phantoms have the potential to become a standard test method for consistent, objective and quantitative inter-comparison of OCT system performance. Towards this end, we have evaluated three PSF phantoms and investigated their ability to compare the performance of four OCT systems. The phantoms are based on 260-nm-diameter gold nanoshells, submicron-diameter iron oxide particles and 1.5-micron-diameter silica particles. The OCT systems included spectral-domain and swept source systems in free-beam geometries as well as a time-domain system in both free-beam and fiberoptic probe geometries. Results indicated that iron oxide particles and gold nanoshells were most effective for measuring spatial variations in the magnitude and shape of PSFs across the image volume. The intensity of individual particles was also used to evaluate spatial variations in signal intensity uniformity. Significant system-to-system differences in resolution and signal intensity and their spatial variation were readily quantified. The phantoms proved useful for identification and characterization of irregularities such as astigmatism. Particle concentrations of 5000 per cubic millimeter or greater provided accurate determination of performance metrics. Our multi-system inter-comparison provides evidence of the effectiveness of PSF-phantom-based test methods for comparison of OCT system resolution and signal uniformity.

Computational modeling of device-tissue interface geometries for time-resolved fluorescence in layered tissue

Temporal measurements of fluorescence emitted from biological tissue provide information on biochemistry and morphology which may be useful in identifying neoplasia onset. Depth-selective detection of time-resolved fluorescence may enable enhanced discrimination of signals originating from individual tissue layers and thus improve device efficacy. In this study, we investigate how illumination-collection design parameters influence a devices ability to measure fluorophore lifetime and changes in superficial layer thickness. A two-layer, time-resolved Monte Carlo model of fluorescence light propagation in colonic polyps was used to simulate temporal decay curves. Several normal- and oblique-incidence geometries were investigated. Also, the efficacy of a convolution-based, bi-exponential lifetime calculation is compared to a full-width-half-max decay curve metric. Results indicate that interface design has a significant effect on the accuracy of fluorophore lifetime estimates and the ability to discriminate changes in tissue morphology. This is due to changes in the relative contribution of each tissue layer to the total detected signal.

Quantitative analysis of low contrast detectability in optical coherence tomography

Optical coherence tomography (OCT) is a high resolution imaging technology that is rapidly being adopted as the standard of care for medical applications such as ocular and intravascular imaging. However, clinical translation has been hampered by the lack of standardized test methods for performance evaluation as well as consensus standards analogous to those that have been developed for established medical imaging modalities (e.g., ultrasound). In this study, we address low contrast detectability, specifically, the ability of systems to differentiate between regions exhibiting small differences in scattering coefficient. Based on standard test methods for established medical imaging modalities, we have developed layered phantoms with well-characterized scattering properties in a biologically relevant range. The phantoms consisted of polydimethylsiloxane (PDMS) doped with varying concentrations of BaSO4 microparticles. Microfabrication processes were used to create layered and channel schemes. Two spectral domain OCT systems - a Fourier domain system at 855 nm and a swept-source device at 1310 nm - were then used to image the phantoms. The detectability of regions with different scattering levels was evaluated for each system by measuring pixel intensity differences. Confounding factors such as the inherent attenuation of the phantoms, signal intensity decay due to focusing and system roll-off were also encountered and addressed. Significant differences between systems were noted. The minimum differences in scattering coefficient that the Fourier domain and swept source systems could differentiate was 1.50 and 0.46 mm-1 respectively. Overall, this approach to evaluating low contrast detectability represents a key step towards the development of standard test methods to facilitate clinical translation of novel OCT systems.

Computational analysis of beveled-tip fiber probes for selective detection of subsurface fluorophores in turbid media

Optimization of illumination-collection parameters may lead to an improvement in the efficacy of fluorescence spectroscopy devices for minimally invasive disease detection. While device-tissue interface geometry has been shown to have a strong influence on the origin of detected fluorescence, prior studies have tended to focus on systems which deliver and collect light at approximately normal incidence to the tissue surface. Optical fibers with beveled surfaces enable the delivery and/or collection of light at a range of angles. This feature may make it possible to interrogate subsurface tissue regions with a greater degree of spatial selectivity. In this study, simulations were performed using a Monte Carlo model of fluorescent light propagation in order to estimate the behavior and potential limitations of this approach. The effect of tissue optical properties, illumination angle, collection angle and illumination-collection fiber separation distance were investigated. Results indicated that beveled fiber probes may significantly improve the ability of fiberoptic probes to selectively interrogate specific tissue regions or layers, however, the advantage over flat fibers is reduced as tissue attenuation increases.

Acute changes associated with electrode insertion measured with optical coherence microscopy

Despite advances in functional neural imaging, penetrating microelectrodes provide the most direct interface for the extraction of neural signals from the nervous system and are a critical component of many high degree-of-freedom braincomputer interface devices. Electrode insertion is a traumatic event that elicits a complex neuroinflammatory response. In this investigation we applied optical coherence microscopy (OCM), particularly optical coherence angiography (OCA), to characterize the immediate tissue response during microelectrode insertion. Microelectrodes of varying dimension and footprint (one-, two-, and four-shank) were inserted into mouse motor cortex beneath a window after craniotomy surgery. The microelectrodes were inserted in 3-4 steps at 15-20°, with approximately 250 μm linear insertion distance for each step. Before insertion and between each step, OCM datasets were collected, including for quantitative capillary velocimetry. A cohort of control animals without microelectrode insertion was also imaged over a similar time period (2-3 hours). Mechanical tissue deformation was observed in all the experimental animals. The quantitative angiography results varied across animals, and were not correlated with device dimensions. In some cases, localized flow drop-out was observed in a small region surrounding the electrode, while in other instances a global disruption in flow occurred, perhaps as a result of large vessel compression caused by mechanical pressure. OCM is a tool that can be used in various neurophotonics applications, including quantification of the neuroinflammatory response to penetrating electrode insertion.

Optical coherence microscopy of mouse cortical vasculature surrounding implanted electrodes

Optical coherence microscopy (OCM) provides real-time, in-vivo, three-dimensional, isotropic micron-resolution structural and functional characterization of tissue, cells, and other biological targets. Optical coherence angiography (OCA) also provides visualization and quantification of vascular flow via speckle-based or phase-resolved techniques. Performance assessment of neuroprosthetic systems, which allow direct thought control of limb prostheses, may be aided by OCA. In particular, there is a need to examine the underlying mechanisms of chronic functional degradation of implanted electrodes. Angiogenesis, capillary network remodeling, and changes in flow velocity are potential indicators of tissue changes that may be associated with waning electrode performance. The overall goal of this investigation is to quantify longitudinal changes in vascular morphology and capillary flow around neural electrodes chronically implanted in mice. We built a 1315-nm OCM system to image vessels in neocortical tissue in a cohort of mice. An optical window was implanted on the skull over the primary motor cortex above a penetrating shank-style microelectrode array. The mice were imaged bi-weekly to generate vascular maps of the region surrounding the implanted microelectrode array. Acute effects of window and electrode implantation included vessel dilation and profusion of vessels in the superficial layer of the cortex (0-200 μm). In deeper layers surrounding the electrode, no qualitative differences were seen in this early phase. These measurements establish a baseline vascular tissue response from the cortical window preparation and lay the ground work for future longitudinal studies to test the hypothesis that vascular changes will be associated with chronic electrode degradation.

Validation of a fiber optic-based UVA-VIS optical property measurement system

Tissue optical properties at ultraviolet A (UVA) and visible (VIS) wavelengths are needed to elucidate light-tissue interaction effects and optimize design parameters for spectroscopy-based neoplasia detection devices. Toward the goal of accurate and useful in vivo measurements, we have constructed and evaluated a system for optical property measurement at UVA-VIS wavelengths. Our approach involves a neural network-based inverse model calibrated with reflectance datasets simulated using a condensed Monte Carlo approach with absorption coefficients as high as 80 cm-1 and reduced scattering coefficients as high as 70 cm-1. Optical properties can be predicted with the inverse model based on spatially resolved reflectance measured with a fiberoptic probe. Theoretical evaluation of the inverse model was performed using simulated reflectance distributions at random optical properties. Experimental evaluation involved the use of tissue phantoms constructed from bovine hemoglobin and polystyrene microspheres. An average accuracy of ±1.0 cm-1 for absorption coefficients and ±2.7 cm-1 for reduced scattering coefficients was found from realistic phantoms at five UVA-VIS wavelengths. While accounting for the very high attenuation levels near the 415 nm Soret absorption band required some modifications, our findings provide evidence that the current approach produces useful data over a wide range of optical properties, and should be particularly useful for in vivo characterization of highly attenuating biological tissues.

Sensitivity and robustness of methods for analyzing time-resolved fluorescence measurements of layered biological tissue

Time-resolved fluorescence (TRF) measurements from layered biological tissue provide chemical and structural information which may be useful for imaging or single-point tissue diagnostics. While several techniques for analyzing TRF data have been proposed in the literature, a rigorous theoretical evaluation of these approaches has not been performed. In this study we have evaluated the sensitivity and robustness of four methods for analyzing TRF signals: biexponential deconvolution, single exponential deconvolution, Laguerre deconvolution, and direct peak width computation. Each of these analyses was performed on a large dataset of synthetic fluorescence decay curves. Each decay curve was generated by numerically convolving a pre-recorded nitrogen laser pulse with a biexponential decay based on fluorescence lifetimes of colonic mucosa. The relative contribution of each mucosal layer to the total TRF signal as well as the superficial layers inherent lifetime were varied so as to investigate sensitivity to morphological and biochemical changes representative of a neoplastic disease process. To evaluate robustness, pre-set levels of Gaussian-distributed noise were added to the convolved curves to achieve variations in the signal-to-noise ratio. The relative merits and pitfalls of each analytical method are discussed.