Browsing by Subject "Spectral imaging"

This thesis describes three computational optical systems and their underlying coding strategies. These codes are useful in a variety of optical imaging and spectroscopic applications. Two multichannel cameras are described. They both use a lenslet array to generate multiple copies of a scene on the detector. Digital processing combines the measured data into a single image. The visible system uses focal plane coding, and the long wave infrared (LWIR) system uses shift coding. With proper calibration, the multichannel interpolation results recover contrast for targets at frequencies beyond the aliasing limit of the individual subimages. This thesis also describes a LWIR imaging system that simultaneously measures four wavelength channels each with narrow bandwidth. In this system, lenses, aperture masks, and dispersive optics implement a spatially varying spectral code.

This dissertation describes two computational sensors that were used to demonstrate applications of generalized sampling of the optical field. The first sensor was an incoherent imaging system designed for compressive measurement of the power spectral density in the scene (spectral imaging). The other sensor was an interferometer used to compressively measure the mutual intensity of the optical field (coherence imaging) for imaging through turbulence. Each sensor made anisomorphic measurements of the optical signal of interest and digital post-processing of these measurements was required to recover the signal. The optical hardware and post-processing software were co-designed to permit acquisition of the signal of interest with sub-Nyquist rate sampling, given the prior information that the signal is sparse or compressible in some basis.

Compressive spectral imaging was achieved by a coded aperture snapshot spectral imager (CASSI), which used a coded aperture and a dispersive element to modulate the optical field and capture a 2D projection of the 3D spectral image of the scene in a snapshot. Prior information of the scene, such as piecewise smoothness of objects in the scene, could be enforced by numerical estimation algorithms to recover an estimate of the spectral image from the snapshot measurement.

Hypothesizing that turbulence between the scene and CASSI would introduce spectral diversity of the point spread function, CASSI's snapshot spectral imaging capability could be used to image objects in the scene through the turbulence. However, no turbulence-induced spectral diversity of the point spread function was observed experimentally. Thus, coherence functions, which are multi-dimensional functions that completely determine optical fields observed by intensity detectors, were considered. These functions have previously been used to image through turbulence after extensive and time-consuming sampling of such functions. Thus, compressive coherence imaging was attempted as an alternative means of imaging through turbulence.

Compressive coherence imaging was demonstrated by using a rotational shear interferometer to measure just a 2D subset of the 4D mutual intensity, a coherence function that captures the optical field correlation between all the pairs of points in the aperture. By imposing a sparsity constraint on the possible distribution of objects in the scene, both the object distribution and the isoplanatic phase distortion induced by the turbulence could be estimated with the small number of measurements made by the interferometer.

This dissertation presents methods and associated performance bounds for spectral image processing tasks such as reconstruction and target detection, which are useful in a variety of applications such as astronomical imaging, biomedical imaging and remote sensing. The key idea behind our spectral image processing methods is the fact that important information in a spectral image can often be captured by low-dimensional manifolds embedded in high-dimensional spectral data. Based on this key idea, our work focuses on the reconstruction of spectral images from photon-limited, and distorted observations.

This dissertation presents a partition-based, maximum penalized likelihood method that recovers spectral images from noisy observations and enjoys several useful properties; namely, it (a) adapts to spatial and spectral smoothness of the underlying spectral image, (b) is computationally efficient, (c) is near-minimax optimal over an anisotropic Holder-Besov function class, and (d) can be extended to inverse problem frameworks.

There are many applications where accurate localization of desired targets in a spectral image is more crucial than a complete reconstruction. Our work draws its inspiration from classical detection theory and compressed sensing to develop computationally efficient methods to detect targets from few projection measurements of each spectrum in the spectral image. Assuming the availability of a spectral dictionary of possible targets, the methods discussed in this work detect targets that either come from the spectral dictionary or otherwise. The theoretical performance bounds offer insight on the performance of our detectors as a function of the number of measurements, signal-to-noise ratio, background contamination and properties of the spectral dictionary.

A related problem is that of level set estimation where the goal is to detect the regions in an image where the underlying intensity function exceeds a threshold. This dissertation studies the problem of accurately extracting the level set of a function from indirect projection measurements without reconstructing the underlying function. Our partition-based set estimation method extracts the level set of proxy observations constructed from such projection measurements. The theoretical analysis presented in this work illustrates how the projection matrix, proxy construction and signal strength of the underlying function affect the estimation performance.

\abstract

Among women, breast cancer has the highest incidence rate worldwide and remains the leading cause of cancer-related deaths in developed countries. Women with stage I or II breast cancer are eligible for a surgical procedure known as breast conserving surgery (BCS) which seeks to optimize the amount of tissue removed.BCS involves removing the tumor and a minimally thin peripheral layer, or margin of disease-free tissue surrounding the tumor. While the procedure dramatically minimizes the amount of tissue removed, an unfortunate concomitant reality is that a significant percentage (around 25$\%$) of patients will be advised to return for a second surgery due to the discovery of malignant cells at the tissue margin edge, suggesting that it is likely not all of the malignant cells were removed in the initial procedure. The fact that margins are analyzed in histopathology post-operatively (in most cases) presents a substantial clinical burden that could be reduced if the surgeon was able to reliably assess suspicious areas intra-operatively.

The primary challenge in addressing this need stems from the need to resolve microscopic cellular morphology within a relatively tremendous amount of benign breast tissue. Many investigative optical tools seek to address this challenge, as the wavelength-dependent nature of light propagation within tissue can be used to assign optical signatures to tissue types derived from the relative tissue constituents.

Among the numerous techniques, quantitative diffuse reflectance spectroscopy (QDRS) is a well-established, comparatively simple technique that has been extensively validated in simulation, tissue-simulating phantoms, and various clinical contexts to robustly provide feature-specific optical signatures related to tissue morphology. We have leveraged QDRS in an evolution of several system formats to describe the morphological state of excised breast tissue based on the endogenous optical chromophores and scatterers within the breast, specifically, the amount of hemoglobin from blood, \betac~ in fat, as well as the size distribution and number density of scatterers.

We have employed multiple hardware embodiments of this technique related to the context of use. Each device leverages the same physical principles: The diffuse reflectance spectrum is measured using an imaging probe with multiple optical channels and is analyzed with a feature extraction algorithm based on a fast, scalable \mc~ model to quantitatively determine the absorption spectrum (\mualam) and reduced scattering spectrum (\musplam). The technology detects varying amounts of malignancy in the presence of benign tissue by quantifying the margin “landscape” as a cumulative distribution function (CDF) of the ratio of \betac~ concentration (absorber) and the wavelength averaged tissue scattering (\bscat), derived from \oprop, respectively. We have established through histopathological validation that the \bscat~ reports on the relative amount of adipose to collagen, glands, and fibrous content; decreased ratios are strongly associated with the presence of residual disease.

Local recurrence in BCS has a compelling association with residual disease, suggesting that QDRS could be used to reduce re-excision rates. The work presented here demonstrates a systematic approach in the development of a pragmatic and clinically viable QDRS imaging system. Two approaches are employed: a robust, research-grade 49-channel system is used to validate previous clinical findings and determine the optimal sampling resolution, and secondly, a low-cost, portable, miniature system based on annular photodiodes is developed and shown to be diagnostically comparable. These systems are accompanied by the development of a unique imaging platform that provides robust quality control and improved resolution, further improving the diagnostic capability. The diagnostic utility of the \bscat parameter is explored in a 100-patient clinical study. The potential for commercialization of the miniature system is informed through deployment of a replica system at a remote institution. Accessibility is improved through the design of a generic, object oriented software package that abstracts the individual hardware components.

The portability, accuracy, and manufacturability provide a realistically translatable path for integration into the clinical standard of care.

This study intends to validate the sensitivity and specificity of coded aperture coherent scatter spectral imaging (CACSSI) by comparison to clinical histological preparation and pathologic analysis methods currently used for the differentiation of normal and neoplastic breast tissues. A composite overlay of the CACSSI rendered image and pathologist interpreted, stained sections validate the ability of coherent scatter imaging to differentiate cancerous tissues from normal, healthy breast structures ex-vivo. Via comparison to the pathologist annotated slides, the CACSSI system may be further optimized to maximized sensitivity and specificity for differentiation of breast carcinomas.

Breast cancer surgical treatment options prove effective at treating breast cancer and reducing breast cancer death rates, prompting women to elect to surgically excise the tumor via a lumpectomy procedure. Despite women choosing lumpectomy over a mastectomy in 60% of cases, and despite the general effectiveness of the lumpectomy procedures, patient recall rates due to missed cancerous tissue are unfavorably high and variable at approximately 25% nationally. In addition, drawn-out processing times due to pathology assessment contribute to sub-optimal patient care and overly onerous costs and workload for hospitals. Therefore, it is the focus of this work to develop, evaluate, and refine a novel imaging modality to aid pathologists and pathologists’ assistants in assessing breast cancer via a more quantified means that would eventually lower the recall rates in breast cancer surgery.

Through previous work, we established a Coded Aperture Coherent Scatter Spectral Imaging (CACSSI) system, characterized several facets of the imaging setup, and evaluated its utility in breast cancer applications. Using Monte Carlo simulations, anthropomorphic breast phantoms, and human breast tissue specimens, we previously validated CACSSI’s utility in differentiating breast tissue types in a clinically relevant manner, which makes the system a promising candidate to act as a supplementary tool to implement in the pathology workflow. This work continues the previous research by applying and implementing the tissue classification ability within a short, clinically feasible timeframe (5-30 minutes) and demonstrating utility in a broader population of 12 patient-derived lumpectomy specimens. The work presented herein is broken into three subprojects: (1) Assessing various characterizations of the system (i.e. the background signal effects, the detector temperature-dependent response, the precision in consecutive scans, and the effect of formalin-fixation) to demonstrate its feasibility for the cancer detection/classification tasks; (2) Evaluating the accuracy of the system in a population of 12 excised breast tissue specimens while establishing and implementing the scan room procedures across multiples specimens; and (3) Utilizing a concurrently-developed classification scheme to more thoroughly compare the system’s fidelity and robustness against pathology-assessed outcomes, which currently serve as the clinical gold standard for breast cancer judgments.

The typical workflow included Surgical Pathology preparing the surgically excised specimens and indicating via palpation the location of the tumor. The specimen, with the preliminary tumor location marked, was then scanned in our imaging system, and spectral scatter signatures were obtained at multiple locations within the tissue. The resulting form factor spectra were then compared with reference spectra to classify the tissue as cancerous or non-cancerous (healthy). The tissue classification mapping was compared against the indicated tumor area or against pathology-stained microslides for verification of tumor diagnosis.

Formalin-fixation was found inconsequential for tissue classification, with fresh-to-formalin-fixed spectra correlations of 0.9782 and 0.9881 over 10 spot scans each for healthy and cancer tissue, respectively. The spatial resolution of the system was found to be 1.5 mm in the lateral direction and 5 mm along the beam path. Our CACSSI system was able to distinguish between cancerous and healthy areas in the tissue slices in a consistent manner, and the system was, on average, 82.93% accurate for the initial classification scheme and 83.70% accurate using a more quantitative classification scheme. Furthermore, we were able to achieve these results in a clinically relevant timeframe on the order of 30 minutes, integrating into the pathology workflow with minimal interruption. Aggregating these results CACSSI will continue to be developed for use as a clinical imaging tool in breast cancer assessment and other diagnostic purposes.