Optical imaging is an emerging technology towards non-invasive breast cancer diagnostics. In recent years, portable and patient comfortable hand-held optical imagers are developed towards two-dimensional (2D) tumor detections. However, these imagers are not capable of three-dimensional (3D) tomography because they cannot register the positional information of the hand-held probe onto the imaged tissue. A hand-held optical imager has been developed in our Optical Imaging Laboratory with 3D tomography capabilities, as demonstrated from tissue phantom studies. The overall goal of my dissertation is towards the translation of our imager to the clinical setting for 3D tomographic imaging in human breast tissues. A systematic experimental approach was designed and executed as follows: (i) fast 2D imaging, (ii) coregistered imaging, and (iii) 3D tomographic imaging studies. (i) Fast 2D imaging was initially demonstrated in tissue phantoms (1% Liposyn solution) and in vitro (minced chicken breast and 1% Liposyn). A 0.45 cm3 fluorescent target at 1:0 contrast ratio was detectable up to 2.5 cm deep. Fast 2D imaging experiments performed in vivo with healthy female subjects also detected a 0.45 cm3 fluorescent target superficially placed ~2.5 cm under the breast tissue. (ii) Coregistered imaging was automated and validated in phantoms with ~0.19 cm error in the probe’s positional information. Coregistration also improved the target depth detection to 3.5 cm, from multi-location imaging approach. Coregistered imaging was further validated in-vivo, although the error in probe’s positional information increased to ~0.9 cm (subject to soft tissue deformation and movement). (iii) Three-dimensional tomography studies were successfully demonstrated in vitro using 0.45 cm3 fluorescence targets. The feasibility of 3D tomography was demonstrated for the first time in breast tissues using the hand-held optical imager, wherein a 0.45 cm3 fluorescent target (superficially placed) was recovered along with artifacts. Diffuse optical imaging studies were performed in two breast cancer patients with invasive ductal carcinoma. The images showed greater absorption at the tumor cites (as observed from x-ray mammography, ultrasound, and/or MRI). In summary, my dissertation demonstrated the potential of a hand-held optical imager towards 2D breast tumor detection and 3D breast tomography, holding a promise for extensive clinical translational efforts.

Diffuse optical tomography uses near infrared (NIR) light as the probing media to recover the distributions of tissue optical properties. It has a potential to become an adjunct imaging modality for breast and brain imaging, that is capable of providing functional information of the tissue under investigation. As NIR light propagation in the tissue is dominated by scattering, the image reconstruction problem (inverse problem) tends to be non-linear and ill-posed, requiring usage of advanced computational methods to compensate this. Traditional image reconstruction methods in diffuse optical tomography employ l2 –norm based regularization, which is known to remove high frequency noises in the re-constructed images and make them appear smooth. The recovered contrast in the reconstructed image in these type of methods are typically dependent on the iterative nature of the method employed, in which the non-linear iterative technique is known to perform better in comparison to linear techniques. The usage of non-linear iterative techniques in the real-time, especially in dynamical imaging, becomes prohibitive due to the computational complexity associated with them. In the rapid dynamic diffuse optical imaging, assumption of a linear dependency in the solutions between successive frames results in a linear inverse problem. This new frame work along with the l1–norm based regularization can provide better robustness to noise and results in a better contrast recovery compared to conventional l2 –based techniques. Moreover, it is shown that the proposed l1-based technique is computationally efficient compared to its counterpart(l2 –based one). The proposed framework requires a reasonably close estimate of the actual solution for the initial frame and any suboptimal estimate leads to erroneous reconstruction results for the subsequent frames. Modern diffuse optical imaging systems are multi-modal in nature, where diffuse optical imaging is combined with traditional imaging modalities such as MRI, CT, and Ultrasound. A novel approach that can more effectively use the structural information provided by the traditional imaging modalities in these scenarios is introduced, which is based on prior image constrained- l1 minimization scheme. This method has been motivated by the recent progress in the sparse image reconstruction techniques. It is shown that the- l1 based frame work is more effective in terms of localizing the tumor region and recovering the optical property values both in numerical and gelatin phantom cases compared to the traditional methods that use structural information.

L’étude du cerveau humain est un domaine en plein essor et les techniques non-invasives de l’étudier sont très prometteuses. Afin de l’étudier de manière non-invasive, notre laboratoire utilise principalement l’imagerie par résonance magnétique fonctionnelle (IRMf) et l’imagerie optique diffuse (IOD) continue pour mesurer et localiser l’activité cérébrale induite par une tâche visuelle, cognitive ou motrice. Le signal de ces deux techniques repose, entre autres, sur les concentrations d’hémoglobine cérébrale à cause du couplage qui existe entre l’activité neuronale et le flux sanguin local dans le cerveau. Pour être en mesure de comparer les deux signaux (et éventuellement calibrer le signal d’IRMf par l’IOD), où chaque signal est relatif à son propre niveau de base physiologique inconnu, une nouvelle technique ayant la capacité de mesurer le niveau de base physiologique est nécessaire. Cette nouvelle technique est l’IOD résolue temporellement qui permet d’estimer les concentrations d’hémoglobine cérébrale. Ce nouveau système permet donc de quantifier le niveau de base physiologique en termes de concentrations d’hémoglobine cérébrale absolue. L’objectif général de ma maîtrise était de développer un tel système afin de l’utiliser dans une large étude portant sur la condition cardiovasculaire, le vieillissement, la neuroimagerie ainsi que les performances cognitives. Il a fallu tout d’abord construire le système, le caractériser puis valider les résultats avant de pouvoir l’utiliser sur les sujets de recherche. La validation s’est premièrement réalisée sur des fantômes homogènes ainsi qu’hétérogènes (deux couches) qui ont été développés. La validation des concentrations d’hémoglobine cérébrale a été réalisée via une tâche cognitive et appuyée par les tests sanguins des sujets de recherche. Finalement, on présente les résultats obtenus dans une large étude employant le système d’IOD résolue temporellement en se concentrant sur les différences reliées au vieillissement. / Our understanding of the functional organization of the human brain has been greatly influenced by the development of new medical imaging techniques. Pr. Hoge’s research has focused on the use of functional magnetic resonance imaging (fMRI) and continuous diffuse optical imaging (DOI) for non-invasive localization and quantification of brain activity associated with behavioral stimuli or tasks (e.g. cognitive, motor or visual). The respective signals of both techniques are based on cerebral haemoglobin concentrations because of the coupling that exists between neuronal activity and cerebral blood flow. Relating BOLD fMRI signals with those acquired using DOI has been complicated by the fact that fMRI yields fractional change values, while the majority of DOI methods have provided absolute changes from an unknown baseline. To address this, we adopted a newer technique known as time-resolved DOI, which allows absolute quantification of cerebral haemoglobin concentrations. Time-resolved DOI thus has the capacity to quantify the subject’s resting hemoglobin concentrations in absolute micromolar units. The main objective of my masters’ project was to implement and optimize a time-resolved DOI system for use in a large study exploring the links between cardiovascular fitness, aging, neuroimaging markers, and cognitive performance. In this thesis we describe the fabrication of the system, followed by its characterisation and validation using solid optical phantoms (homogeneous and heterogeneous) developed for this purpose. Haemoglobin concentrations obtained non-invasively with the system are validated against blood draws, while the sensitivity to variations in concentration are assessed during a cognitive task. Finally, we present the results of a large study in which the time-resolved DOI system was used to characterize age-related vascular changes in the brain.

L’étude du cerveau humain est un domaine en plein essor et les techniques non-invasives de l’étudier sont très prometteuses. Afin de l’étudier de manière non-invasive, notre laboratoire utilise principalement l’imagerie par résonance magnétique fonctionnelle (IRMf) et l’imagerie optique diffuse (IOD) continue pour mesurer et localiser l’activité cérébrale induite par une tâche visuelle, cognitive ou motrice. Le signal de ces deux techniques repose, entre autres, sur les concentrations d’hémoglobine cérébrale à cause du couplage qui existe entre l’activité neuronale et le flux sanguin local dans le cerveau. Pour être en mesure de comparer les deux signaux (et éventuellement calibrer le signal d’IRMf par l’IOD), où chaque signal est relatif à son propre niveau de base physiologique inconnu, une nouvelle technique ayant la capacité de mesurer le niveau de base physiologique est nécessaire. Cette nouvelle technique est l’IOD résolue temporellement qui permet d’estimer les concentrations d’hémoglobine cérébrale. Ce nouveau système permet donc de quantifier le niveau de base physiologique en termes de concentrations d’hémoglobine cérébrale absolue. L’objectif général de ma maîtrise était de développer un tel système afin de l’utiliser dans une large étude portant sur la condition cardiovasculaire, le vieillissement, la neuroimagerie ainsi que les performances cognitives. Il a fallu tout d’abord construire le système, le caractériser puis valider les résultats avant de pouvoir l’utiliser sur les sujets de recherche. La validation s’est premièrement réalisée sur des fantômes homogènes ainsi qu’hétérogènes (deux couches) qui ont été développés. La validation des concentrations d’hémoglobine cérébrale a été réalisée via une tâche cognitive et appuyée par les tests sanguins des sujets de recherche. Finalement, on présente les résultats obtenus dans une large étude employant le système d’IOD résolue temporellement en se concentrant sur les différences reliées au vieillissement. / Our understanding of the functional organization of the human brain has been greatly influenced by the development of new medical imaging techniques. Pr. Hoge’s research has focused on the use of functional magnetic resonance imaging (fMRI) and continuous diffuse optical imaging (DOI) for non-invasive localization and quantification of brain activity associated with behavioral stimuli or tasks (e.g. cognitive, motor or visual). The respective signals of both techniques are based on cerebral haemoglobin concentrations because of the coupling that exists between neuronal activity and cerebral blood flow. Relating BOLD fMRI signals with those acquired using DOI has been complicated by the fact that fMRI yields fractional change values, while the majority of DOI methods have provided absolute changes from an unknown baseline. To address this, we adopted a newer technique known as time-resolved DOI, which allows absolute quantification of cerebral haemoglobin concentrations. Time-resolved DOI thus has the capacity to quantify the subject’s resting hemoglobin concentrations in absolute micromolar units. The main objective of my masters’ project was to implement and optimize a time-resolved DOI system for use in a large study exploring the links between cardiovascular fitness, aging, neuroimaging markers, and cognitive performance. In this thesis we describe the fabrication of the system, followed by its characterisation and validation using solid optical phantoms (homogeneous and heterogeneous) developed for this purpose. Haemoglobin concentrations obtained non-invasively with the system are validated against blood draws, while the sensitivity to variations in concentration are assessed during a cognitive task. Finally, we present the results of a large study in which the time-resolved DOI system was used to characterize age-related vascular changes in the brain.

Conventional medical imaging technologies for cancer diagnosis utilize fixed geometric configuration of the source and the detector to image the target. In this dissertation, we hypothesize that dynamic utilization of source and detector geometry will lead to better performance of medical imaging devices. Interrogating a target in a three dimensional space requires cooperation and coordination between the source and detector positions. The goal of this dissertation is to develop a dynamic imaging method, which will improve the tumor characterization performance, and provide a control scheme appropriate for the dynamic interrogation. This dissertation proposes a bimodal dynamic imaging (BDI) method for improving tumor characterization and a hybrid hierarchical statistical control scheme for the autonomous control of the sources and detectors. The tactile imaging sensor has high specificity but low sensitivity in tumor characterization. The spectral sensor has high sensitivity but low specificity. The BDI system integrates the tactile sensing and the spectral sensing modalities with the capability of dynamic positioning of the source and detector to determine the mechanical and spectral properties of a tumor. The tactile sensing can estimate the mechanical properties of the tumor, such as size, depth, and elastic modulus, while the spectral sensing can determine the absorption coefficient of the tumor through diffuse optical imaging. These properties help us characterize the tumor, and differentiate cancerous tissues from healthy tissues. We designed and experimentally evaluated the BDI system for estimating the size, depth, elastic modulus, and absorption coefficient of embedded inclusions. The system performance in characterizing mechanical properties was then compared to that of the tactile imaging sensor. The proposed BDI method was experimentally validated using fabricated bimodal phantom. The experimental results showed that the tactile imaging system (TIS) estimated the tumor phantom size with 7.23% error; BDI measured the size with 0.8% error. The TIS depth estimation error was 41.83%; BDI reduced the depth measurement error to 20.00%. The TIS elastic modulus estimation error was 96.80%; the BDI method showed 74.79% error. Additionally, BDI estimated the absorption coefficient with 14%-25% estimation error. For further improvement the system performance, this bimodal imaging system is implemented on a dual-arm robot, Baxter, where the laser source and the tactile imaging sensors were mounted on the end-effectors. Each arm of Baxter robot has seven Degree-of- Freedom. This provides more flexibility in terms of interrogating the target compared to the fixed geometric configuration. We devised a hybrid statistical controller for maneuvering the source and the detector of the system. In this control architecture, a high-level supervisory controller was used for the functions at a higher level for coordinating two arms. At lower level, a full-state feedback statistical controller was used to facilitate the minimum position variation. A linear model for the dual-arm Baxter robot was derived for testing the proposed architecture. We performed the simulations of hybrid hierarchical statistical controller on the Baxter model for trajectory tracking. The simulation studies demonstrated accurate sequential task execution for the bimodal dynamic imaging system using a hybrid hierarchical statistical control. / Electrical and Computer Engineering

New advances in computational modeling and instrumentation in the past decade has enabled the use of electromagnetic radiation for non-invasive monitoring of the physio-logical state of biological tissues. The near infrared (NIR) light having the wavelength range of 600 nm -1000 nm has been the main contender in these emerging molecular imaging modalities. Assessment of accurate pathological condition of the tissue under investigation relies on the contrast in the molecular images, where the endogenous contrast may not be sufficient in these scenarios. The fluorescence (exogenous) contrast agents have been deployed to overcome these difficulties, where the preferential uptake by the tumor vasculature leads to high contrast,making this modality one of the biggest contenders in small-animal and soft-tissue molecular imaging modalities. In Fluorescence diffuse optical spectroscopy/imaging, this exogenous drug is excited by NIR laser light causing the emission of the fluorescence light. The emitted fluorescence light is typically dependent on the life time and concentration of the exogenous drug coupled with physiology associated with the tissue under investigation. As there is an excitation and emission of the light,the underlying physics of the problem is described by a coupled diffusion equations. These coupled diffusion equations are typically solved by advanced numerical methods, which tend to be computationally demanding. In this work, analytical solutions for these coupled partial differential equations (PDEs) for the regular geometries for both time-domain and frequency-domain cases were developed. Till now, the existing literature has not dealt with all regular geometries and derived analytical solutions were only for couple of geometries. Here a universally acceptable generic solution was developed based on Green’s function approach that is applicable to any regular geometry. Using this, the analytical solutions for the regular geometries that is encountered in diffuse fluorescence spectroscopy/imaging were obtained. These solutions can play an important role in determining the bulk fluorescence properties of the tissue, which could act as good initial guesses for the advanced image reconstruction techniques and/or can also facilitate the calibration of experimental fluorescence data by removing biases and source-detector variations. In the second part of this work, the developed analytical models for regular geometries were validated through comparison with the established numerical models that are traditionally used in the diffuse fluorescence spectroscopy/imaging. This comparison not only validated the developed analytical models, but also showed that analytical models are capable of providing bulk fluorescence properties with at least one order of magnitude less computational cost compared to the highly optimized traditional numerical models.

Computational imaging has been playing an important role in understanding and analysing the captured images. Both image segmentation and restoration has been in-tegral parts of computational imaging. The studies performed in this thesis has been focussed toward developing novel algorithms for image segmentation and restoration. Study related to usage of Morozov Discrepancy Principle in Di use Optical Imaging was also presented here to show that hyper parameter selection could be performed with ease. The Laplacian of Gaussian (LoG) and Canny operators use Gaussian smoothing be-fore applying the derivative operator for edge detection in real images. The LoG kernel was based on second derivative and is highly sensitive to noise when compared to the Canny edge detector. A new edge detection kernel, called as Helmholtz of Gaussian (HoG), which provides higher di suavity is developed in this thesis and it was shown that it is more robust to noise. The formulation of the developed HoG kernel is similar to LoG. It was also shown both theoretically and experimentally that LoG is a special case of HoG. This kernel when used as an edge detector exhibited superior performance compared to LoG, Canny and wavelet based edge detector for the standard test cases both in one- and two-dimensions. The linear inverse problem encountered in restoration of blurred noisy images is typically solved via Tikhonov minimization. The outcome (restored image) of such min-imitation is highly dependent on the choice of regularization parameter. In the absence of prior information about the noise levels in the blurred image, ending this regular-inaction/hyper parameter in an automated way becomes extremely challenging. The available methods like Generalized Cross Validation (GCV) may not yield optimal re-salts in all cases. A novel method that relies on minimal residual method for ending the regularization parameter automatically was proposed here and was systematically compared with the GCV method. It was shown that the proposed method performance was superior to the GCV method in providing high quality restored images in cases where the noise levels are high Di use optical tomography uses near infrared (NIR) light as the probing media to recover the distributions of tissue optical properties with an ability to provide functional information of the tissue under investigation. As NIR light propagation in the tissue is dominated by scattering, the image reconstruction problem (inverse problem) is non-linear and ill-posed, requiring usage of advanced computational methods to compensate this. An automated method for selection of regularization/hyper parameter that incorporates Morozov discrepancy principle(MDP) into the Tikhonov method was proposed and shown to be a promising method for the dynamic Di use Optical Tomography.

Diffuse optical tomography (DOT) is one of the promising imaging modalities that pro- vides functional information of the soft biological tissues in-vivo, such as breast and brain tissues. The near infrared (NIR) light (600-1000 nm) is the interrogating radiation, which is typically delivered and collected using fiber bundles placed on the boundary of the tissue. The internal optical property distribution is estimated via model-based image reconstruction algorithm using these limited boundary measurements. Image reconstruction problem in DOT is known to be non-linear, ill-posed, and some times under-determined due to the multiple scattering of NIR light in the tissue. Solving this inverse problem requires regularization to obtain meaningful results, with Tikhonov-type regularization being the most popular one. The choice of the regularization parameter dictates the reconstructed optical image quality and is typically chosen empirically or based on prior experience. An automated method for optimal selection of regularization parameter that is based on regularized minimal residual method (MRM) is proposed and is compared with the traditional generalized cross-validation method. The results obtained using numerical and gelatin phantom data indicate that the MRM-based method is capable of providing the optimal regularization parameter. A new approach that can easily incorporate any generic penalty function into the diffuse optical tomographic image reconstruction is introduced to show the utility of non-quadratic penalty functions. The penalty functions that were used include, quadratic (`2), absolute (`1), Cauchy, and Geman-McClure. The regularization parameter in each of these cases were obtained automatically using the generalized cross-validation (GCV) method. The reconstruction results were systematically compared with each other via utilization of quantitative metrics, such as relative error and Pearson correlation. The reconstruction results indicate that while quadratic penalty may be able to provide better separation between two closely spaced targets, it's contrast recovery capability is limited and the sparseness promoting penalties, such as `1, Cauchy, Geman-McClure have better utility in reconstructing high-contrast and complex-shaped targets with Geman-McClure penalty being the most optimal one. Effective usage of image guidance by incorporating the refractive index (RI) variation in computational modeling of light propagation in tissue is investigated to assess its impact on optical-property estimation. With the aid of realistic patient breast three-dimensional models, the variation in RI for different regions of tissue under investigation is shown to influence the estimation of optical properties in image-guided diffuse optical tomography (IG-DOT) using numerical simulations. It is also shown that by assuming identical RI for all regions of tissue would lead to erroneous estimation of optical properties. The a priori knowledge of the RI for the segmented regions of tissue in IG-DOT, which is difficult to obtain for the in vivo cases, leads to more accurate estimates of optical properties. Even inclusion of approximated RI values, obtained from the literature, for the regions of tissue resulted in better estimates of optical properties, with values comparable to that of having the correct knowledge of RI for different regions of tissue. Image reconstruction in IG-DOT procedure involves reduction of the number of optical parameters to be reconstructed equal to the number of distinct regions identified in the structural information provided by the traditional imaging modality. This makes the image reconstruction problem to be well-determined compared to traditional under- determined case. Still, the methods that are deployed in this case are same as the one used for traditional diffuse optical image reconstruction, which involves regularization term as well as computation of the Jacobian. A gradient-free Nelder-Mead simplex method was proposed here to perform the image reconstruction procedure and shown to be providing solutions that are closely matching with ones obtained using established methods. The proposed method also has the distinctive advantage of being more efficient due to being regularization free, involving only repeated forward calculations.

Diffuse optical tomography is a promising imaging modality that provides functional information of the soft biological tissues, with prime imaging applications including breast and brain tissue in-vivo. This modality uses near infrared light( 600nm-900nm) as the probing media, giving an advantage of being non-ionizing imaging modality. The image reconstruction problem in diffuse optical tomography is typically posed as a least-squares problem that minimizes the difference between experimental and modeled data with respect to optical properties. This problem is non-linear and ill-posed, due to multiple scattering of the near infrared light in the biological tissues, leading to infinitely many possible solutions. The traditional methods employ a regularization term to constrain the solution space as well as stabilize the solution, with Tikhonov type regularization being the most popular one. The choice of this regularization parameter, also known as hyper parameter, dictates the reconstructed optical image quality and is typically chosen empirically or based on prior experience. In this thesis, a simple back projection type image reconstruction algorithm is taken up, as they are known to provide computationally efficient solution compared to regularized solutions. In these algorithms, the hyper parameter becomes equivalent to filter factor and choice of which is typically dependent on the sampling interval used for acquiring data in each projection and the angle of projection. Determining these parameters for diffuse optical tomography is not so straightforward and requires usage of advanced computational models. In this thesis, a computationally efficient simplex Method based optimization scheme for automatically finding this filter factor is proposed and its performances is evaluated through numerical and experimental phantom data. As back projection type algorithms are approximations to traditional methods, the absolute quantitative accuracy of the reconstructed optical properties is poor .In scenarios, like dynamic imaging, where the emphasis is on recovering relative difference in the optical properties, these algorithms are effective in comparison to traditional methods, with an added advantage being highly computationally efficient. In the second part of this thesis, this hyper parameter choice for traditional Tikhonov type regularization is attempted with the help of Least-Squares QR-decompisition (LSQR) method. The established techniques that enable the automated choice of hyper parameters include Generalized Cross-Validation(GCV) and regularized Minimal Residual Method(MRM), where both of them come with higher over head of computation time, making it prohibitive to be used in the real-time. The proposed LSQR algorithm uses bidiagonalization of the system matrix to result in less computational cost. The proposed LSQR-based algorithm for automated choice of hyper parameter is compared with MRM methods and is proven to be computationally optimal technique through numerical and experimental phantom cases.