Search

Medical systems and assistive technologies, human-computer interaction, human-robot interaction, industrial automation, virtual environment control, sign language translation, crisis and disaster management, entertainment and computer games, and so on all use RGB cameras for hand gesture recognition. However, their performance is limited especially in low-light conditions. In this paper, we propose a robust hand gesture recognition system based on high-resolution thermal imaging that is light-independent. A dataset of 14,400 thermal hand gestures is constructed, separated into two color tones. We also propose using a deep CNN to classify high-resolution hand gestures accurately. The proposed models were also tested on Raspberry Pi 4 and Nvidia AGX edge computing devices, and the results were compared to the benchmark models. The model also achieves an accuracy of 98.81% and an inference time of 75.138 ms on Nvidia Jetson AGX. In contrast to hand gesture recognition systems based on RGB cameras, which have limited performance in the dark-light conditions, the proposed system based on reliable high resolution thermal images is well-suited to a wide range of applications.

In this paper, we present a novel multiclass-target classification method for mmWave frequency modulated continuous wave (FMCW) radar operating in the frequency range of 77 - 81 GHz, based on custom range-angle heatmaps and machine learning tools. The elevation field of view (FoV) is increased by orienting the Radar antennas in elevation. In this orientation, the radar focuses the beam in elevation to improve the elevation FoV. The azimuth FoV is improved by mechanically rotating the Radar horizontally, which has antenna elements oriented in the elevation direction. The data from the Radar measurements obtained by mechanical rotation of the Radar in Azimuth are used to generate a range-angle heatmap. The measurements are taken in a variety of real-world scenarios with various objects such as humans, a car, and an unmanned aerial vehicle (UAV), also known as a drone. The proposed technique achieves accuracy of 97.6 % and 99.6 % for classifying the UAV and humans, respectively, and accuracy of 98.1 % for classifying the car from the range-angle FoV heatmap. Such a Radar classification technique will be extremely useful for a wide range of applications in cost-effective and dependable autonomous systems, including ground station traffic monitoring and surveillance, as well as control systems for both on-ground and aerial vehicles.

Digital rock is an emerging area of rock physics, which involves scanning reservoir rocks using X-ray micro computed tomography (XCT) scanners and using it for various petrophysical computations and evaluations. The acquired micro CT projections are used to reconstruct the X-ray attenuation maps of the rock. The image reconstruction problem can be solved by utilization of analytical (such as Feldkamp–Davis–Kress (FDK) algorithm) or iterative methods. Analytical schemes are typically computationally more efficient and hence preferred for large datasets such as digital rocks. Iterative schemes like maximum likelihood expectation maximization (MLEM) are known to generate accurate image representation over analytical scheme in limited data (and/or noisy) situations, however iterative schemes are computationally expensive. In this work, we have parallelized the forward and inverse operators used in the MLEM algorithm on multiple graphics processing units (multi-GPU) platforms. The multi-GPU implementation involves dividing the rock volumes and detector geometry into smaller modules (along with overlap regions). Each of the module was passed onto different GPU to enable computation of forward and inverse operations. We observed an acceleration of $$\sim 30$$ ∼ 30 times using our multi-GPU approach compared to the multi-core CPU implementation. Further multi-GPU based MLEM obtained superior reconstruction compared to traditional FDK algorithm.

In this article, we present a novel localization and activity classification method for aerial vehicle using mmWave frequency modulated continuous wave (FMCW) Radar. The localization and activity classification for aerial vehicle enables the utilization of mmWave Radars in security surveillance and privacy monitoring applications. In the proposed method, Radar’s antennas are oriented vertically to measure the elevation angle of arrival of the aerial vehicle from ground station. The height of the aerial vehicle and horizontal distance of the aerial vehicle from Radar station on ground are estimated using the measured radial range and the elevation angle of arrival. The aerial vehicle’s activity is classified using machine learning methods on micro-Doppler signatures extracted from Radar measurements taken in an outdoor environment. To evaluate performance, various light weight classification models such as logistic regression, support vector machine (SVM), Light gradient boosting machine (GBM), and a custom lightweight convolutional neural network (CNN) are investigated. Based on the results, the logistic regression, SVM, and Light GBM achieve an accuracy of 93%. Furthermore, the custom lightweight CNN can achieve activity classification accuracy of 95%. The performance of the proposed lightweight CNN is also compared with the pre-trained models (VGG16, VGG19, ResNet50, ResNet101, and InceptionResNet). The proposed lightweight CNN suits best for embedded and/or edge computing devices.

The sign language digits based on hand gestures have been utilized in various applications such as human-computer interaction, robotics, health and medical systems, health assistive technologies, automotive user interfaces, crisis management and disaster relief, entertainment, and contactless communication in smart devices. The color and depth cameras are commonly deployed for hand gesture recognition, but the robust classification of hand gestures under varying illumination is still a challenging task. This work presents the design and deployment of a complete end-to-end edge computing system that can accurately provide the classification of hand gestures captured from thermal images. A thermal dataset of 3200 images was created with each sign language digit having 320 thermal images. The solution presented here utilizes live images taken from a low-resolution thermal camera of $32\times 32$ pixels, feeding into a novel light weight deep learning model based on bottleneck motivated from deep residual learning for classification of hand gestures. The edge computing system presented here utilizes Raspberry pi with a thermal camera making it highly portable. The designed system achieves an accuracy of 99.52% on the test data set with an added advantage of accuracy being invariable to background lighting conditions as it is based on thermal imaging.

PURPOSE To propose a generic deep learning based medical image reconstruction model (named as SpiNet) that can enforce any Schatten p-norm regularization with 0

The reconstruction methods for solving the ill-posed inverse problem of photoacoustic tomography with limited noisy data are iterative in nature to provide accurate solutions. These methods performance is highly affected by the noise level in the photoacoustic data. A singular value decomposition (SVD) based plug and play priors method for solving photoacoustic inverse problem was proposed in this work to provide robustness to noise in the data. The method was shown to be superior as compared to total variation regularization, basis pursuit deconvolution and Lanczos Tikhonov based regularization and provided improved performance in case of noisy data. The numerical and experimental cases show that the improvement can be as high as 8.1 dB in signal to noise ratio of the reconstructed image and 67.98% in root mean square error in comparison to the state of the art methods. Published version Department of Science and Technology, Ministry of Science and Technology, India (DST-ICPS (T-851)). PKY acknowledges the DST-ICPS cluster funding (T-851) for the data science program.

Purpose To compare the performance of iterative direct and indirect parametric reconstruction methods with indirect deep learning-based reconstruction methods in estimating tracer-kinetic parameters from highly undersampled DCE-MR Imaging breast data and provide a systematic comparison of the same. Methods Estimation of tracer-kinetic parameters using indirect methods from undersampled data requires to reconstruct the anatomical images initially by solving an inverse problem. This reconstructed images gets utilized in turn to estimate the tracer-kinetic parameters. In direct estimation, the parameters are estimated without reconstructing the anatomical images. Both problems are ill-posed and are typically solved using prior-based regularization or using deep learning. In this study, for indirect estimation, two deep learning-based reconstruction frameworks namely, ISTA-Net+ and MODL, were utilized. For direct and indirect parametric estimation, sparsity inducing priors (L1 and Total-Variation) and limited memory Broyden-Fletcher-Goldfarb-Shanno algorithm as solver was deployed. The performance of these techniques were compared systematically in estimation of vascular permeability ( K trans ) from undersampled DCE-MRI breast data using Patlak as pharmaco-kinetic model. The experiments involved retrospective undersampling of the data 20×, 50×, and 100× and compared the results using PSNR, nRMSE, SSIM, and Xydeas metrics. The K trans maps estimated from fully sampled data were utilized as ground truth. The developed code was made available as https://github.com/Medical-Imaging-Group/DCE-MRI-Compare open-source for enthusiastic users. Results The reconstruction methods performance was evaluated using ten patients breast data (five patients each for training and testing). Consistent with other studies, the results indicate that direct parametric reconstruction methods provide improved performance compared to the indirect parameteric reconstruction methods. The results also indicate that for 20× undersampling, deep learning-based methods performs better or at par with direct estimation in terms of PSNR, SSIM, and nRMSE. However, for higher undersampling rates (50× and 100×) direct estimation performs better in all metrics. For all undersampling rates, direct reconstruction performed better in terms of Xydeas metric, which indicated fidelity in magnitude and orientation of edges. Conclusion Deep learning-based indirect techniques perform at par with direct estimation techniques for lower undersampling rates in the breast DCE-MR imaging. At higher undersampling rates, they are not able to provide much needed generalization. Direct estimation techniques are able to provide more accurate results than both deep learning- and parametric-based indirect methods in these high undersampling scenarios.

This work presents two variants of Born-series methods, for the first time, providing a numerical solution to the time-independent photoacoustic wave equation. These methods are effective in provid...

Photoacoustic tomography involves reconstructing the initial pressure rise distribution from the measured acoustic boundary data. The recovery of the initial pressure rise distribution tends to be an ill-posed problem in presence of noise and when limited independent data is available, necessitating regularization. The standard regularization schemes include, Tikhonov, `1-norm, and total-variation. These regularization schemes weigh the singular values equally irrespective of the noise level present in the data. This work introduces a fractional framework, to weigh the singular values with respect to a fractional power. This fractional framework was implemented for Tikhonov, `1-norm, and total-variation regularization schemes. Moreover, an automated method for choosing the fractional power was also proposed. It was shown theoretically and with numerical experiments that the fractional power is inversely related to the data noise level for fractional Tikhonov scheme. The fractional framework outperforms the standard regularization schemes, Tikhonov, `1-norm, and total-variation by 54% in numerical simulations, experimental phantoms and in vivo rat data in terms of observed contrast/signal-to-noise-ratio of the reconstructed images. NMRC (Natl Medical Research Council, S’pore) Accepted version

The low-dose computed tomography (CT) perfusion data has low signal-to-noise ratio resulting in derived perfusion maps being noisy. These low-quality maps typically requires a denoising step to improve their utility in real-time. The existing methods, including state-of-the-art online sparse perfusion deconvolution (SPD), largely relies on the convolutional model that may not be applicable in all cases of brain perfusion. In this paper, a denoising convolutional neural network (DCNN) was proposed that relies only on computed perfusion maps for performing the denoising step. The network was trained with a large number of low-dose digital brain phantom perfusion maps to provide an approximation to the corresponding high-dose perfusion maps. The batch normalization coupled with residual learning makes the trained model invariant to the dynamic range of the input low-dose perfusion maps. The denoising of the raw-data using the convolutional neural network was also attempted here and shown to have limited applicability in the low-dose CT perfusion cases. The digital perfusion phantom as well as in-vivo results indicate that the proposed DCNN applied in the derived map domain provides superior improvement compared to the online SPD with an added advantage of being computationally efficient.

The recovery of the initial pressure rise distribution tends to be an ill-posed problem in the presence of noise and when limited independent data is available, necessitating regularization. The standard regularization schemes include Tikhonov, l1 -norm, and total-variation. These regularization schemes weigh the singular values equally irrespective of the noise level present in the data. This work introduces a fractional framework to weigh the singular values with respect to a fractional power. This fractional framework was implemented for Tikhonov, l1-norm, and total-variation regularization schemes. The fractional framework outperformed the standard regularization schemes by 54% in terms of observed contrast/signal-to-noise-ratio.

Radars with mmWave frequency modulated continuous wave (FMCW) technology accurately estimate the range and velocity of targets in their field of view (FoV). The targeted angle of arrival (AoA) estimation can be improved by increasing receiving antennas or by using multiple-input multiple-output (MIMO). However, obtaining target features such as target type remains challenging. In this paper, we present a novel target classification method based on machine learning and features extracted from a range fast Fourier transform (FFT) profile by using mmWave FMCW radars operating in the frequency range of 77–81 GHz. The measurements are carried out in a variety of realistic situations, including pedestrian, automotive, and unmanned aerial vehicle (UAV) (also known as drone). Peak, width, area, variance, and range are collected from range FFT profile peaks and fed into a machine learning model. In order to evaluate the performance, various light weight classification machine learning models such as logistic regression, Naive Bayes, support vector machine (SVM), and lightweight gradient boosting machine (GBM) are used. We demonstrate our findings by using outdoor measurements and achieve a classification accuracy of 95.6% by using LightGBM. The proposed method will be extremely useful in a wide range of applications, including cost-effective and dependable ground station traffic management and control systems for autonomous operations, and advanced driver-assistance systems (ADAS). The presented classification technique extends the potential of mmWave FMCW radar beyond the detection of range, velocity, and AoA to classification. mmWave FMCW radars will be more robust in computer vision, visual perception, and fully autonomous ground control and traffic management cyber-physical systems as a result of the added new feature.

Hand gestures based sign language digits have several contactless applications. Applications include communication for impaired people, such as elderly and disabled people, health-care applications, automotive user interfaces, and security and surveillance. This work presents the design and implementation of a complete end-to-end deep learning based edge computing system that can verify a user contactlessly using &lsquo, authentication code&rsquo, The &lsquo, is an &lsquo, n&rsquo, digit numeric code and the digits are hand gestures of sign language digits. We propose a memory-efficient deep learning model to classify the hand gestures of the sign language digits. The proposed deep learning model is based on the bottleneck module which is inspired by the deep residual networks. The model achieves classification accuracy of 99.1% on the publicly available sign language digits dataset. The model is deployed on a Raspberry pi 4 Model B edge computing system to serve as an edge device for user verification. The edge computing system consists of two steps, it first takes input from the camera attached to it in real-time and stores it in the buffer. In the second step, the model classifies the digit with the inference rate of 280 ms, by taking the first image in the buffer as input.

Photoacoustic/Optoacoustic tomography aims to reconstruct maps of the initial pressure rise induced by the absorption of light pulses in tissue. This reconstruction is an ill-conditioned and under-determined problem, when the data acquisition protocol involves limited detection positions. The aim of the work is to develop an inversion method which integrates denoising procedure within the iterative model-based reconstruction to improve quantitative performance of optoacoustic imaging. Among the model-based schemes, total-variation (TV) constrained reconstruction scheme is a popular approach. In this work, a two-step approach was proposed for improving the TV constrained optoacoustic inversion by adding a non-local means based filtering step within each TV iteration. Compared to TV-based reconstruction, inclusion of this non-local means step resulted in signal-to-noise ratio improvement of 2.5 dB in the reconstructed optoacoustic images. Accepted version Phaneendra K. Yalavarthy acknowledges the DST-ICPS cluster funding (T-851) for the data science program. Part of the work is supported through IFTAS-CDS Collaborative Laboratory of Data Science and Engineering. Jaya Prakash acknowledges the IISc Startup Grant as well as DBT-IYBA Program.

Photoacoustic tomography (PAT) is an imaging technique that enables us to visualize vasculature networks at high resolution at depths much greater than optical microscopy.1,2 PAT can also resolve tissue chromophores such as oxyhemoglobin and deoxyhemoglobin, which allows estimation of tissue oxygentation levels.1,3 Accurate estimation of tissue oxygentation is very important in the fields of neuroimaging, cancer diagnostics, and critical care medicine.4,5 PAT involves collection of broadband acoustic waves generated by thermoelastic expansion of tissue upon absorption of light (illuminated at near-infrared wavelengths).6,7 PAT image reconstruction involves recovery of the acoustic source distribution from the detected broadband acoustic waves, which can be performed using analytical or model-based methods.8 Analytical methods are computationally less expensive compared to model-based image reconstruction.8,9 In contrary, model-based schemes are known to generate accurate images compared to analytical schemes, specifically in scenarios where irregular acquisition geometry is involved or limited data is acquired.9,10 Even with the model-based schemes, PAT inversion becomes difficult when the number of ultrasound detector positions is limited, and the situation gets aggravated due to the presence of noise in the data. More recently deep learning-based image postprocessing schemes have been developed for artifact removal and for limited data situations in PAT/microscopy;11–13 however, these schemes were trained on simulated data and generating large experimental data is very difficult. Segmentation approaches were proposed to delineate skin lining in mesoscopic photoacoustic images.14 Active contours-based image segmentation was used to obtain the PAT region of interest (ROI), the obtained ROI was then used to perform fluence correction to make PAT more quantitative.15,16 Different computational techniques have been reported in the literature to obtain vascular network in photoacoustic imaging. Vesselness filter has been proposed to enable accurate recovery of vasculature in PAT;17–19 however, questions have been raised on the accuracy of using the vesselness filter (due to generation of artificial vasculature).20 Probabilistic approach involving different steps such as smoothing and filtering the data, clustering, vessel-segmentation using clusters, and morphological filling was developed for segmenting vessels in photoacoustic images.21 More recently convolution neural network-based approaches were also developed for jointly performing the image reconstruction and segmentation.22 Further, deep learning-based approach was proposed for segmenting animal boundary using multi-modal photoacoustic and ultrasound data.23 Further, deep learning-based method has also been developed for animal brain imaging.24 Notably segmenting vasculature is very important in the context of diagnosis and surgical guidance, specifically for cardiovascular diseases.25,26 In this work, we developed a binary tomography approach wherein we assume the reconstruction distribution contains only two unknowns namely the background and the absorbers. The major contributions of this work include: (a) application of binary tomographic methods for photoacoustic tomographic imaging involving spherical wave propagation as opposed to straight line approximation applied in x-ray computed tomography; (b) showing the utility of binary tomographic methods for improving photoacoustic tomographic imaging of vasculature; and (c) establishing the binary tomographic method superiority in photoacoustic tomographic imaging for providing approximate priors to improve model-based image reconstruction. Binary tomography problem is solved based on a dual optimization approach with an auxiliary variable posed as a constrained optimization problem. We compared the binary tomography algorithm against standard backprojection approach and model-based reconstruction schemes such as Tikhonov regularization and sparsity-based regularization.27 The performance of these different algorithms was validated with in-silico data, physical phantom, and in-vivo brain vasculature imaging data.

Lung ultrasound (US) imaging has the potential to be an effective point-of-care test for detection of COVID-19, due to its ease of operation with minimal personal protection equipment along with easy disinfection. The current state-of-the-art deep learning models for detection of COVID-19 are heavy models that may not be easy to deploy in commonly utilized mobile platforms in point-of-care testing. In this work, we develop a lightweight mobile friendly efficient deep learning model for detection of COVID-19 using lung US images. Three different classes including COVID-19, pneumonia, and healthy were included in this task. The developed network, named as Mini-COVIDNet, was bench-marked with other lightweight neural network models along with state-of-the-art heavy model. It was shown that the proposed network can achieve the highest accuracy of 83.2% and requires a training time of only 24 min. The proposed Mini-COVIDNet has 4.39 times less number of parameters in the network compared to its next best performing network and requires a memory of only 51.29 MB, making the point-of-care detection of COVID-19 using lung US imaging plausible on a mobile platform. Deployment of these lightweight networks on embedded platforms shows that the proposed Mini-COVIDNet is highly versatile and provides optimal performance in terms of being accurate as well as having latency in the same order as other lightweight networks. The developed lightweight models are available at https://github.com/navchetan-awasthi/Mini-COVIDNet.

Low-cost automated histopathology microscopy systems usually suffer from optical imperfections, producing images that are slightly Out of Focus (OoF). In this work, a guided filter (GF) based image preprocessing is proposed for compensating focal errors and its efficacy is demonstrated on images of healthy and malaria infected red blood cells (h-RBCs and i-RBCs), and PAP smears. Since contrast enhancement has been widely used as an image preprocessing step for the analysis of histopathology images, a systematic comparison is made with six such prominently used methods, namely Contrast Limited Adaptive Histogram Equalization (CLAHE), RIQMC-based optimal histogram matching (ROHIM), modified L0, Morphological Varying(MV)-Bitonic filter, unsharp mask filter and joint bilateral filter. The images enhanced using GF approach lead to better segmentation accuracy (upto 50% improvement over native images) and visual quality compared to other approaches, without any change in the color tones. Thus, the proposed GF approach is a viable solution for rectifying the OoF microscopy images without the loss of the valuable diagnostic information presented by the color tone.

The quality of the reconstructed photoacoustic image largely depends on the amount of photoacoustic (PA) boundary data available, which in turn is proportional to the number of detectors employed. In case of limited data (owing to less number of detectors due to cost/instrumentation constraints), the reconstructed PA images suffer from artifacts and are often noisy. In this work, for the first time, a deep learning based model was developed to super resolve and denoise the photoacoustic sinogram data. The proposed method was compared with existing nearest neighbor interpolation and wavelet based denoising techniques and was shown to outperform them both in numerical and in-vivo cases. The improvement obtained in Root Mean Square Error (RMSE) and Peak Signal to Noise Ratio (PSNR) for the reconstructed PA image using the sinogram data that was super-resolved and denoised using proposed neural network based method was as high as 41.70 % and 6.93 dB respectively compared to utilizing limited sinogram data.

Photoacoustic tomography (PAT) is a noninvasive imaging modality combining the benefits of optical contrast at ultrasonic resolution. Analytical reconstruction algorithms for photoacoustic (PA) signals require a large number of data points for accurate image reconstruction. However, in practical scenarios, data are collected using the limited number of transducers along with data being often corrupted with noise resulting in only qualitative images. Furthermore, the collected boundary data are band-limited due to limited bandwidth (BW) of the transducer, making the PA imaging with limited data being qualitative. In this work, a deep neural network-based model with loss function being scaled root-mean-squared error was proposed for super-resolution, denoising, as well as BW enhancement of the PA signals collected at the boundary of the domain. The proposed network has been compared with traditional as well as other popular deep-learning methods in numerical as well as experimental cases and is shown to improve the collected boundary data, in turn, providing superior quality reconstructed PA image. The improvement obtained in the Pearson correlation, structural similarity index metric, and root-mean-square error was as high as 35.62%, 33.81%, and 41.07%, respectively, for phantom cases and signal-to-noise ratio improvement in the reconstructed PA images was as high as 11.65 dB for in vivo cases compared with reconstructed image obtained using original limited BW data. Code is available at https://sites.google.com/site/sercmig/home/dnnpat. Accepted version

Implementing image-guidance for diffuse optical tomographic imaging involves reducing the spatial optical parameter space within the discrete tissue types being estimated. This makes the inverse problem overdetermined which indicates that the required number of measurements could be less than all available measurements. In this work, we propose a data-optimization approach to curtail the algorithmic complexity and implicitly reconstruct optical absorption image based on direct sensitivity approach. The performance of the proposed method was validated using numerical and gelatin phantom data indicating that this perturbation-like approach can quantify embedded regions with good accuracy and is free of bias errors associated with regularization approaches. The testing of the algorithm on human data of fast pulsatile NIR imaging in breast tissue showed that fast updates are possible and the required number of measurements is equal to discrete tissue types. The proposed method offers high level of measurement optimization for the dynamic imaging, compared to traditional methods of full iterative regularized tomography or recently proposed data resolution-based methods.

Purpose: A prior image based temporally constrained reconstruction ( PITCR) algorithm was developed for obtaining accurate temperature maps having better volume coverage, and spatial, and temporal resolution than other algorithms for highly undersampled data in magnetic resonance (MR) thermometry. Methods: The proposed PITCR approach is an algorithm that gives weight to the prior image and performs accurate reconstruction in a dynamic imaging environment. The PITCR method is compared with the temporally constrained reconstruction (TCR) algorithm using pork muscle data. Results: The PITCR method provides superior performance compared to the TCR approach with highly undersampled data. The proposed approach is computationally expensive compared to the TCR approach, but this could be overcome by the advantage of reconstructing with fewer measurements. In the case of reconstruction of temperature maps from 16% of fully sampled data, the PITCR approach was 1.57x slower compared to the TCR approach, while the root mean square error using PITCR is 0.784 compared to 2.815 with the TCR scheme. Conclusions: The PITCR approach is able to perform more accurate reconstructions of temperature maps compared to the TCR approach with highly undersampled data in MR guided high intensity focused ultrasound. (C) 2015 American Association of Physicists in Medicine.

Purpose: Development of simple and computationally efficient extrapolated Tikhonov filtering reconstruction methods for photoacoustic tomography. Methods: The model-based reconstruction algorithms in photoacoustic tomography typically utilize Tikhonov regularization scheme for the reconstruction of initial pressure distribution from the measured boundary acoustic data. The automated choice of regularization parameter in these cases is computationally expensive. Moreover, the Tikhonov scheme promotes the smooth features in the reconstructed image due to the smooth regularizer, thus leading to loss of sharp features. The proposed extrapolation method estimates the solution at zero regularization assuming the solution being a function of regularization parameter and thus posing it as a zero value problem. Thus the numerically computed zero regularization solution is expected to have better features compared to standard Tikhonov solution, with an added advantage of removing the necessity of automated choice of regularization. The reconstructed results using this method, were shown in three variants (Lanczos, Traditional, and Exponential) of Tikhonov filtering and were compared with the stan- dard error estimate technique. Results: Four numerical (including realistic breast phantom) and two experimental phantom data was utilized to show the e ectiveness of the proposed method. It was shown that the proposed method performance was superior than the standard error estimate technique, being at least four times faster in terms of computation, and provides an improvement as high as 2.6 times in terms of standard gures of merit. Conclusion: The developed extrapolated Tikhonov ltering methods overcome the di culty of obtaining a suitable regularization parameter and shown to be reconstructing high quality photoacoustic images with additional advantage of being computationally efficient, making it more appealing in real-time applications. Accepted version

CT and MR perfusion weighted imaging (PWI) enable quantification of perfusion parameters in stroke studies. These parameters are calculated from the residual impulse response function (IRF) based on a physiological model for tissue perfusion. The standard approach for estimating the IRF is deconvolution using oscillatory-limited singular value decomposition (oSVD) 1 or Frequency Domain Deconvolution (FDD). 2 FDD is widely recognized as the fastest approach currently available for deconvolution of CT Perfusion/MR PWI. 2 In this work, three faster methods are proposed. The first is a direct (model based) crude approximation to the final perfusion quantities (Blood flow, Blood volume, Mean Transit Time and Delay) using the Welch-Satterthwaite approximation for gamma fitted concentration time curves (CTC). The second method is a fast accurate deconvolution method, we call Analytical Fourier Filtering (AFF). The third is another fast accurate deconvolution technique using Showalter's method, we call Analytical Showalter's Spectral Filtering (ASSF). Through systematic evaluation on phantom and clinical data, the proposed methods are shown to be computationally more than twice as fast as FDD. The two deconvolution based methods, AFF and ASSF, are also shown to be quantitatively accurate compared to FDD and oSVD.

The sparse recovery methods utilize the l(p)-normbased regularization in the estimation problem with 0

The methods available for solving the inverse problem of photoacoustic tomography promote only one feature–either being smooth or sharp–in the resultant image. The fusion of photoacoustic images reconstructed from distinct methods improves the individually reconstructed images, with the guided filter based approach being state-of-the-art, which requires that implicit regularization parameters are chosen. In this work, a deep fusion method based on convolutional neural networks has been proposed as an alternative to the guided filter based approach. It has the combined benefit of using less data for training without the need for the careful choice of any parameters and is a fully data-driven approach. The proposed deep fusion approach outperformed the contemporary fusion method, which was proved using experimental, numerical phantoms and in-vivo studies. The improvement obtained in the reconstructed images was as high as 95.49% in root mean square error and 7.77 dB in signal to noise ratio (SNR) in comparison to the guided filter approach. Also, it was demonstrated that the proposed deep fuse approach, trained on only blood vessel type images at measurement data SNR being 40 dB, was able to provide a generalization that can work across various noise levels in the measurement data, experimental set-ups as well as imaging objects. Published version

PurposeTo extend the previously developed temporally constrained reconstruction (TCR) algorithm to allow for real-time availability of three-dimensional (3D) temperature maps capable of monitoring MR-guided high intensity focused ultrasound applications. MethodsA real-time TCR (RT-TCR) algorithm is developed that only uses current and previously acquired undersampled k-space data from a 3D segmented EPI pulse sequence, with the image reconstruction done in a graphics processing unit implementation to overcome computation burden. Simulated and experimental data sets of HIFU heating are used to evaluate the performance of the RT-TCR algorithm. ResultsThe simulation studies demonstrate that the RT-TCR algorithm has subsecond reconstruction time and can accurately measure HIFU-induced temperature rises of 20 degrees C in 15 s for 3D volumes of 16 slices (RMSE = 0.1 degrees C), 24 slices (RMSE = 0.2 degrees C), and 32 slices (RMSE = 0.3 degrees C). Experimental results in ex vivo porcine muscle demonstrate that the RT-TCR approach can reconstruct temperature maps with 192 x 162 x 66 mm 3D volume coverage, 1.5 x 1.5 x 3.0 mm resolution, and 1.2-s scan time with an accuracy of 0.5 degrees C. ConclusionThe RT-TCR algorithm offers an approach to obtaining large coverage 3D temperature maps in real-time for monitoring MR-guided high intensity focused ultrasound treatments. Magn Reson Med 71:1394-1404, 2014. (c) 2013 Wiley Periodicals, Inc.

Photoacoustic (PA) signals collected at the boundary of tissue are always band-limited. A deep neural network (DNN) was proposed to enhance the bandwidth of the detected PA signal, thereby improving the quantitative accuracy of the reconstructed PA images. A least square based deconvolution method that utilizes the Tikhonov regularization framework was used for comparison with the proposed network. The proposed method was evaluated using both numerical and experimental data. The results indicate that the proposed method was capable of enhancing the bandwidth of the detected PA signal, which inturn improves the contrast recovery and quality of reconstructed PA images without adding any significant computational burden. NRF (Natl Research Foundation, S’pore) MOE (Min. of Education, S’pore) NMRC (Natl Medical Research Council, S’pore) Published version

The limited data photoacoustic image reconstruction problem is typically solved using either weighted or ordinary least squares (LS), with regularization term being added for stability, which account only for data imperfections (noise). Numerical modeling of acoustic wave propagation requires discretization of imaging region and is typically developed based on many assumptions, such as speed of sound being constant in the tissue, making it imperfect. In this work, two variants of total least squares (TLS), namely ordinary TLS and Sparse TLS were developed, which account for model imperfections. The ordinary TLS is implemented in the Lanczos bidiagonalization framework to make it computationally efficient. The Sparse TLS utilizes the total variation penalty to promote recovery of high frequency components in the reconstructed image. The Lanczos truncated TLS (Lanczos T-TLS) and Sparse TLS methods were compared with the recently established state-of-the-art methods, such as Lanczos Tikhonov and Exponential Filtering. The TLS methods exhibited better performance for experimental data as well as in cases where modeling errors were present, such as few acoustic detectors malfunctioning and speed of sound variations. Also, the TLS methods does not require any prior information about the errors present in the model or data, making it attractive for real-time scenarios. NRF (Natl Research Foundation, S’pore) MOE (Min. of Education, S’pore) NMRC (Natl Medical Research Council, S’pore) MOH (Min. of Health, S’pore) Accepted version

The linear inverse problem encountered in restoration of blurred noisy images is typically solved via Tikhonov minimization. The outcome (restored image) of such minimization is highly dependent on the choice of regularization parameter. In the absence of prior information about the noise levels in the blurred image, finding this regularization parameter in an automated way becomes extremely challenging. The available methods like Generalized Cross Validation (GCV) may not yield optimal results in all cases. A novel method that relies on minimal residual method for finding the regularization parameter automatically is proposed here and was systematically compared with the GCV method. It was shown that the proposed method performance is superior to the GCV method in providing high quality restored images in cases where the noise levels are high.

The model-based image reconstruction techniques for photoacoustic (PA) tomography require an explicit regularization. An error estimate (η2) minimization-based approach was proposed and developed for the determination of a regularization parameter for PA imaging. The regularization was used within Lanczos bidiagonalization framework, which provides the advantage of dimensionality reduction for a large system of equations. It was shown that the proposed method is computationally faster than the state-of-the-art techniques and provides similar performance in terms of quantitative accuracy in reconstructed images. It was also shown that the error estimate (η2) can also be utilized in determining a suitable regularization parameter for other popular techniques such as Tikhonov, exponential, and nonsmooth (l1 and total variation norm based) regularization methods.

Brain templates provide a standard anatomical platform for population based morphometric assessments. Typically, standard brain templates for such assessments are created using Caucasian brains, which may not be ideal to analyze brains from other ethnicities. To effectively demonstrate this, we compared brain morphometric differences between T1 weighted structural MRI images of 27 healthy Indian and Caucasian subjects of similar age and same sex ratio. Furthermore, a population specific brain template was created from MRI images of healthy Indian subjects and compared with standard Montreal Neurological Institute (MNI-152) template. We also examined the accuracy of registration of by acquiring a different T1 weighted MRI data set and registering them to newly created Indian template and MNI-152 template. The statistical analysis indicates significant difference in global brain measures and regional brain structures of Indian and Caucasian subjects. Specifically, the global brain measurements of the Indian brain template were smaller than that of the MNI template. Also, Indian brain images were better realigned to the newly created template than to the MNI-152 template. The notable variations in Indian and Caucasian brains convey the need to build a population specific Indian brain template and atlas.

The attenuation of near-infrared (NIR) light intensity as it propagates in a turbid medium like biological tissue is described by modified the Beer–Lambert law (MBLL). The MBLL is generally used to quantify the changes in tissue chromophore concentrations for NIR spectroscopic data analysis. Even though MBLL is effective in terms of providing qualitative comparison, it suffers from its applicability across tissue types and tissue dimensions. In this work, we introduce Lambert-W function-based modeling for light propagation in biological tissues, which is a generalized version of the Beer–Lambert model. The proposed modeling provides parametrization of tissue properties, which includes two attenuation coefficients μ0 and η. We validated our model against the Monte Carlo simulation, which is the gold standard for modeling NIR light propagation in biological tissue. We included numerous human and animal tissues to validate the proposed empirical model, including an inhomogeneous adult human head model. The proposed model, which has a closed form (analytical), is first of its kind in providing accurate modeling of NIR light propagation in biological tissues.

The diffusion equation-based modeling of near infrared light propagation in tissue is achieved by using finite-element mesh for imaging real-tissue types, such as breast and brain. The finite-element mesh size (number of nodes) dictates the parameter space in the optical tomographic imaging. Most commonly used finite-element meshing algorithms do not provide the flexibility of distinct nodal spacing in different regions of imaging domain to take the sensitivity of the problem into consideration. This study aims to present a computationally efficient mesh simplification method that can be used as a preprocessing step to iterative image reconstruction, where the finite-element mesh is simplified by using an edge collapsing algorithm to reduce the parameter space at regions where the sensitivity of the problem is relatively low. It is shown, using simulations and experimental phantom data for simple meshes/domains, that a significant reduction in parameter space could be achieved without compromising on the reconstructed image quality. The maximum errors observed by using the simplified meshes were less than 0.27% in the forward problem and 5% for inverse problem.