Diffuse Optical Tomography Imaging Research Articles

Improving diffuse optical tomography imaging quality using APU-Net: an attention-based physical U-Net model.

Traditional diffuse optical tomography (DOT) reconstructions are hampered by image artifacts arising from factors such as DOT sources being closer to shallow lesions, poor optode-tissue coupling, tissue heterogeneity, and large high-contrast lesions lacking information in deeper regions (known as shadowing effect). Addressing these challenges is crucial for improving the quality of DOT images and obtaining robust lesion diagnosis. We address the limitations of current DOT imaging reconstruction by introducing an attention-based U-Net (APU-Net) model to enhance the image quality of DOT reconstruction, ultimately improving lesion diagnostic accuracy. We designed an APU-Net model incorporating a contextual transformer attention module to enhance DOT reconstruction. The model was trained on simulation and phantom data, focusing on challenges such as artifact-induced distortions and lesion-shadowing effects. The model was then evaluated by the clinical data. Transitioning from simulation and phantom data to clinical patients' data, our APU-Net model effectively reduced artifacts with an average artifact contrast decrease of 26.83% and improved image quality. In addition, statistical analyses revealed significant contrast improvements in depth profile with an average contrast increase of 20.28% and 45.31% for the second and third target layers, respectively. These results highlighted the efficacy of our approach in breast cancer diagnosis. The APU-Net model improves the image quality of DOT reconstruction by reducing DOT image artifacts and improving the target depth profile.

Ultrasound and diffuse optical tomography-transformer model for assessing pathological complete response to neoadjuvant chemotherapy in breast cancer.

We evaluate the efficiency of integrating ultrasound (US) and diffuse optical tomography (DOT) images for predicting pathological complete response (pCR) to neoadjuvant chemotherapy (NAC) in breast cancer patients. The ultrasound-diffuse optical tomography (USDOT)-Transformer model represents a significant step toward accurate prediction of pCR, which is critical for personalized treatment planning. We aim to develop and assess the performance of the USDOT-Transformer model, which combines US and DOT images with tumor receptor biomarkers to predict the pCR of breast cancer patients under NAC. We developed the USDOT-Transformer model using a dual-input transformer to process co-registered US and DOT images along with tumor receptor biomarkers. Our dataset comprised imaging data from 60 patients at multiple time points during their chemotherapy treatment. We used fivefold cross-validation to assess the model's performance, comparing its results against a single modality of US or DOT. The USDOT-Transformer model demonstrated excellent predictive performance, with a mean area under the receiving characteristic curve of 0.96 (95%CI: 0.93 to 0.99) across the fivefold cross-validation. The integration of US and DOT images significantly enhanced the model's ability to predict pCR, outperforming models that relied on a single imaging modality (0.87 for US and 0.82 for DOT). This performance indicates the potential of advanced deep learning techniques and multimodal imaging data for improving the accuracy (ACC) of pCR prediction. The USDOT-Transformer model offers a promising non-invasive approach for predicting pCR to NAC in breast cancer patients. By leveraging the structural and functional information from US and DOT images, the model offers a faster and more reliable tool for personalized treatment planning. Future work will focus on expanding the dataset and refining the model to further improve its accuracy and generalizability.

Deep learning-enabled high-speed, multi-parameter diffuse optical tomography.

Frequency-domain diffuse optical tomography (FD-DOT) could enhance clinical breast tumor characterization. However, conventional diffuse optical tomography (DOT) image reconstruction algorithms require case-by-case expert tuning and are too computationally intensive to provide feedback during a scan. Deep learning (DL) algorithms front-load computational and tuning costs, enabling high-speed, high-fidelity FD-DOT. We aim to demonstrate a simultaneous reconstruction of three-dimensional absorption and reduced scattering coefficients using DL-FD-DOT, with a view toward real-time imaging with a handheld probe. A DL model was trained to solve the DOT inverse problem using a realistically simulated FD-DOT dataset emulating a handheld probe for human breast imaging and tested using both synthetic and experimental data. Over a test set of 300 simulated tissue phantoms for absorption and scattering reconstructions, the DL-DOT model reduced the root mean square error by and , increased the spatial similarity by and , increased the anomaly contrast accuracy by ( ), and reduced the crosstalk by and , respectively, compared with model-based tomography. The average reconstruction time was reduced from 3.8min to 0.02s for a single reconstruction. The model was successfully verified using two tumor-emulating optical phantoms. There is clinical potential for real-time functional imaging of human breast tissue using DL and FD-DOT.

Functional hemodynamic imaging markers for the prediction of pathological outcomes in breast cancer patients treated with neoadjuvant chemotherapy.

Achieving pathologic complete response (pCR) after neoadjuvant chemotherapy (NACT) is a significant predictor of increased likelihood of survival in breast cancer patients. Early prediction of pCR is of high clinical value as it could allow personalized adjustment of treatment regimens in non-responding patients for improved outcomes. We aim to assess the association between hemoglobin-based functional imaging biomarkers derived from diffuse optical tomography (DOT) and the pathological outcome represented by pCR at different timepoints along the course of NACT. Twenty-two breast cancer patients undergoing NACT were enrolled in a multimodal DOT and X-ray digital breast tomosynthesis (DBT) imaging study in which their breasts were imaged at different compression levels. Logistic regressions were used to study the associations between DOT-derived imaging markers evaluated after the first and second cycles of chemotherapy, respectively, with pCR status determined after the conclusion of NACT at the time of surgery. Receiver operating characteristic curve analysis was also used to explore the predictive performance of selected DOT-derived markers. Normalized tumor HbT under half compression was significantly lower in the pCR group compared to the non-pCR group after two chemotherapy cycles (). In addition, the change in normalized tumor upon reducing compression from full to half mammographic force was identified as another potential indicator of pCR at an earlier time point, i.e., after the first chemo cycle (). Exploratory predictive assessments showed that AUCs using DOT-derived functional imaging markers as predictors reach as high as 0.75 and 0.71, respectively, after the first and second chemo cycle, compared to AUCs of 0.50 and 0.53 using changes in tumor size measured on DBT and MRI. These findings suggest that breast DOT could be used to assist response assessment in women undergoing NACT, a critical but unmet clinical need, and potentially enable personalized adjustments of treatment regimens.

Mapping brain function in adults and young children during naturalistic viewing with high-density diffuse optical tomography.

Human studies of early brain development have been limited by extant neuroimaging methods. MRI scanners present logistical challenges for imaging young children, while alternative modalities like functional near-infrared spectroscopy have traditionally been limited by image quality due to sparse sampling. In addition, conventional tasks for brain mapping elicit low task engagement, high head motion, and considerable participant attrition in pediatric populations. As a result, typical and atypical developmental trajectories of processes such as language acquisition remain understudied during sensitive periods over the first years of life. We evaluate high-density diffuse optical tomography (HD-DOT) imaging combined with movie stimuli for high resolution optical neuroimaging in awake children ranging from 1 to 7 years of age. We built an HD-DOT system with design features geared towards enhancing both image quality and child comfort. Furthermore, we characterized a library of animated movie clips as a stimulus set for brain mapping and we optimized associated data analysis pipelines. Together, these tools could map cortical responses to movies and contained features such as speech in both adults and awake young children. This study lays the groundwork for future research to investigate response variability in larger pediatric samples and atypical trajectories of early brain development in clinical populations.

DOTnet 2.0: Deep learning network for diffuse optical tomography image reconstruction

Breast cancer is the most common cancer worldwide. The standard imaging modality for breast cancer screening is X-ray mammography, which suffers from low sensitivities in women with dense breasts and can potentially cause cancers despite a low radiation dosage. Diffuse Optical Tomography (DOT) is a noninvasive imaging technique that can potentially be employed to improve breast cancer early detection. However, conventional model-based algorithms for reconstructing DOT images usually produce low-quality images with limited resolution and low reconstruction accuracy. We propose to integrate deep neural networks (DNNs) with the conventional DOT reconstruction methods. This hybrid framework significantly enhances image quality. The DNNs have been trained and tested with sample data derived from clinically relevant breast models. The sample dataset contains blood vessel structures from breast structures and artificially created vessels using the Lindenmayer-system algorithm. By comparing the hybrid reconstruction with the ground truth image, we demonstrated a multi scale - structural similarity index measure (MS-SSIM) score of 0.80–0.90. Whereas using conventional reconstruction, MS-SSIM provided a much inferior score of 0.36–0.59. In terms of DOT image quality, both qualitative and quantitative assessments of the reconstructed images signify that the hybrid approach is superior to conventional methods. This improvement suggests that DOT can potentially become a viable alternative to breast cancer screening, providing a step towards the next-generation device for optical mammography.

Automated pipeline for breast cancer diagnosis using US assisted diffuse optical tomography.

Ultrasound (US)-guided diffuse optical tomography (DOT) is a portable and non-invasive imaging modality for breast cancer diagnosis and treatment response monitoring. However, DOT data pre-processing and imaging reconstruction often require labor intensive manual processing which hampers real-time diagnosis. In this study, we aim at providing an automated US-assisted DOT pre-processing, imaging and diagnosis pipeline to achieve near real-time diagnosis. We have developed an automated DOT pre-processing method including motion detection, mismatch classification using deep-learning approach, and outlier removal. US-lesion information needed for DOT reconstruction was extracted by a semi-automated lesion segmentation approach combined with a US reading algorithm. A deep learning model was used to evaluate the quality of the reconstructed DOT images and a two-step deep-learning model developed earlier is implemented to provide final diagnosis based on US imaging features and DOT measurements and imaging results. The presented US-assisted DOT pipeline accurately processed the DOT measurements and reconstruction and reduced the procedure time to 2 to 3 minutes while maintained a comparable classification result with manually processed dataset.

Gauss-Seidel based spatially varying optimal regularization improves reconstruction in diffuse optical tomography

ABSTRACT The inverse problem associated with Diffuse optical tomography image reconstruction is known to be highly nonlinear, under-determined, and ill-posed. The Levenberg-Marquardt technique is employed in solving it and is known to produce low-resolution reconstructed images. To stabilize the inversion of the large matrix, a heuristically chosen regularization parameter is used. A novel methodology is developed using Gauss-Seidel, Modified Richardson, and Kaczmarz recursive methods to solve the inverse problem and to obtain spatially varying regularization parameters, which are optimally obtained for every node automatically, which is otherwise not possible. The proposed methods are thoroughly compared with the existing traditional methods in both 2-D and 3-D imaging domains using numerically simulated noisy data and also real-life phantom data. Of all the proposed methods, the Gauss-Seidel-based method provides a quantitatively accurate estimation of spatially varying regularization using the model-resolution-matrix-based method and hence improves the quality of the reconstructed images with better resolution characteristics.

FDU-Net: Deep Learning-Based Three-Dimensional Diffuse Optical Image Reconstruction.

Near-infrared diffuse optical tomography (DOT) is a promising functional modality for breast cancer imaging; however, the clinical translation of DOT is hampered by technical limitations. Specifically, conventional finite element method (FEM)-based optical image reconstruction approaches are time-consuming and ineffective in recovering full lesion contrast. To address this, we developed a deep learning-based reconstruction model (FDU-Net) comprised of a Fully connected subnet, followed by a convolutional encoder-Decoder subnet, and a U-Net for fast, end-to-end 3D DOT image reconstruction. The FDU-Net was trained on digital phantoms that include randomly located singular spherical inclusions of various sizes and contrasts. Reconstruction performance was evaluated in 400 simulated cases with realistic noise profiles for the FDU-Net and conventional FEM approaches. Our results show that the overall quality of images reconstructed by FDU-Net is significantly improved compared to FEM-based methods and a previously proposed deep-learning network. Importantly, once trained, FDU-Net demonstrates substantially better capability to recover true inclusion contrast and location without using any inclusion information during reconstruction. The model was also generalizable to multi-focal and irregularly shaped inclusions unseen during training. Finally, FDU-Net, trained on simulated data, could successfully reconstruct a breast tumor from a real patient measurement. Overall, our deep learning-based approach demonstrates marked superiority over the conventional DOT image reconstruction methods while also offering over four orders of magnitude acceleration in computational time. Once adapted to the clinical breast imaging workflow, FDU-Net has the potential to provide real-time accurate lesion characterization by DOT to assist the clinical diagnosis and management of breast cancer.

Short-separation regression incorporated diffuse optical tomography image reconstruction modeling for high-density functional near-infrared spectroscopy.

Short-separation (SS) regression and diffuse optical tomography (DOT) image reconstruction, two widely adopted methods in functional near-infrared spectroscopy (fNIRS), were demonstrated to individually facilitate the separation of brain activation and physiological signals, with further improvement using both sequentially. We hypothesized that doing both simultaneously would further improve the performance. Motivated by the success of these two approaches, we propose a method, SS-DOT, which applies SS and DOT simultaneously. The method, which employs spatial and temporal basis functions to represent the hemoglobin concentration changes, enables us to incorporate SS regressors into the time series DOT model. To benchmark the performance of the SS-DOT model against conventional sequential models, we use fNIRS resting state data augmented with synthetic brain response as well as data acquired during a ball squeezing task. The conventional sequential models comprise performing SS regression and DOT. The results show that the SS-DOT model improves the image quality by increasing the contrast-to-background ratio by a threefold improvement. The benefits are marginal at small brain activation. The SS-DOT model improves the fNIRS image reconstruction quality.

An Integrated-Circuit Node for High-Spatiotemporal Resolution Time-Domain Near-Infrared Diffuse Optical Tomography Imaging Arrays

Next-generation brain–computer interfaces (BCIs) for healthy individuals are expected to largely rely on noninvasive functional imaging methods to record cortex-wide neural activity because of the risk associated with surgically implanted devices. In this work, we present a fully integrated 1.8 × 1.8 mm single chip that can be arrayed on a wearable patch to perform noninvasive, functional brain imaging over large cortical areas. This chip node contains two bonded vertical-cavity surface-emitting lasers (VCSELs), an 8 × 8 single-photon avalanche diode (SPAD) array with event-driven time-to-digital converters (TDCs) per row, and a digital back-end for on-chip histogramming and time-gating. We achieved 70-ps resolution for time-of-flight (ToF) imaging at the record-high 100-MHz laser repetition rate with 80-mW total power. We showed that time-gating improves the imaging contrast by as much as 36% using a brain-skull phantom. The fully integrated and compact node presented here is the key enabler for future high-spatiotemporal-resolution time-domain diffuse optical tomography (TD-DOT) imaging arrays.

A Review of Image Reconstruction Algorithms for Diffuse Optical Tomography

Diffuse optical tomography (DOT) is a biomedical imaging modality that can reconstruct hemoglobin concentration and associated oxygen saturation by using detected light passing through a biological medium. Various clinical applications of DOT such as the diagnosis of breast cancer and functional brain imaging are expected. However, it has been difficult to obtain high spatial resolution and quantification accuracy with DOT because of diffusive light propagation in biological tissues with strong scattering and absorption. In recent years, various image reconstruction algorithms have been proposed to overcome these technical problems. Moreover, with progress in related technologies, such as artificial intelligence and supercomputers, the circumstances surrounding DOT image reconstruction have changed. To support the applications of DOT image reconstruction in clinics and new entries of related technologies in DOT, we review the recent efforts in image reconstruction of DOT from the viewpoint of (i) the forward calculation process, including the radiative transfer equation and its approximations to simulate light propagation with high precision, and (ii) the optimization process, including the use of sparsity regularization and prior information to improve the spatial resolution and quantification.

Fusion deep learning approach combining diffuse optical tomography and ultrasound for improving breast cancer classification.

Diffuse optical tomography (DOT) is a promising technique that provides functional information related to tumor angiogenesis. However, reconstructing the DOT function map of a breast lesion is an ill-posed and underdetermined inverse process. A co-registered ultrasound (US) system that provides structural information about the breast lesion can improve the localization and accuracy of DOT reconstruction. Additionally, the well-known US characteristics of benign and malignant breast lesions can further improve cancer diagnosis based on DOT alone. Inspired by a fusion model deep learning approach, we combined US features extracted by a modified VGG-11 network with images reconstructed from a DOT deep learning auto-encoder-based model to form a new neural network for breast cancer diagnosis. The combined neural network model was trained with simulation data and fine-tuned with clinical data: it achieved an AUC of 0.931 (95% CI: 0.919-0.943), superior to those achieved using US images alone (0.860) or DOT images alone (0.842).

Compact breast shape acquisition system for improving diffuse optical tomography image reconstructions.

Diffuse optical tomography (DOT) has been investigated for diagnosing malignant breast lesions, but its accuracy relies on model-based image reconstructions, which in turn depends on the accuracy of breast shape acquisition. In this work, we have developed a dual-camera structured light imaging (SLI) breast shape acquisition system tailored for a mammography-like compression setting. Illumination pattern intensity is dynamically adjusted to account for skin tone differences, while thickness-informed pattern masking reduces artifacts due to specular reflections. This compact system is affixed to a rigid mount that can be installed into existing mammography or parallel-plate DOT systems without the need for camera-projector re-calibration. Our SLI system produces sub-millimeter resolution with a mean surface error of 0.26 mm. This breast shape acquisition system results in more accurate surface recovery, with an average 1.6-fold reduction in surface estimation errors over a reference method via contour extrusion. Such improvement translates to 25% to 50% reduction in mean squared error in the recovered absorption coefficient for a series of simulated tumors 1-2 cm below the skin.

Deep Learning of Diffuse Optical Tomography Based on Time-Domain Radiative Transfer Equation

Near infrared diffuse optical tomography (DOT) is a potential tool for diagnosing cancer by image reconstruction of tissue optical properties. A variety of image reconstruction methods for DOT have been attempted, in general, based on the diffusion equation (DE). However, the image quality is still insufficient to clinical use, which is mainly attributed to the fact that the DE is invalid in some regions, such as low-scattering regions, and the inverse problem is inherently ill-posed. In contrast, the radiative transfer equation (RTE) accurately describes light propagation in biological tissue and also the DOT by deep learning is recently thought to be an alternative approach to the inverse problem. Distribution of time of flight (DTOF) of photons estimated by the time-domain RTE lends itself to deep learning along a temporal sequence. In this study, we propose a new DOT image reconstruction algorithm based on a long-short-term memory and the time-domain RTE. In simulation studies, using this algorithm, we succeeded in detection of an absorbing inclusion with a diameter of 5 mm, an absorber mimicking cancer, which was embedded in a two-dimensional square model (4 cm × 4 cm) with an optically homogeneous background. Multiple absorbers and a bigger absorber embedded in this model were also detected. We also demonstrate that, if simulation data by beam injection from multiple directions are employed as a training set, the accuracy of detection is improved especially for multiple absorbers.

Tikhonov regularization-based extended Kalman filter technique for robust and accurate reconstruction in diffuse optical tomography.

Diffuse optical tomography (DOT) is a non-invasive imaging modality that uses near-infrared light to probe the optical properties of tissue. In conventionally used deterministic methods for DOT inversion, the measurement errors were not taken into account, resulting in unsatisfactory noise robustness and, consequently, affecting the DOT image reconstruction quality. In order to overcome this defect, an extended Kalman filter (EKF)-based DOT reconstruction algorithm was introduced first, which improved the reconstruction results by incorporating a priori information and measurement errors to the model. Further, to mitigate the instability caused by the ill-condition of the observation matrix in the tomographic imaging problem, a new, to the best of our knowledge, estimation algorithm was derived by incorporating Tikhonov regularization to the EKF method. To verify the effectiveness of the EKF algorithm and Tikhonov regularization-based EKF algorithm for DOT imaging, a series of numerical simulations and phantom experiments were conducted, and the experimental results were quantitatively evaluated and compared with two conventionally used deterministic methods involving the algebraic reconstruction technique and Levenberg-Marquardt algorithm. The results show that the two EKF-based algorithms can accurately estimate the location and size of the target, and the imaging accuracy and noise robustness are obviously improved. Furthermore, the Tikhonov regularization-based EKF obtained optimal parameter estimations, especially under the circumstance of low absorption contrast (1.2) and high noise level (10%).

Method to improve the localization accuracy and contrast recovery of lesions in separately acquired X-ray and diffuse optical tomographic breast imaging.

Near-infrared diffuse optical tomography (DOT) has the potential to improve the accuracy of breast cancer diagnosis and aid in monitoring the response of breast tumors to chemotherapy by providing hemoglobin-based functional imaging. The use of structural lesion priors derived from clinical breast imaging methods, such as mammography, can improve recovery of tumor optical contrast; however, accurate lesion prior placement is essential to take full advantage of prior-guided DOT image reconstruction. Simultaneous optical and anatomical imaging may not always be possible or desired, which can make the accurate registration of the lesion prior challenging. In this paper, we present a three-step lesion prior scanning approach to facilitate improved accuracy in lesion localization based on the optical contrast quantified by the total hemoglobin concentration (HbT) for non-simultaneous multimodal DOT and digital breast tomosynthesis (DBT) imaging. In three challenging breast cancer patient cases, where no clear optical contrast was present initially, we have demonstrated consistent improvement in the recovered HbT lesion contrast by utilizing this method.

Prefrontal cortex supports speech perception in listeners with cochlear implants.

Cochlear implants are neuroprosthetic devices that can restore hearing in people with severe to profound hearing loss by electrically stimulating the auditory nerve. Because of physical limitations on the precision of this stimulation, the acoustic information delivered by a cochlear implant does not convey the same level of acoustic detail as that conveyed by normal hearing. As a result, speech understanding in listeners with cochlear implants is typically poorer and more effortful than in listeners with normal hearing. The brain networks supporting speech understanding in listeners with cochlear implants are not well understood, partly due to difficulties obtaining functional neuroimaging data in this population. In the current study, we assessed the brain regions supporting spoken word understanding in adult listeners with right unilateral cochlear implants (n=20) and matched controls (n=18) using high-density diffuse optical tomography (HD-DOT), a quiet and non-invasive imaging modality with spatial resolution comparable to that of functional MRI. We found that while listening to spoken words in quiet, listeners with cochlear implants showed greater activity in the left prefrontal cortex than listeners with normal hearing, specifically in a region engaged in a separate spatial working memory task. These results suggest that listeners with cochlear implants require greater cognitive processing during speech understanding than listeners with normal hearing, supported by compensatory recruitment of the left prefrontal cortex.

Regression-based neural network for improving image reconstruction in diffuse optical tomography.

Diffuse optical tomography (DOT) is a non-invasive imaging technique utilizing multi-scattered light at visible and infrared wavelengths to detect anomalies in tissues. However, the DOT image reconstruction is based on solving the inverse problem, which requires massive calculations and time. In this article, for the first time, to the best of our knowledge, a simple, regression-based cascaded feed-forward deep learning neural network is derived to solve the inverse problem of DOT in compressed breast geometry. The predicted data is subsequently utilized to visualize the breast tissues and their anomalies. The dataset in this study is created using a Monte-Carlo algorithm, which simulates the light propagation in the compressed breast placed inside a parallel plate source-detector geometry (forward process). The simulated DL-DOT system's performance is evaluated using the Pearson correlation coefficient (R) and the Mean squared error (MSE) metrics. Although a comparatively smaller dataset (50 nos.) is used, our simulation results show that the developed feed-forward network algorithm to solve the inverse problem delivers an increment of ∼30% over the analytical solution approach, in terms of R. Furthermore, the proposed network's MSE outperforms that of the analytical solution's MSE by a large margin revealing the robustness of the network and the adaptability of the system for potential applications in medical settings.

Multitask Deep Learning Reconstruction and Localization of Lesions in Limited Angle Diffuse Optical Tomography.

Diffuse optical tomography (DOT) leverages near-infrared light propagation through tissue to assess its optical properties and identify abnormalities. DOT image reconstruction is an ill-posed problem due to the highly scattered photons in the medium and the smaller number of measurements compared to the number of unknowns. Limited-angle DOT reduces probe complexity at the cost of increased reconstruction complexity. Reconstructions are thus commonly marred by artifacts and, as a result, it is difficult to obtain an accurate reconstruction of target objects, e.g., malignant lesions. Reconstruction does not always ensure good localization of small lesions. Furthermore, conventional optimization-based reconstruction methods are computationally expensive, rendering them too slow for real-time imaging applications. Our goal is to develop a fast and accurate image reconstruction method using deep learning, where multitask learning ensures accurate lesion localization in addition to improved reconstruction. We apply spatial-wise attention and a distance transform based loss function in a novel multitask learning formulation to improve localization and reconstruction compared to single-task optimized methods. Given the scarcity of real-world sensor-image pairs required for training supervised deep learning models, we leverage physics-based simulation to generate synthetic datasets and use a transfer learning module to align the sensor domain distribution between in silico and real-world data, while taking advantage of cross-domain learning. Applying our method, we find that we can reconstruct and localize lesions faithfully while allowing real-time reconstruction. We also demonstrate that the present algorithm can reconstruct multiple cancer lesions. The results demonstrate that multitask learning provides sharper and more accurate reconstruction.