In-vivo Datasets Research Articles

Overlapping microbubble localization based on multiscale statistical features for ultrasound super-resolution imaging

Ultrasound localization microscopy enables visualization of the microvasculature through the localization and tracking of isolated microbubbles (MBs). Currently, improving the precision for localizing overlapping MBs at high concentration to decrease the acquisition time remains a challenge. This study introduces multiscale statistical feature localization (MSFL) to address this challenge. First, the overlapping MBs are separated by local energy and blobness filters constructed from the Hessian matrix of texture features and its eigenvalues. Then, multiscale and orientation kernels are modeled to adapt to the appearance of various MBs; their amplitude and gradient convolution response results jointly determine the credibility of the MBs. Finally, precise MB localization is obtained by the weighted average of the local maxima of the response maps at each scale to achieve super-resolution (SR) imaging. The proposed method is evaluated using simulation and in-vivo datasets. Simulation results demonstrate that MSFL can adapt to asymmetric Gaussian distributions and improve overlapping MB localization, with a 9.28 % improvement in localization precision and a 5 % greater Jaccard index compared to the Gaussian fitting method. Additionally, the strong robustness of MSFL under noise interference maintains its ability to locate more MBs, and reconstruct the same saturation vessels with 36 % acquisition time reduction compared to Radial symmetry method. For SR images, MSFL can draw smoother interlaced vessels when more overlapping MBs are present, and quickly reconstructs more saturated in vivo microvasculature with a resolution is 10.5 μm(λ/10). These results validate the advantages of MSFL in achieving precise SR imaging with a short acquisition time.

Unsupervised MRI motion artifact disentanglement: introducing MAUDGAN

Objective. This study developed an unsupervised motion artifact reduction method for magnetic resonance imaging (MRI) images of patients with brain tumors. The proposed novel design uses multi-parametric multicenter contrast-enhanced T1W (ceT1W) and T2-FLAIR MRI images. Approach. The proposed framework included two generators, two discriminators, and two feature extractor networks. A 3-fold cross-validation was used to train and fine-tune the hyperparameters of the proposed model using 230 brain MRI images with tumors, which were then tested on 148 patients’ in-vivo datasets. An ablation was performed to evaluate the model’s compartments. Our model was compared with Pix2pix and CycleGAN. Six evaluation metrics were reported, including normalized mean squared error (NMSE), structural similarity index (SSIM), multi-scale-SSIM (MS-SSIM), peak signal-to-noise ratio (PSNR), visual information fidelity (VIF), and multi-scale gradient magnitude similarity deviation (MS-GMSD). Artifact reduction and consistency of tumor regions, image contrast, and sharpness were evaluated by three evaluators using Likert scales and compared with ANOVA and Tukey’s HSD tests. Main results. On average, our method outperforms comparative models to remove heavy motion artifacts with the lowest NMSE (18.34±5.07%) and MS-GMSD (0.07 ± 0.03) for heavy motion artifact level. Additionally, our method creates motion-free images with the highest SSIM (0.93 ± 0.04), PSNR (30.63 ± 4.96), and VIF (0.45 ± 0.05) values, along with comparable MS-SSIM (0.96 ± 0.31). Similarly, our method outperformed comparative models in removing in-vivo motion artifacts for different distortion levels except for MS- SSIM and VIF, which have comparable performance with CycleGAN. Moreover, our method had a consistent performance for different artifact levels. For the heavy level of motion artifacts, our method got the highest Likert scores of 2.82 ± 0.52, 1.88 ± 0.71, and 1.02 ± 0.14 (p-values ≪ 0.0001) for our method, CycleGAN, and Pix2pix respectively. Similar trends were also found for other motion artifact levels. Significance. Our proposed unsupervised method was demonstrated to reduce motion artifacts from the ceT1W brain images under a multi-parametric framework.

Deep learning model for predicting genetic diseases using DNA sequence data

The emergence of machine learning in the recent decade has excelled in determining new potential features and nonlinear relationships existing between the data derived from the DNA sequences of genetic diseases. Machine learning also enhances the process of handling data with maximum predicted variables compared to observations during the data mining process of prediction. In this context, our study presents a deep learning model for predicting Transcription Factor Binding Sites (TFBS) in DNA sequences, with a focus on features within genetic data associated with diseases. Transcription Factors (TFs) play a crucial role in modulating gene expression by binding to TFBS. The accurate prediction of TFBS is essential for understanding genome function and evolution. Thus, we develop an efficient deep learning model that considers TFBS prediction as a nucleotide-level binary classification task. In our proposed model, first we create an input matrix using the original DNA sequences. Next, we encode these DNA sequences using one-hot encoding, representing them as a sequence of numerical values. We then employ three convolutional layers, allowing our model to capture intricate patterns and motif features over a larger spatial range. To capture important features within the DNA sequence and to focus on them, we incorporate an attention layer. Finally, a dense layer, consisting of two fully connected layers and a dropout layer, calculates the probability of TF binding site occurrence based on the features learned by the proposed model. Our experimental results, using in-vivo datasets obtained from Chip-seq, demonstrate the superior performance of our proposed deep learning model in TFBS prediction compared to other existing state-of-the-art methods. The improvement in accuracy is due to additional layers of CNN and then an attention layer in the model. Thus, this result in a better performance of our approach in predicting the transcription factor binding sites and enhancing our understanding of gene regulation and genome function.

A statistical approach for identifying single nucleotide variants that affect transcription factor binding

Non-coding variants located within regulatory elements may alter gene expression by modifying transcription factor (TF) binding sites, thereby leading to functional consequences. Different TF models are being used to assess the effect of DNA sequence variants, such as single nucleotide variants (SNVs). Often existing methods are slow and do not assess statistical significance of results. We investigated the distribution of absolute maximal differential TF binding scores for general computational models that affect TF binding. We find that a modified Laplace distribution can adequately approximate the empirical distributions. A benchmark on invitro and invivo datasets showed that our approach improves upon an existing method in terms of performance and speed. Applications on eQTLs and on a genome-wide association study illustrate the usefulness of our statistics by highlighting cell type-specific regulators and target genes. An implementation of our approach is freely available on GitHub and as bioconda package.

Competitive Swarm Optimized SVD Clutter Filtering for Ultrafast Power Doppler Imaging.

Ultrafast power Doppler imaging (uPDI) can significantly increase the sensitivity of resolving small vascular paths in ultrasound. While clutter filtering is a fundamental and essential method to realize uPDI, it commonly uses singular value decomposition (SVD) to suppress clutter signals and noise. However, current SVD-based clutter filters using two cutoffs cannot ensure sufficient separation of tissue, blood, and noise in uPDI. This article proposes a new competitive swarm-optimized SVD clutter filter to improve the quality of uPDI. Specifically, without using two cutoffs, such a new filter introduces competitive swarm optimization (CSO) to search for the counterparts of blood signals in each singular value. We validate the CSO-SVD clutter filter on public in vivo datasets. The experimental results demonstrate that our method can achieve higher contrast-to-noise ratio (CNR), signal-to-noise ratio (SNR), and blood-to-clutter ratio (BCR) than the state-of-the-art SVD-based clutter filters, showing a better balance between suppressing clutter signals and preserving blood signals. Particularly, our CSO-SVD clutter filter improves CNR by 0.99 ± 0.08 dB, SNR by 0.79 ± 0.08 dB, and BCR by 1.95 ± 0.03 dB when comparing a spatial-similarity-based SVD clutter filter in the in vivo dataset of rat brain bolus.

A parametrization study of TEULM’s image reconstruction performance

Time efficient ultrasound localization microscopy (TEULM) allows imaging the microvascular structures with reduced frame rates compared to standard ULM. The quality of reconstructed images relies on different parameters. In this study, we aim at understanding the links between TEULM performance and three key parameters: SVD k-value (representing the number of components considered as noise), the maximum linking distance (indicating the assumed maximum distance that a microbubble can cover between two successive frames), and the minimum length of a track. We generated TEULM density and dynamics maps, singularly varying these three parameters. Next, we evaluated the quality of TEULM images by means of a root mean squared error (RMSE), dice score, and Fourier ring correlation (FRC) analysis. A publicly available preclinical in-vivo dataset was used for this study. Results demonstrate that minimum track length and k-value have the greatest impact on both density and dynamic maps RMSE, whereas the minimum linking distance is the key factor influencing the dice score. Additionally, the relationship between these parameters and metrics is non-linear. Furthermore, the resolution measured by the FRC is negatively affected by increasing the SVD k-value and by increasing the maximum linking distance but positively affected by increasing the minimum track length.

Advanced waveform analysis of the photoplethysmogram signal using complementary signal processing techniques for the extraction of biomarkers of cardiovascular function.

Photoplethysmogram signals from wearable devices typically measure heart rate and blood oxygen saturation, but contain a wealth of additional information about the cardiovascular system. In this study, we compared two signal-processing techniques: fiducial point analysis and Symmetric Projection Attractor Reconstruction, on their ability to extract new cardiovascular information from a photoplethysmogram signal. The aim was to identify fiducial point analysis and Symmetric Projection Attractor Reconstruction indices that could classify photoplethysmogram signals, according to age, sex and physical activity. Three datasets were used: an in-silico dataset of simulated photoplethysmogram waves for healthy male participants (25-75 years old); an in-vivo dataset containing 10-min photoplethysmogram recordings from 57 healthy subjects at rest (18-39 or > 70 years old; 53% female); and an in-vivo dataset containing photoplethysmogram recordings collected for 4 weeks from a single subject, in daily life. The best-performing indices from the in-silico study (5/48 fiducial point analysis and 6/49 Symmetric Projection Attractor Reconstruction) were applied to the in-vivo datasets. Key fiducial point analysis and Symmetric Projection Attractor Reconstruction indices, which showed the greatest differences between groups, were found to be consistent across datasets. These indices were related to systolic augmentation, diastolic peak positioning and prominence, and waveform variability. Both fiducial point analysis and Symmetric Projection Attractor Reconstruction techniques provided indices that supported the classification of age and physical activity, but not sex. Both fiducial point analysis and Symmetric Projection Attractor Reconstruction techniques demonstrated utility in identifying cardiovascular differences between individuals and within an individual over time. Future research should investigate the potential utility of these techniques for extracting information on fitness and disease, to support healthcare-decision making.

BEAS-Net: a Shape-Prior-Based Deep Convolutional Neural Network for Robust Left Ventricular Segmentation in 2D Echocardiography.

Left ventricle (LV) segmentation of 2D echocardiography images is an essential step in the analysis of cardiac morphology and function and - more generally - diagnosis of cardiovascular diseases. Several deep learning (DL) algorithms have recently been proposed for the automatic segmentation of the LV, showing significant performance improvement over the traditional segmentation algorithms. However, unlike the traditional methods, prior information about the segmentation problem, e.g. anatomical shape information, is not usually incorporated for training the DL algorithms. This can degrade the generalization performance of the DL models on unseen images if their characteristics are somewhat different from those of the training images, e.g. low-quality testing images. In this study, a new shape-constrained deep convolutional neural network (CNN) - called BEAS-Net - is introduced for automatic LV segmentation. The BEAS-Net learns how to associate the image features, encoded by its convolutional layers, with anatomical shape-prior information derived by the B-spline explicit active surface (BEAS) algorithm to generate physiologically meaningful segmentation contours when dealing with artifactual or low-quality images. The performance of the proposed network was evaluated using three different in-vivo datasets and was compared a deep segmentation algorithm based on the U-Net model. Both networks yielded comparable results when tested on images of acceptable quality, but the BEAS-Net outperformed the benchmark DL model on artifactual and low-quality images.

Spherical harmonics-based deep learning achieves generalized and accurate diffusion tensor imaging.

Diffusion tensor imaging (DTI) is a prevalent magnetic resonance imaging (MRI) technique, widely used in clinical and neuroscience research. However, the reliability of DTI is affected by the low signal-to-noise ratio inherent in diffusion-weighted (DW) images. Deep learning (DL) has shown promise in improving the quality of DTI, but its limited generalization to variable acquisition schemes hinders practical applications. This study aims to develop a generalized, accurate, and efficient DL-based DTI method. By leveraging the representation of voxel-wise diffusion MRI (dMRI) signals on the sphere using spherical harmonics (SH), we propose a novel approach that utilizes SH coefficient maps as input to a network for predicting the diffusion tensor (DT) field, enabling improved generalization. Extensive experiments were conducted on simulated and in-vivo datasets, covering various DTI application scenarios. The results demonstrate that the proposed SH-DTI method achieves advanced performance in both quantitative and qualitative analyses of DTI. Moreover, it exhibits remarkable generalization capabilities across different acquisition schemes, centers, and scanners, ensuring its broad applicability in diverse settings.

Microwave ablation modeling with AMICA antenna: Validation by means a numerical analysis

Background and objectivesPercutaneous microwave thermal ablation is based on electromagnetic waves that generate dielectric heating, and it is widely recognized as one of the mostly used techniques for tumor treatment. The aim of this work is to validate a predictive model capable of providing physicians with guidelines to be used during thermal ablation procedures avoiding collateral damage. MethodsA finite element commercial software, COMSOL Multiphysics, is employed to implement a tuning-parameter approach. Governing equations are written with reference to variable-porosity and Local Thermal Non-Equilibrium (LTNE) equations are employed. The simulations results are compared with available ex-vivo and in-vivo data with the help of regression analysis. For in-vivo data simulations, velocity vector modulus and direction are varied between 0.0007 and 0.0009 m/s and 90–270°, respectively, in order to use this parameter as a tuning one to simulate – and lately optimize with respect to the differences from experimental outcomes - all the possible directions of the blood flow with respect to the antenna, whose insertion angle is not registered in the dataset. ResultsThe model is validated using reference data provided by the manufacturer (AMICA), which is obtained from ex-vivo bovine liver. The model accurately predicts the size and shape of the ablated area, resulting in an overestimation lesser than 10 %. Additionally, predictive data are compared to an in-vivo dataset. The ablated volume is accurately predicted with a mean underestimation of 6 %. The sphericity index is calculated as 0.75 and 0.62 for the predictions and in-vivo data, respectively. ConclusionThis study developed a predictive model for microwave ablation of liver tumors that showed good performance in predicting ablation dimensions and sphericity index for ex-vivo bovine liver and for in-vivo human liver data with the tuning technique. The study emphasizes the necessity for additional development and validation to enhance the accuracy and reliability of in-vivo application.

Laparoscopic Image-based Critical Action Recognition and Anticipation with Explainable Features.

Surgical workflow analysis integrates perception, comprehension, and prediction of the surgical workflow, which helps real-time surgical support systems provide proper guidance and assistance for surgeons. This paper promotes the idea of critical actions, which refer to the essential surgical actions that progress towards the fulfillment of the operation. Fine-grained workflow analysis involves recognizing current critical actions and previewing the moving tendency of instruments in the early stage of critical actions. Aiming at this, we propose a framework that incorporates operational experience to improve the robustness and interpretability of action recognition in in-vivo situations. High-dimensional images are mapped into an experience-based explainable feature space with low-dimension to achieve critical action recognition through a hierarchical classification structure. To forecast the instrument's motion tendency, we model the motion primitives in the polar coordinate system (PCS) to represent patterns of complex trajectories. Given the laparoscopy variance, the adaptive pattern recognition (APR) method, which adapts to uncertain trajectories by modifying model parameters, is designed to improve prediction accuracy. The in-vivo dataset validations show that our framework fulfilled the surgical awareness tasks with exceptional accuracy and real-time performance.

Frame rate improvement in ultrafast coherent plane wave compounding

Coherent plane wave compounding (CPWC), as an ultrafast ultrasound imaging technique, makes a significant breakthrough in frame rate enhancement. However, there exists a compromise between the quality of the final image and the frame rate in CPWC. In this paper, we propose an efficient method to minimize the number of required emissions, and consequently, improve the frame rate, while maintaining the image quality. To this end, we down-sample the angle interval using two specific sampling factors. More precisely, we construct two different subsets, each of which consists of a few numbers of emissions. The optimal values of the angle intervals are achieved based on the beampattern that corresponds to the reference case (that is, the case where all plane waves are used). Finally, in order to keep the image quality comparable with the reference case, we apply some modifications to the image reconstruction procedure. In the proposed algorithm, the Delay-and-Sum beamformed images of two considered subsets are convolved to achieve the final reconstructed image. The obtained results confirm the efficiency of the proposed method in terms of frame rate improvement compared to the reference case. In particular, by using the proposed method, the required emissions in PICMUS data reduce to 16, which is 4.6 times smaller compared to the reference case. Also, the gCNR values of the proposed method and the reference case are obtained as 0.98 and 0.97, respectively, for in-vivo dataset. This demonstrates that the proposed method successfully preserves the quality of the reconstructed image by using much fewer emissions.

Relevance and utility of the in-vivo and ex-vivo optical properties of the skin reported in the literature: a review [Invited

Imaging non-invasively into the human body is currently limited by cost (MRI and CT scan), image resolution (ultrasound), exposure to ionising radiation (CT scan and X-ray), and the requirement for exogenous contrast agents (CT scan and PET scan). Optical imaging has the potential to overcome all these issues but is currently limited by imaging depth due to the scattering and absorption properties of human tissue. Skin is the first barrier encountered by light when imaging non-invasively, and therefore a clear understanding of the way that light interacts with skin is required for progress on optical medical imaging to be made. Here we present a thorough review of the optical properties of human skin measured in-vivo and compare these to the previously collated ex-vivo measurements. Both in-vivo and ex-vivo published data show high inter- and intra-publication variability making definitive answers regarding optical properties at given wavelengths challenging. Overall, variability is highest for ex-vivo absorption measurements with differences of up to 77-fold compared with 9.6-fold for the in-vivo absorption case. The impact of this variation on optical penetration depth and transport mean free path is presented and potential causes of these inconsistencies are discussed. We propose a set of experimental controls and reporting requirements for future measurements. We conclude that a robust in-vivo dataset, measured across a broad spectrum of wavelengths, is required for the development of future technologies that significantly increase the depth of optical imaging.

Development of a Novel Tool To Demystify Drug Distribution at Tissue-Blood Barriers.

Tissue endothelial cells express ABC-transporter enzymes that change the concentration of small molecules within different tissue compartments. These "blood-tissue barriers" have been shown to directly affect the efficacy and toxicity of anticancer, antimicrobial, psychiatric, and anti-epileptic drugs. Currently this phenomenon is best studied for the blood-brain barrier, but remains enigmatic for most other tissues. In addition, canonical pharmacokinetic theory specifically assumes an equal concentration of free drug within all tissue compartments. Inspired by Lipinski's "rule of 5," we here clarify current knowledge on drug-tissue distribution by: 1) curating the in-vivo literature on 73 drugs across 23 tissues and 2) developing two graphical web-based applications to visually describe and interpret data. These curated in-vivo dataset and visualization tools enabled us to achieve new insights into the logic of the barrier-tissue organization and showed remarkable correspondence to whole-body imaging of radiolabeled molecules.

Proteomic and phosphoproteomic characteristics of the cortex, hippocampus, thalamus, lung, and kidney in COVID-19-infected female K18-hACE2 mice.

Neurological damage caused by coronavirus disease 2019 (COVID-19) has attracted increasing attention. Recently, through autopsies of patients with COVID-19, the direct identification of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in their central nervous system (CNS) has been reported, indicating that SARS-CoV-2 might directly attack the CNS. The need to prevent COVID-19-induced severe injuries and potential sequelae is urgent, requiring the elucidation of large-scale molecular mechanisms invivo. In this study, we performed liquid chromatography-mass spectrometry-based proteomic and phosphoproteomic analyses of the cortex, hippocampus, thalamus, lungs, and kidneys of SARS-CoV-2-infected K18-hACE2 female mice. We then performed comprehensive bioinformatic analyses, including differential analyses, functional enrichment, and kinase prediction, to identify key molecules involved in COVID-19. We found that the cortex had higher viral loads than did the lungs, and the kidneys did not have SARS-COV-2. After SARS-CoV-2 infection, RIG-I-associated virus recognition, antigen processing and presentation, and complement and coagulation cascades were activated to different degrees in all five organs, especially the lungs. The infected cortex exhibited disorders of multiple organelles and biological processes, including dysregulated spliceosome, ribosome, peroxisome, proteasome, endosome, and mitochondrial oxidative respiratory chain. The hippocampus and thalamus had fewer disorders than did the cortex; however, hyperphosphorylation of Mapt/Tau, which may contribute to neurodegenerative diseases, such as Alzheimer's disease, was found in all three brain regions. Moreover, SARS-CoV-2-induced elevation of human angiotensin-converting enzyme 2 (hACE2) was observed in the lungs and kidneys, but not in the three brain regions. Although the virus was not detected, the kidneys expressed high levels of hACE2 and exhibited obvious functional dysregulation after infection. This indicates that SARS-CoV-2 can cause tissue infections or damage via complicated routes. Thus, the treatment of COVID-19 requires a multipronged approach. This study provides observations and invivo datasets for COVID-19-associated proteomic and phosphoproteomic alterations in multiple organs, especially cerebral tissues, of K18-hACE2 mice. In mature drug databases, the differentially expressed proteins and predicted kinases in this study can be used as baits to identify candidate therapeutic drugs for COVID-19. This study can serve as a solid resource for the scientific community. The data in this manuscript will serve as a starting point for future research on COVID-19-associated encephalopathy. This study was supported by grants from the Chinese Academy of Medical Sciences Innovation Fund for Medical Sciences, the National Natural Science Foundation of China, and the Natural Science Foundation of Beijing.

Reconstruction of Quantitative Susceptibility Mapping from Total Field Maps with Local Field Maps Guided UU-Net.

Quantitative susceptibility mapping (QSM) is an emerging computational technique based on the magnetic resonance imaging (MRI) phase signal, which can provide magnetic susceptibility values of tissues. The existing deep learning-based models mainly reconstruct QSM from local field maps. However, the complicated inconsecutive reconstruction steps not only accumulate errors for inaccurate estimation, but also are inefficient in clinical practice. To this end, a novel local field maps guided UU-Net with Self- and Cross-Guided Transformer (LGUU-SCT-Net) is proposed to reconstruct QSM directly from the total field maps. Specifically, we propose to additionally generate the local field maps as the auxiliary supervision during the training stage. This strategy decomposes the more complicated mapping from total maps to QSM into two relatively easier ones, effectively alleviating the difficulty of direct mapping. Meanwhile, an improved U-Net model, named LGUU-SCT-Net, is further designed to promote the nonlinear mapping ability. The long-range connections are designed between two sequentially stacked U-Nets to bring more feature fusions and facilitate the information flow. The Self- and Cross-Guided Transformer integrated into these connections further captures multi-scale channel-wise correlations and guides the fusion of multiscale transferred features, assisting in the more accurate reconstruction. The experimental results on an in-vivo dataset demonstrate the superior reconstruction results of our proposed algorithm.

A tailor-made 3-dimensional directional Haar semi-tight framelet for pMRI reconstruction

In this paper, we propose a model for parallel magnetic resonance imaging (pMRI) reconstruction, regularized by a carefully designed tight framelet system, that can lead to reconstructed images with much less artifacts in comparison to those from existing models. Our model is motivated from the observations that each receiver coil in a pMRI system is more sensitive to the specific object nearest to the coil, and all coil images are correlated. To exploit these observations, we first stack all coil images together as a 3-dimensional (3D) data matrix, and then design a 3D directional Haar tight framelet (3DHTF) to represent it. After analyzing sparse information of the coil images provided by the high-pass filters of the 3DHTF, we separate the high-pass filters into effective ones and ineffective ones, and we then devise a 3D directional Haar semi-tight framelet (3DHSTF) from the 3DHTF by replacing its ineffective filters with only one filter. This 3DHSTF is tailor-made for coil images, meanwhile, giving a significant saving in computation comparing to the 3DHTF. With the 3DHSTF, we propose an ℓ 1 -3DHSTF model for pMRI reconstruction. Numerical experiments for MRI phantom and in-vivo data sets are provided to demonstrate the superiority of our ℓ 1 -3DHSTF model in terms of the efficiency of reducing aliasing artifacts in the reconstructed images.

Estimating Echocardiographic Myocardial Strain of Left Ventricle with Deep Learning.

The global longitudinal strain of the myocardial tissue has been shown to be a better indicator of cardiac pathologies in the subclinical stage than other indices, such as the ejection fraction. This article presents a new deep learning approach for strain estimation in 2D echocardiograms. The proposed method improves the performance of the state of the art without losing stability with noisy echocardiograms and achieved an average end point error of 0.14 ± 0.17 pixels in the estimation of the optical flow in the myocardium and an error of 1.34 ± 2.34 % in the estimation of the global longitudinal strain indicator when evaluated in a synthetic echocardiographic dataset. Further research will validate the proposed method by a clinical in-vivo dataset. Clinical relevance- This paper presents a method to estimate the global longitudinal strain index in noisy echocardiograms, which promises to be a better indicator of cardiac pathologies in the subclinical stage than other indices such as the ejection fraction.

Medical Instrument Segmentation in 3D US by Hybrid Constrained Semi-Supervised Learning.

Medical instrument segmentation in 3D ultrasound is essential for image-guided intervention. However, to train a successful deep neural network for instrument segmentation, a large number of labeled images are required, which is expensive and time-consuming to obtain. In this article, we propose a semi-supervised learning (SSL) framework for instrument segmentation in 3D US, which requires much less annotation effort than the existing methods. To achieve the SSL learning, a Dual-UNet is proposed to segment the instrument. The Dual-UNet leverages unlabeled data using a novel hybrid loss function, consisting of uncertainty and contextual constraints. Specifically, the uncertainty constraints leverage the uncertainty estimation of the predictions of the UNet, and therefore improve the unlabeled information for SSL training. In addition, contextual constraints exploit the contextual information of the training images, which are used as the complementary information for voxel-wise uncertainty estimation. Extensive experiments on multiple ex-vivo and in-vivo datasets show that our proposed method achieves Dice score of about 68.6%-69.1% and the inference time of about 1 sec. per volume. These results are better than the state-of-the-art SSL methods and the inference time is comparable to the supervised approaches.

Integrated deep learning framework for accelerated optical coherence tomography angiography

Label-free optical coherence tomography angiography (OCTA) has become a premium imaging tool in clinics to obtain structural and functional information of microvasculatures. One primary technical drawback for OCTA, however, is its imaging speed. The current protocols require high sampling density and multiple acquisitions of cross-sectional B-scans to form one image frame, resulting in low acquisition speed. Recently, deep learning (DL)-based methods have gained attention in accelerating the OCTA acquisition process. They achieve faster acquisition using two independent reconstructing approaches: high-quality angiograms from a few repeated B-scans and high-resolution angiograms from undersampled data. While these approaches have shown promising results, they provide limited solutions that only partially account for the OCTA scanning mechanism. Herein, we propose an integrated DL method to simultaneously tackle both factors and further enhance the reconstruction performance in speed and quality. We designed an end-to-end deep neural network (DNN) framework with a two-staged adversarial training scheme to reconstruct fully-sampled, high-quality (8 repeated B-scans) angiograms from their corresponding undersampled, low-quality (2 repeated B-scans) counterparts by successively enhancing the pixel resolution and the image quality. Using an in-vivo mouse brain vasculature dataset, we evaluate our proposed framework through quantitative and qualitative assessments and demonstrate that our method can achieve superior reconstruction performance compared to the conventional means. Our DL-based framework can accelerate the OCTA imaging speed from 16 to 256times while preserving the image quality, thus enabling a convenient software-only solution to enhance preclinical and clinical studies.