3D Vessel Structures Research Articles

Investigation of 3D vessel reconstruction under Doppler imaging with phantoms: Towards reconstruction of the Circle of Willis

BackgroundStroke is the second leading cause of death across the globe. Early screening and risk detection could provide early intervention and possibly prevent its incidence. Imaging modalities, including 1D-Transcranial Doppler Ultrasound (1D-TCD) or Transcranial Color-code sonography (TCCS), could only provide low spatial resolution or 2D image information, respectively. Notably, 3D imaging modalities including CT have high radiation exposure, whereas MRI is expensive and cannot be adopted in patients with implanted devices. This study proposes an alternative imaging solution for reconstructing 3D Doppler ultrasound geared towards providing a screening tool for the 3D vessel structure of the brain. MethodsThe system comprises an ultrasound phased array attached to a servo motor, which can rotate 180˚ at a speed of 2˚/s. We extracted the color Doppler ROI from the image before reconstructing it into a 3D view using a customized pixel-based algorithm. Different vascular diameters, flow velocity, and depth were tested using a vascular phantom with a pumped flow to confirm the system for imaging blood flow. These variables were set to mimic the vessel diameter, flow speed, and depth of the Circle of Willis (CoW) during a transcranial screening. Results and conclusionsThe lower values of absolute error and ratio were found in the larger vascular channels, and vessel diameter overrepresentation was observed. Under different flow velocities, such diameter overrepresentation in the reconstructed flow did not change much; however, it did change with different depths. Meanwhile, the setting of the velocity scale and the color gain affected the dimension of reconstructed objectives. Moreover, we presented a 3D image of CoW from a subject to demonstrate its potential. The findings of this work can provide a good reference for further studies on the reconstruction of the CoW or other blood vessels using Doppler imaging.

A semiparametric Gaussian mixture model for chest CT-based 3D blood vessel reconstruction.

Computed tomography (CT) has been a powerful diagnostic tool since its emergence in the 1970s. Using CT data, 3D structures of human internal organs and tissues, such as blood vessels, can be reconstructed using professional software. This 3D reconstruction is crucial for surgical operations and can serve as a vivid medical teaching example. However, traditional 3D reconstruction heavily relies on manual operations, which are time-consuming, subjective, and require substantial experience. To address this problem, we develop a novel semiparametric Gaussian mixture model tailored for the 3D reconstruction of blood vessels. This model extends the classical Gaussian mixture model by enabling nonparametric variations in the component-wise parameters of interest according to voxel positions. We develop a kernel-based expectation-maximization algorithm for estimating the model parameters, accompanied by a supporting asymptotic theory. Furthermore, we propose a novel regression method for optimal bandwidth selection. Compared to the conventional cross-validation-based (CV) method, the regression method outperforms the CV method in terms of computational and statistical efficiency. In application, this methodology facilitates the fully automated reconstruction of 3D blood vessel structures with remarkable accuracy.

Self-supervised medical slice interpolation network using controllable feature flow

Deep learning-based image interpolation methods are confronted with various challenges in its application to anisotropic medical volumetric data (i.e., CT and MR images) out of the complex nonlinear deformation and the scarcity of high-quality images. This paper proposes a self-supervised multiple medical slice interpolation network using controllable feature flow. Firstly, a Controllable Feature Flow Network (CFFNet) is proposed to estimate the complex nonlinear deformation of medical images. In CFFNet, we first use a deformation-aware network and a spatial channel modulation module to predict the bi-directional feature flows from the source slices to the target slices by considering an additional position parameter, and then the learned feature flows are used to synthesize the target intermediate features by using the deformable convolution. Secondly, an improved two-stage self-supervised framework is proposed to alleviate the dependency on high-quality medical images. In the first stage, the synthesized training pairs along the dense sagittal and coronal directions are adopted to pre-train the CFFNet. In the second one, the sparse axial slices are used to fine-tune the CFFNet with the cycle-consistency constraint and a feature domain smooth loss. Experimental results illustrate that the proposed CFFNet assumes superior performance on medical slice interpolation tasks with fewer parameters and the proposed self-supervised CFFNet obtains competitive results compared with other fully supervised algorithms. We also evaluate our model on HepaticVessel segmentation task, the proposed method can effectively enhance the continuity and smoothness of the 3D vessel structures, showing promising results on clinical application. The code is available at: https://codeocean.com/capsule/8945297/tree/v1.

Graph Convolution Based Cross-Network Multiscale Feature Fusion for Deep Vessel Segmentation.

Vessel segmentation is widely used to help with vascular disease diagnosis. Vessels reconstructed using existing methods are often not sufficiently accurate to meet clinical use standards. This is because 3D vessel structures are highly complicated and exhibit unique characteristics, including sparsity and anisotropy. In this paper, we propose a novel hybrid deep neural network for vessel segmentation. Our network consists of two cascaded subnetworks performing initial and refined segmentation respectively. The second subnetwork further has two tightly coupled components, a traditional CNN-based U-Net and a graph U-Net. Cross-network multi-scale feature fusion is performed between these two U-shaped networks to effectively support high-quality vessel segmentation. The entire cascaded network can be trained from end to end. The graph in the second subnetwork is constructed according to a vessel probability map as well as appearance and semantic similarities in the original CT volume. To tackle the challenges caused by the sparsity and anisotropy of vessels, a higher percentage of graph nodes are distributed in areas that potentially contain vessels while a higher percentage of edges follow the orientation of potential nearby vessels. Extensive experiments demonstrate our deep network achieves state-of-the-art 3D vessel segmentation performance on multiple public and in-house datasets.

Vascular regression in the kidney: changes in 3D vessel structure with time post-irradiation.

Though angiogenesis has been investigated in depth, vascular regression and rarefaction remain poorly understood. Regression of renal vasculature accompanies many pathological states such as diabetes, hypertension, atherosclerosis, and radiotherapy. Radiation decreases microvessel density in multiple organs, though the mechanism is not known. By using a whole animal (rat) model with a single dose of partial body irradiation to the kidney, changes in the volume of renal vasculature were recorded at two time points, 60 and 90 days after exposure. Next, a novel vascular and metabolic imaging (VMI) technique was used to computationally assess 3D vessel diameter, volume, branch depth, and density over multiple levels of branching down to 70 µm. Four groups of rats were studied, of which two groups received a single dose of 12.5 Gy X-rays. The kidneys were harvested after 60 or 90 days from one irradiated and one non-irradiated group at each time point. Measurements of the 3D vasculature showed that by day-90 post-radiation, when renal function is known to deteriorate, total vessel volume, vessel density, maximum branch depth, and the number of terminal points in the kidneys decreased by 55%, 57%, 28%, and 53%, respectively. Decreases in the same parameters were not statistically significant at 60 days post-irradiation. Smaller vessels with internal diameters of 70-450 µm as well as large vessels of diameter 451-850 µm, both decreased by 90 days post-radiation. Vascular regression in the lungs of the same strain of irradiated rats has been reported to occur before 60 days supporting the hypothesis that this process is regulated in an organ-specific manner and occurs by a concurrent decrease in luminal diameters of small as well as large blood vessels.

Multiplexed Ultrasound Imaging Using Spectral Analysis on Gas Vesicles.

Current advances in ultrasound imaging techniques combined with the next generation contrast agents such as gas vesicles (GV) revolutionize the visualization of biological tissues with spatiotemporal precision. In optics, fluorescent proteins enable understanding of molecular and cellular functions in biological systems due to their multiplexed imaging capability. Here, a panel of GVs is investigated using mid-band fit (MBF) spectral imaging to realize multiplexed ultrasound imaging to uniquely visualize locations of different types of stationary GVs. The MBF spectral imaging technique demonstrates that stationary clustered GVs are efficiently localized and distinguished from unclustered GVs in agarose gel phantom and 3D vessel structures are visualized in ex vivo mouse liver specimens. Mouse macrophages serve as carriers of clustered and unclustered GVs and multiplexing beacons to report cells' spatial locations by emitting distinct spectral signals. 2D MBF spectral images are reconstructed, and pixels in these images are classified depending on MBF values by comparing predetermined filters that predict the existence of cells with clustered and unclustered GVs. This pseudo-coloring scheme clearly distinguishes the locations of two classes of cells like pseudo-color images in fluorescence microscopy.

Robust Deep 3D Blood Vessel Segmentation Using Structural Priors.

Deep learning has enabled significant improvements in the accuracy of 3D blood vessel segmentation. Open challenges remain in scenarios where labeled 3D segmentation maps for training are severely limited, as is often the case in practice, and in ensuring robustness to noise. Inspired by the observation that 3D vessel structures project onto 2D image slices with informative and unique edge profiles, we propose a novel deep 3D vessel segmentation network guided by edge profiles. Our network architecture comprises a shared encoder and two decoders that learn segmentation maps and edge profiles jointly. 3D context is mined in both the segmentation and edge prediction branches by employing bidirectional convolutional long-short term memory (BCLSTM) modules. 3D features from the two branches are concatenated to facilitate learning of the segmentation map. As a key contribution, we introduce new regularization terms that: a) capture the local homogeneity of 3D blood vessel volumes in the presence of biomarkers; and b) ensure performance robustness to domain-specific noise by suppressing false positive responses. Experiments on benchmark datasets with ground truth labels reveal that the proposed approach outperforms state-of-the-art techniques on standard measures such as DICE overlap and mean Intersection-over-Union. The performance gains of our method are even more pronounced when training is limited. Furthermore, the computational cost of our network inference is among the lowest compared with state-of-the-art.

Kinetic equilibrium reconstructions of plasmas in the MAST database and preparation for reconstruction of the first plasmas in MAST upgrade

Reconstructions of plasma equilibria using magnetic sensors were routine during operation of the Mega Ampere Spherical Tokamak (MAST) device, but reconstructions using kinetic profiles were not. These are necessary for stability and disruption analysis of the MAST database, as well as for operation in the upgrade to the device, MAST-U. The three-dimensional (3D) code VALEN is used to determine eddy currents in the 3D vessel structures for vacuum coil test shots, which are then mapped to effective resistances in the two-dimensional vessel groupings in the EFIT equilibrium reconstruction code to be used in conjunction with nearby loop voltage measurements for estimated currents in the structures during reconstruction. Kinetic equilibrium reconstructions with EFIT, using all available magnetic sensors as well as Thomson scattering measurements of electron temperature and density, charge exchange recombination spectroscopy measurements of ion temperature, and internal magnetic field pitch angle measurements from a motional Stark effect (MSE) diagnostic are performed for a large database of MAST discharges. Excellent convergence errors are obtained for the portions of the discharges where the stored energy was not too low, and it is found that reconstructions performed with temperature and density measurements but without MSE data usually already match the pitch angle measurements well. A database of 275 kinetic equilibria is used to test the ideal MHD stability calculation capability for MAST. Finally, the necessary changes to conducting structure in VALEN, and diagnostic setup in EFIT have been completed for the upgrade from MAST to MAST-U, enabling kinetic reconstructions to commence from the first plasma discharges of the upgraded device.

VC-Net: Deep Volume-Composition Networks for Segmentation and Visualization of Highly Sparse and Noisy Image Data.

The fundamental motivation of the proposed work is to present a new visualization-guided computing paradigm to combine direct 3D volume processing and volume rendered clues for effective 3D exploration. For example, extracting and visualizing microstructures in-vivo have been a long-standing challenging problem. However, due to the high sparseness and noisiness in cerebrovasculature data as well as highly complex geometry and topology variations of micro vessels, it is still extremely challenging to extract the complete 3D vessel structure and visualize it in 3D with high fidelity. In this paper, we present an end-to-end deep learning method, VC-Net, for robust extraction of 3D microvascular structure through embedding the image composition, generated by maximum intensity projection (MIP), into the 3D volumetric image learning process to enhance the overall performance. The core novelty is to automatically leverage the volume visualization technique (e.g., MIP - a volume rendering scheme for 3D volume images) to enhance the 3D data exploration at the deep learning level. The MIP embedding features can enhance the local vessel signal (through canceling out the noise) and adapt to the geometric variability and scalability of vessels, which is of great importance in microvascular tracking. A multi-stream convolutional neural network (CNN) framework is proposed to effectively learn the 3D volume and 2D MIP feature vectors, respectively, and then explore their inter-dependencies in a joint volume-composition embedding space by unprojecting the 2D feature vectors into the 3D volume embedding space. It is noted that the proposed framework can better capture the small/micro vessels and improve the vessel connectivity. To our knowledge, this is the first time that a deep learning framework is proposed to construct a joint convolutional embedding space, where the computed vessel probabilities from volume rendering based 2D projection and 3D volume can be explored and integrated synergistically. Experimental results are evaluated and compared with the traditional 3D vessel segmentation methods and the state-of-the-art in deep learning, by using extensive public and real patient (micro- )cerebrovascular image datasets. The application of this accurate segmentation and visualization of sparse and complicated 3D microvascular structure facilitated by our method demonstrates the potential in a powerful MR arteriogram and venogram diagnosis of vascular disease.

PAvessel

The blood vessels are the most critical part of the human circulatory system. Information acquired on the structure and status of blood vessels drives the development of numerous medical and biometric applications. Therefore, it is of paramount importance to find an effective way to sense vessel structures. Traditional methods, including infrared and Doppler sensing modalities, are limited by optical diffusion and ultrasonic scattering that are not good at vessel sensing with high performance. In comparison, we argue photoacoustic (PA) sensing is an emerging technique that can image 3D vessel structure deep in tissue with high-resolution visualization, maintaining the advantages of both optical and ultrasound methods. In this work, we propose and develop PAvessel, a practical 3D vessel structure sensing system based on PA effects. The entire sensing system comprises two key components, PA sensing hardware and PA sensing software. Specifically, the hardware mainly consists of a linear ultrasound transducer array, an ultrasound data acquisition system, and a neodymium-doped yttrium aluminum garnet (Nd:YAG) laser. After receiving the PA raw data, we use the advanced image reconstruction and 3D photoacoustic vein model to establish the 3D vessel structure model. We validated its effectiveness, cost-effectiveness and high resolution of PAvessel in the evaluation. The system achieves 52% higher signal-to-noise ratio (SNR) compared to the other methods. Furthermore, considering the 3D palm vein contains high dimensional human features and is almost impossible to forge, we also explored its applications in palm biometrics. In a pilot study with 10 participants, PAvessel, combined with a 3D vessel structure matching algorithm (EMD-VT), has proven to possess high accuracy and robustness as a biometric. PAvessel achieves the precision and recall of 98.33% and 97.37%, respectively.

A novel micro-CT-based method to monitor the morphology of blood vessels in the rabbit endplate

The aim of this study was to develop a novel method for observing the morphology of the blood vessels in the rabbit endplate. Twenty 6-month-old rabbits were used in this study. The blood vessels in the L5 endplate in Group A were injected with iohexol and Group B with barium sulfate. Group C was the control group with saline. To optimize the study, Group B was divided into two subgroups: Group B-1 was injected with 100% (w/v) barium sulfate and Group B-2 with 50% (w/v). After injection, the L4-L5 vertebral body was excised and the cranial endplate of L5 was scanned using a micro-CT scanner. Models of the vertebral endplate and vessels were reconstructed using the 3D reconstruction software (Mimics 16.0) by calculating a bone threshold value, and then merged these two models to create a superimposed model. The 3D vessel models could not be created in Groups A and C, but they were clearly visualized in Group B. In the 3D model, the blood vessels extended from the subchondral bone to the endplate, and the density of the blood vessels in the area of the nucleus pulposus (NP) was higher than in the annulus fibrosus. The results of this study suggest that the blood vessels in the rabbit endplate can be clearly observed by the method described using barium sulfate [the 50% (w/v) gave better results compared with the 100% (w/v)]. The information from the 3D vessel structure could provide essential data to help us understand the nutrient pathways within the vertebral endplate.

A Framework for 3D Vessel Analysis using Whole Slide Images of Liver Tissue Sections.

Three-dimensional (3D) high resolution microscopic images have high potential for improving the understanding of both normal and disease processes where structural changes or spatial relationship of disease features are significant. In this paper, we develop a complete framework applicable to 3D pathology analytical imaging, with an application to whole slide images of sequential liver slices for 3D vessel structure analysis. The analysis workflow consists of image registration, segmentation, vessel cross-section association, interpolation, and volumetric rendering. To identify biologically-meaningful correspondence across adjacent slides, we formulate a similarity function for four association cases. The optimal solution is then obtained by constrained Integer Programming. We quantitatively and qualitatively compare our vessel reconstruction results with human annotations. Validation results indicate a satisfactory concordance as measured both by region-based and distance-based metrics. These results demonstrate a promising 3D vessel analysis framework for whole slide images of liver tissue sections.

LIVER WHOLE SLIDE IMAGE ANALYSIS FOR 3D VESSEL RECONSTRUCTION.

The emergence of digital pathology has enabled numerous quantitative analyses of histopathology structures. However, most pathology image analyses are limited to two-dimensional datasets, resulting in substantial information loss and incomplete interpretation. To address this, we have developed a complete framework for three-dimensional whole slide image analysis and demonstrated its efficacy on 3D vessel structure analysis with liver tissue sections. The proposed workflow includes components on image registration, vessel segmentation, vessel cross-section association, object interpolation, and volumetric rendering. For 3D vessel reconstruction, a cost function is formulated based on shape descriptors, spatial similarity and trajectory smoothness by taking into account four vessel association scenarios. An efficient entropy-based Relaxed Integer Programming (eRIP) method is proposed to identify the optimal inter-frame vessel associations. The reconstructed 3D vessels are both quantitatively and qualitatively validated. Evaluation results demonstrate high efficiency and accuracy of the proposed method, suggesting its promise to support further 3D vessel analysis with whole slide images.

Endothelial invasive response in a co-culture model with physically-induced osteodifferentiation

Manipulation of stem cells using physicochemical stimuli has emerged as an important tool in regenerative medicine. While 2D substrates with tunable elasticity have been studied for control of stem cell differentiation, we recently developed a stratified co-culture model of angiogenesis of human mesenchymal stem cells (hMSCs) that differentiate on a tunable polydimethylsiloxane (PDMS) substrate, thereby creating a physiologic context for elasticity-induced differentiation. Endothelial cells (EC) were cultured on top of the hMSC construct on a collagen gel to monitor network formation. Media composition influenced EC invasion due to the conditioning media, the reduction of serum and supplemental growth factors, and the addition of recombinant growth factors. Conditioned media, recombinant growth factors and direct co-culture were compared for endothelial cell invasive response using quantitative image analysis. As anticipated, use of recombinant vascular endothelial growth factor (VEGF) induced the deepest EC invasions while direct co-culture caused shallow invasions compared to other conditions. However, endothelial cells displayed lumen-like morphology, suggesting that cell-cell interaction in the co-culture model could mimic sprouting behaviour. In summary, an engineered suitable biochemical and physical environment facilitated endothelial cells to form 3D vessel structures onto hMSCs. These structures were plated on a stiff surface known to induce osteodifferentiation of stem cells. This low cost co-culture system, with its minimal chemical supplementation and physically controllable matrix, could potentially model in vivo potential in engineered and pre-vascularized bone grafts.

A realistic elastic rod model for real-time simulation of minimally invasive vascular interventions

Simulating intrinsic deformation behaviors of guidewire and catheters for interventional radiology (IR) procedures, such as minimally invasive vascular interventions is a challenging task. Especially real-time simulations for interactive training systems require not only the accuracy of guidewire manipulations, but also the efficiency of computations. The insertion of guidewires and catheters is an essential task for IR procedures and the success of these procedures depends on the accurate navigation of guidewires in complex 3D blood vessel structures to a clinical target, whilst avoiding complications or mistakes of damaging vital tissues and blood vessel walls. In this paper, a novel elastic model for modeling guidewires is presented and evaluated. Our interactive guidewire simulator models the medical instrument as thin flexible elastic rods with arbitrary cross sections, treating the centerline as dynamic and the deformation as quasi-static. Constraints are used to enforce inextensibility of guidewires, providing an efficient computation for bending and twisting modes of the physically-based simulation model. We demonstrate the effectiveness of the new model with a number of simulation examples.

Multiple hypothesis template tracking of small 3D vessel structures

A multiple hypothesis tracking approach to the segmentation of small 3D vessel structures is presented. By simultaneously tracking multiple hypothetical vessel trajectories, low contrast passages can be traversed, leading to an improved tracking performance in areas of low contrast. This work also contributes a novel mathematical vessel template model, with which an accurate vessel centerline extraction is obtained. The tracking is fast enough for interactive segmentation and can be combined with other segmentation techniques to form robust hybrid methods. This is demonstrated by segmenting both the liver arteries in CT angiography data, which is known to pose great challenges, and the coronary arteries in 32 CT cardiac angiography data sets in the Rotterdam Coronary Artery Algorithm Evaluation Framework, for which ground-truth centerlines are available.

Deformable 2D-3D Registration of Vascular Structures in a One View Scenario

Alignment of angiographic 3D scans to 2D projections is an important issue for 3D depth perception and navigation during interventions. Currently, in a setting where only one 2D projection is available, methods employing a rigid transformation model present the state of the art for this problem. In this work, we introduce a method capable of deformably registering 3D vessel structures to a respective single projection of the scene. Our approach addresses the inherent ill-posedness of the problem by incorporating a priori knowledge about the vessel structures into the formulation. We minimize the distance between the 2D points and corresponding projected 3D points together with regularization terms encoding the properties of length preservation of vessel structures and smoothness of deformation. We demonstrate the performance and accuracy of the proposed method by quantitative tests on synthetic examples as well as real angiographic scenes.