Voxel Shape Research Articles

Kiosk 10R-TA-09 - Evaluation of Voxel Volume and Shape for Cardiac Diffusion Tensor Imaging

Kiosk 10R-TA-09 - Evaluation of Voxel Volume and Shape for Cardiac Diffusion Tensor Imaging

Extraction of Urban Road Boundary Points from Mobile Laser Scanning Data Based on Cuboid Voxel

The accuracy of point cloud processing results is greatly dependent on the determination of the voxel size and shape during the point cloud voxelization process. Previous studies predominantly set voxel sizes based on point cloud density or the size of ground objects. Voxels are mostly considered square in shape by default. However, conventional square voxels are not applicable to all surfaces. This study proposes a method of using cuboid voxels to extract urban road boundary points using curb points as road boundary points. In comparison with conventional cubic voxels, cuboid voxels reduce the probability of mixed voxels at the road curb, highlight two geometric features of road curb voxels (i.e., normal vector and distribution dimension), and improve the accuracy of road curb point extraction. In this study, ground points were obtained using cloth simulation filtering. First, the cuboid-based voxelization of ground points was performed. Then, taking the voxel as a unit, two geometric features, namely, the normal vector of the voxel and the linear dimension of the point distribution in the voxel, were calculated. According to these geometric features, the voxels that met the conditions were regarded as candidate road curb voxels, and the points in them as candidate road curb points. Afterward, filtering was applied using the intensity value to eliminate the bottom points of fences, street trees, and other ground objects in the candidate road curb points. Finally, noise points were eliminated according to the clustering results of the density based spatial clustering of applications with noise (DBSCAN) algorithm. In this study, point cloud data obtained by the SSW vehicle-mounted mobile mapping system and three-point cloud datasets in the IQmulus & TerraMobilita competition dataset were used to experimentally extract road curbs. Results showed that this method could effectively extract road curb points as the precision of the four groups of data results was over 90% and the quality coefficient reached over 75%.

Enhanced Schlieren System for In Situ Observation of Dynamic Light–Resin Interactions in Projection-Based Stereolithography Process

Abstract Digital maskless lithography is gaining popularity due to its unique ability to quickly fabricate high-resolution parts without the use of physical masks. By implementing controlled grayscaling and exposure control, it has the potential to replace conventional lithography altogether. However, despite the existence of a theoretical foundation for photopolymerization, observing the voxel growth process in situ is a significant challenge. This difficulty can be attributed to several factors, including the microscopic size of the parts, the low refractive index difference between cured and uncured resin, and the rapid rate of photopolymerization once it crosses a certain threshold. As such, there is a pressing need for a system that can address these issues. To tackle these challenges, the paper proposes a modified Schlieren-based observation system that utilizes confocal magnifying optics to create a virtual screen at the camera's focal plane. This system allows for the visualization of the minute changes in refractive indices made visible by the use of Schlieren optics, specifically the deflection of light by the changing density-induced refractive index gradient. The use of focusing optics provides the system with the flexibility needed to position the virtual screen and implement optical magnification. The researchers employed single-shot binary images with different pixel numbers to fabricate voxels and examine the various factors affecting voxel shape, including chemical composition and energy input. The observed results were then compared against simulations based on Beer–Lambert's law, photopolymerization curve, and Gaussian beam propagation theory. The physical experimental results validated the effectiveness of the proposed observation system. The paper also briefly discusses the application of this system in fabricating microlenses and its advantages over theoretical model-based profile predictions.

Improved two-photon polymerization through an optical fiber using coherent beam shaping

The state of the art for high resolution three-dimensional light-based additive manufacturing mainly relies on two-photon polymerization. It consists of focusing and scanning a femtosecond laser beam on a photosensitive resin to initiate crosslinking and fabricate a 3D part. This is usually achieved with a bulky objective lens which prevents printing in confined spaces. Recently, it was proposed to replace it with a multimode fiber whose core diameter is 400 μm. Thanks to wavefront shaping it was shown that the refocused beam can be digitally scanned for two-photon 3D printing. This provided a new tool to deliver light in hard-to-reach places without any motion of the fiber distal end, but it requires multiple and fine calibration measurement with a relatively complex optical setup. Here, we go a step further and demonstrate a finer digital scanning method based on a transmission matrix approach. The latter is a faster and much easier method than previously reported techniques to manipulate light at the output of the fiber. In addition, it provides smaller hatching distances (down to 200 nm) which translates into smoother 3D printed structures and it also offers the possibility of engineering the focal spot size and shape. We demonstrate the fabrication of a woodpile structure with a focused beam size of 0.8 μm along x-y and 7 μm along z. The printing time for a layer is reduced from 29 minutes to 6 minutes by engineering the voxel shape which opens up new perspectives for high resolution, fast and smooth printing in confined areas.

3D Model Classification Based on Bayesian Classifier with AdaBoost

With the exponential increasement of 3D models, 3D model classification is crucial to the effective management and retrieval of model database. Feature descriptor has important influence on 3D model classification. Voxel descriptor expresses surface and internal information of 3D model. However, it does not contain topological structure information. Shape distribution descriptor expresses geometry relationship of random points on model surface and has rotation invariance. They can all be used to classify 3D models, but accuracy is low due to insufficient description of 3D model. This paper proposes a 3D model classification algorithm that fuses voxel descriptor and shape distribution descriptor. 3D convolutional neural network (CNN) is used to extract voxel features, and 1D CNN is adopted to extract shape distribution features. AdaBoost algorithm is applied to combine several Bayesian classifiers to get a strong classifier for classifying 3D models. Experiments are conducted on ModelNet10, and results show that accuracy of the proposed method is improved.

Multi-view secondary input collaborative deep learning for lung nodule 3D segmentation

BackgroundConvolutional neural networks (CNNs) have been extensively applied to two-dimensional (2D) medical image segmentation, yielding excellent performance. However, their application to three-dimensional (3D) nodule segmentation remains a challenge.MethodsIn this study, we propose a multi-view secondary input residual (MV-SIR) convolutional neural network model for 3D lung nodule segmentation using the Lung Image Database Consortium and Image Database Resource Initiative (LIDC-IDRI) dataset of chest computed tomography (CT) images. Lung nodule cubes are prepared from the sample CT images. Further, from the axial, coronal, and sagittal perspectives, multi-view patches are generated with randomly selected voxels in the lung nodule cubes as centers. Our model consists of six submodels, which enable learning of 3D lung nodules sliced into three views of features; each submodel extracts voxel heterogeneity and shape heterogeneity features. We convert the segmentation of 3D lung nodules into voxel classification by inputting the multi-view patches into the model and determine whether the voxel points belong to the nodule. The structure of the secondary input residual submodel comprises a residual block followed by a secondary input module. We integrate the six submodels to classify whether voxel points belong to nodules, and then reconstruct the segmentation image.ResultsThe results of tests conducted using our model and comparison with other existing CNN models indicate that the MV-SIR model achieves excellent results in the 3D segmentation of pulmonary nodules, with a Dice coefficient of 0.926 and an average surface distance of 0.072.Conclusionour MV-SIR model can accurately perform 3D segmentation of lung nodules with the same segmentation accuracy as the U-net model.

Fabrication of a microfluidic device by using two-photon lithography on a positive photoresist

Organ-on-chip (OoC) devices are increasingly used for biomedical research and to speed up the process of bringing a medicinal drug from the lab to the market. The technology also addresses ethical issues linked to animal testing as it offers a reduction in animal experiments through its possibility to mimic the environment in the human body. Hence, it produces more relevant results than plain cell-culture experiments, while still sharing the possibility of high-throughput testing by simultaneously operating many devices in parallel. The main components of an OoC device are microfluidic channels and porous membranes. Current chips are often assembled from several parts. In the development phase a small change in design will cause a delay in the research because a new prototype has to be built again step-by-step. This research addresses this problem by targeting the fabrication of a microfluidic device in a single lithography and development step. A device consisting of two crossed channels at different heights separated by a membrane was chosen. The manufacturing method used was two-photon lithography (TPL) on positive photoresist AZ 4562. TPL exploits the fact that two-photon absorption non-linearly depends on the light intensity. The required intensity is only achieved in a small volume around the focal point, called the voxel. By positioning and moving this voxel inside the resist, 3D manufacturing is possible. Therefore the technique can be used for the single step fabrication of a closed channel. By vertically moving the voxel through the resist, pores can be manufactured. In literature an ellipsoidal voxel shape is assumed, leading to the assumption that the maximum voxel, and therefore pore diameter is found at the focal plane, i.e. at z = 0. However, this research shows that the voxel may also have an hourglass-shape for a high enough laser intensity. In this case the maximum voxel diameter is found at a distance z from the focal plane. For the production of pores, therefore the maximum voxel diameter instead of the diameter at the focal plane has to be taken into account. The smallest pores produced measured 250 nm - 290 nm. AZ 4562 was used as a stamp and a mold for the manufacturing of 3D topographical features on a PDMS surface. Finally the microfluidic device was successfully produced in a layer of 50 µm thick resist. The channel width was 100 µm and the height was 10 µm. The channels were separated by 10 µm in height. The sizes of the input and output holes were adapted to the diameter of the smallest available blunt needle, which is 210 µm.

Size and shape matter: The impact of voxel geometry on the identification of small nuclei.

How, and to what extent do size and shape of a voxel measured with magnetic resonance imaging (MRI) affect the ability to visualize small brain nuclei? Despite general consensus that voxel geometry affects volumetric properties of regions of interest, particularly those of small brain nuclei, no quantitative data on the influence of voxel size and shape on labeling accuracy is available. Using simulations, we investigated the selective influence of voxel geometry by reconstructing simulated ellipsoid structures with voxels varying in shape and size. For each reconstructed ellipsoid, we calculated differences in volume and similarity between the labeled volume and the predefined dimensions of the ellipsoid. Probability functions were derived from one or two individual raters and a simulated ground truth for reference. As expected, larger voxels (i.e., coarser resolution) and increasing anisotropy results in increased deviations of both volume and shape measures, which is of particular relevance for small brain structures. Our findings clearly illustrate the anatomical inaccuracies introduced by the application of large and/or anisotropic voxels. To ensure deviations occur within the acceptable range (Dice coefficient scores; DCS > 0.75, corresponding to < 57% volume deviation), the volume of isotropic voxels should not exceed 5% of the total volume of the region of interest. When high accuracy is required (DCS > 0.90, corresponding to a < 19% volume deviation), the volumes of isotropic voxels should not exceed 0.08%, of the total volume. Finally, when large anisotropic factors (>3) are used, and the ellipsoid is orthogonal to the slice axes, having its long axis in the imaging plane, the voxel volume should not exceed 0.005% of the total volume. This allows sufficient compensation of anisotropy effects, in order to reach accuracy in the acceptable range (DCS > 0.75, corresponding to >57% volume deviation).

A new method of equiangular sectorial voxelization of single-scan terrestrial laser scanning data and its applications in forest defoliation estimation

Voxelization is an efficient and frequently used data process that is applied to terrestrial laser scanning (TLS) data to facilitate data management and reduce storage size. In this study, an innovative method of equiangular sectorial voxelization is presented based on the distinctive point distribution characteristic of single-scan TLS. It has the function of containing the same number of laser beams going through each voxel, which results in metrics that can be applied to delineate forest conditions. To verify the effectiveness of the new voxelization method and to illustrate its application, 48 plots and 1098 individual trees with different degrees of defoliation were scanned using single-scan TLS. Their defoliation could be linearly regressed by using only point density metrics derived from this new shape of voxels. A 0.89 R2 value and a 12 RMSE (% of defoliation) were obtained for individual-tree-scale estimation, and a 0.83 R2 value and a 12 RMSE (% of defoliation) were obtained for plot-scale estimation. We conclude that the new voxelization method was effective, and the point density that was thus calculated was an efficient feature that revealed forest attributes.

Fiber tractography of the optic radiations: impact of diffusion model, voxel shape and orientation.

Reliable visualization of the optic radiations (OR) is of major importance in tumor surgery close to the OR to prevent permanent visual field deficits. Diffusion tensor imaging (DTI) based fiber tractography (FT) has become a standard tool to visualize major white matter tracts and to support the prevention of postoperative deficits. Nevertheless, FT of the OR is notoriously challenging due to its high neuroanatomical complexity. To improve FT of the OR we analyzed the effect of a more complex diffusion model and the effect of different voxel shapes and orientations. MRI data of 21 healthy subjects was acquired using isometric and anisometric voxel sizes and standard and adapted slice angulation. FT was performed using the DTI based approach and an orientation distribution function (ODF) based approach. Results were visually inspected, and fiber tract volumes were compared. DTI based FT led to poor results, failing to reconstruct plausible tracts at all in up to 26.11% of all cases. The ODF based approach resulted in more compound and solid tracts showing also significantly larger tract volumes. Voxel shape or orientation did not influence DTI, but ODF based FT. Isometric or anisometric voxels with standard slice orientation revealed highest tract volumes. Adapted orientation in combination with anisometric voxels led to significantly smaller tract volumes. Plausible tractography of the OR can be achieved using ODF based fiber tracking within a clinically feasible timeframe. Voxel shape and orientation seem to be of minor importance and might be kept to isometric voxel for flexible application of FT.

Voxel cores

We present a novel algorithm for computing the medial axes of 3D shapes. We make the observation that the medial axis of a voxel shape can be simply yet faithfully approximated by the interior Voronoi diagram of the boundary vertices, which we call the voxel core. We further show that voxel cores can approximate the medial axes of any smooth shape with homotopy equivalence and geometric convergence. These insights motivate an algorithm that is simple, efficient, numerically stable, and equipped with theoretical guarantees. Compared with existing voxel-based methods, our method inherits their simplicity but is more scalable and can process significantly larger inputs. Compared with sampling-based methods that offer similar theoretical guarantees, our method produces visually comparable results but more robustly captures the topology of the input shape.

Two-photon polymerization setup enables experimental mapping and correction of spherical aberrations for improved macroscopic structure fabrication.

The two-photon photopolymerization of resins by focused laser light in principle enables the fabrication of structures with details below the diffraction limit. However, the method can be highly susceptible to aberrations, which hinders the fabrication of structures that are larger than, e.g., the working distance of the microscope objective. Here, two-photon polymerization is extended to the fabrication of macroscopic structures by making use of medium numerical-aperture microscope objectives. By introducing a substrate holder movable in the axial direction it is possible to keep the focusing conditions constant and to fabricate very large structures with heights that are not limited by the working distance of the objective. Moreover, the constant focusing conditions enable us to quantify spherical aberrations by experimental mapping of the optical point-spread function, which manifests itself in the shape of singe photo-polymerized voxels. By monitoring such shapes it is possible to minimize aberrations. Effective aberration control enables us to fabricate large but detailed biomedical scaffolds with interconnected pores, e.g., in the shape of a human stirrup bone.

Abruptly autofocusing beams enable advanced multiscale photo-polymerization

Tunable-ring Airy beams are experimentally generated and employed for the fabrication of large three-dimensional structures with high resolution using multi-photon polymerization. We demonstrate that these beams can be adjusted to abruptly autofocus over an extended range of working distances while keeping their voxel shape and dimensions almost invariant. This striking property together with the real-time electronically controlled focus tuning makes these beams ideal candidates for long-range multi-photon polymerization. Moreover, the well-controlled remote localized deposition of energy can also impact many other fields of linear and nonlinear optics, like filamentation and remote high-power terahertz generation.

削り残し形状に対応した工具の干渉回避手法の開発

In this study, we propose a new automatic planning method of tool posture for continuous 5-axis machining, which can avoid collision between spindle shape and machined workpiece surface. In the proposed method, shape of the machined workpiece surface is simulated with voxel representation. Then interference among the voxel shape and took holder and shank shape are estimated by each tool moving step. Then, the interfered tool posture is modified to new tool posture without collision. In the proposed method, we developed a new efficient algorithm to set and detect new candidates of tool posture. As a result, the proposed method can realize both avoidance of rapid tool posture change and reduction of computational time to plan a new tool posture.

Laser-induced forward transfer (LIFT) of congruent voxels

Laser-induced forward transfer (LIFT) of functional materials offers unique advantages and capabilities for the rapid prototyping of electronic, optical and sensor elements. The use of LIFT for printing high viscosity metallic nano-inks and nano-pastes can be optimized for the transfer of voxels congruent with the shape of the laser pulse, forming thin film-like structures non-lithographically. These processes are capable of printing patterns with excellent lateral resolution and thickness uniformity typically found in 3-dimensional stacked assemblies, MEMS-like structures and free-standing interconnects. However, in order to achieve congruent voxel transfer with LIFT, the particle size and viscosity of the ink or paste suspensions must be adjusted to minimize variations due to wetting and drying effects. When LIFT is carried out with high-viscosity nano-suspensions, the printed voxel size and shape become controllable parameters, allowing the printing of thin-film like structures whose shape is determined by the spatial distribution of the laser pulse. The result is a new level of parallelization beyond current serial direct-write processes whereby the geometry of each printed voxel can be optimized according to the pattern design. This work shows how LIFT of congruent voxels can be applied to the fabrication of 2D and 3D microstructures by adjusting the viscosity of the nano-suspension and laser transfer parameters.

A Chemical Display: Generating Animations by Controlled Diffusion from Porous Voxels

The concept of utilizing precisely patterned and chemically loaded 3D porous containers akin to chemical voxels to enable dynamic visual patterns via spatial and temporal control of both local and global chemical release is described. Using numerical simulations and experiments, it is shown how variations in porosity, volume, shape, and relative positioning of the chemical voxels can be used to control the types of images that are formed with control in both space and time. Moving images are generated via controlled chemical release from an array of voxels to create an animation of a running man thereby illustrating proof‐of‐concept for a chemical display.

High resolution MRI imaging at 9.4 Tesla of the osteochondral unit in a translational model of articular cartilage repair.

BackgroundNon-destructive structural evaluation of the osteochondral unit is challenging. Here, the capability of high-field magnetic resonance imaging (μMRI) at 9.4 Tesla (T) was explored to examine osteochondral repair ex vivo in a preclinical large animal model. A specific aim of this study was to detect recently described alterations of the subchondral bone associated with cartilage repair.MethodsOsteochondral samples of medial femoral condyles from adult ewes containing full-thickness articular cartilage defects treated with marrow stimulation were obtained after 6 month in vivo and scanned in a 9.4 T μMRI. Ex vivo imaging of small osteochondral samples (typical volume: 1–2 cm3) at μMRI was optimised by variation of repetition time (TR), time echo (TE), flip angle (FA), spatial resolution and number of excitations (NEX) from standard MultiSliceMultiEcho (MSME) and three-dimensional (3D) spoiled GradientEcho (SGE) sequences.ResultsA 3D SGE sequence with the parameters: TR = 10 ms, TE = 3 ms, FA = 10 °, voxel size = 120 × 120 × 120 μm3 and NEX = 10 resulted in the best fitting for sample size, image quality, scanning time and artifacts. An isovolumetric voxel shape allowed for multiplanar reconstructions. Within the osteochondral unit articular cartilage, cartilaginous repair tissue and bone marrow could clearly be distinguished from the subchondral bone plate and subarticular spongiosa. Specific alterations of the osteochondral unit associated with cartilage repair such as persistent drill holes, subchondral bone cysts, sclerosis of the subchondral bone plate and of the subarticular spongiosa and intralesional osteophytes were precisely detected.ConclusionsHigh resolution, non-destructive ex vivo analysis of the entire osteochondral unit in a preclinical large animal model that is sufficient for further analyses is possible using μMRI at 9.4 T. In particular, 9.4 T is capable of accurately depicting alterations of the subchondral bone that are associated with osteochondral repair.

Comparison between voxel and consistent meso-scale models of woven composites

The applicability of voxel meshes to model the mechanical behavior of woven composites at the mesoscopic scale is studied and compared to consistent Finite Element (FE) meshes. The methods are illustrated by mechanically modeling a Representative Unit Cell (RUC) of a composite made of four layers of glass fiber plain weave fabric embedded in an epoxy matrix. Mesh convergence is studied to determine the minimum element size necessary to obtain a correct yarn volume fraction. The comparison between both methods is based on (i) homogenized macroscopic elastic properties, (ii) local stress fields, and (iii) first damage prediction. Even if a good agreement is obtained for the elastic properties, the stress concentrations due to the steplike shape of voxels induce significant differences between both methods in terms of first damage prediction.

Characterizing, measuring, and utilizing the resolution of CT imagery for improved quantification of fine-scale features

Quantitative results extracted from computed tomographic (CT) data sets should be the same across resolutions and between different instruments and laboratory groups. Despite the proliferation of scanners and data processing methods and tools, and scientific studies utilizing them, relatively little emphasis has been given to ensuring that these results are comparable or reproducible. This issue is particularly pertinent when the features being imaged and measured are of the same order size as data voxels, as is often the case with fracture apertures, pore throats, and cell walls. We have created a tool that facilitates quantification of the spatial resolution of CT data via its point-spread function (PSF), in which the user draws a traverse across a sharp interface between two materials and a Gaussian PSF is fitted to the blurring across that interface. Geometric corrections account for voxel shape and the angle of the traverse to the interface, which does not need to be orthogonal. We use the tool to investigate a series of grid phantoms scanned at varying conditions and observe how the PSF varies within and between slices. The PSF increases with increasing radial distance within slices, and can increase tangentially with increasing radial distance in CT data sets acquired with relatively few projections. The PSF between CT slices is similar to that within slices when a 2-D detector is used, but is much sharper when the data are acquired one slice at a time with a collimated linear detector array. The capability described here can be used not only to calibrate processing algorithms that use deconvolution operations, but it can also help evaluate scans on a routine basis within and between CT research groups, and with respect to the features within the imagery that are being measured.

Direct laser writing of three-dimensional network structures as templates for disordered photonic materials

In the present article we substantially expand on our recent study about the fabrication of mesoscale polymeric templates of disordered photonic network materials. We present a detailed analysis and discussion of important technical aspects related to the fabrication and characterization of these fascinating materials. Compared to our initial report we were able to reduce the typical structural length scale of the seed pattern from $a=3.3\mu m$ to $a=2\mu m$, bringing it closer to the technologically relevant fiber-optic communications wavelength range around $\lambda \sim 1.5 \mu m$. We have employed scanning electron microscopy coupled with focused ion beam cutting to look inside the bulk of the samples of different height. Moreover we demonstrate the use of laser scanning confocal microscopy to assess the real space structure of the samples fabricated by direct laser writing. We address in detail question about scalability, finite size effects and geometrical distortions. We also study the effect of the lithographic voxel shape, that is the ellipsoidal shape of the laser pen used in the fabrication process. To this end we employ detailed numerical modelling of the scattering function using a discrete dipole approximation scheme.