3D investigation of a PS / ABS polymer using different microscopy techniques

To examine whether the third-trimester fetus can be assessed sonographically using 3-dimensional (3D) volume data sets. Twenty-seven consecutive third-trimester fetuses were evaluated. Fetuses were scanned using 2-dimensional (2D) imaging followed by 5 3D volume acquisitions. The initial scan was interpreted and reported based on the 2D images. The 3D volume data sets were independently reviewed offline several weeks later by 2 sonologists. Parameters evaluated included fetal presentation, placental location, amniotic fluid volume, fetal biometry including a calculation of estimated fetal weight, and major fetal anatomic structures. The result of the interpretation via 3D reconstruction of the volume from each of the 2 sonologists was compared with the original 2D sonography report. Fetal presentation, amniotic fluid volume, and placental location with respect to the cervix were correctly identified 100% of the time by each sonologist. The estimated fetal weight was within 10% of the 2D estimate 89% (95% CI, 0.71-0.98) of the time for sonologist A and 96% (95% CI, 0.81-0.99) of the time for sonologist B. The majority of major anatomic landmarks were adequately seen by both sonologists. Offline review of 3D volume data sets is a reliable method for determining fetal presentation, amniotic fluid volume, placental location, and estimating fetal weight in the third trimester.

ABSTRACTThe aim of this paper is to image particularly buried archaeological remains with a new technique for 3D display in the Temple of Augustus (Monumentum Ancyranum) relating to the Roman time, around and in Agora in the Ulus district of Ankara, Turkey. 2D ground‐penetrating radar (GPR) data were acquired on constant spaced parallel profiles in the study area. After data processing, a 3D data volume was built using a parallel 2D data set. The full data block was divided into sub‐data blocks in time. A weak amplitude range was made invisible with zero opaque by applying zero opacity to these values in the visualization. The arranged visible maximum amplitude range was enlarged or the amplitude scale was weighted with a constant coefficient, which was greater than one and smaller than two and had a decimal number, according to the time range. In this way, a transparent 3D image was obtained for determining buried remains according to the depth range. Interactive visualization was carried out by constructing sub‐blocks of the transparent 3D volume.The interactive transparent 3D visualization was provided to identify the archaeological remains on native locations with depth in a 3D volume. Very complex and deep wall structures were visualized with any depth range inside (cella) of the Temple of Augustus and a few very narrow cubic anomalies exceeding 4 m deep were determined at the East side of the temple. An excavation could not be carried out in the cella because of the sensitivity problem of the cella walls. However, excavations in front of the East wall of the temple encouraged the new 3D image results. In addition, a lot of iron clamps connecting wall marble stones were determined on a profile gathered on the East wall.

In this study, four-dimensional (3D+t) ultrasound imaging techniques were used for the development and in vivo verification of 3D strain imaging. Two different iterative coarse- to-fine 3D strain estimation methods were developed. One method was based on measuring displacements using 3D kernels and a 3D cross-correlation function. The second method used 2D kernels and cross-correlation, and estimated 3D displacements in an iterative process. A 3D or 2D parabolic interpolation was used for sub-sample displacement estimates. The strain estimation methods were experimentally validated using a gelatin phantom with a hard cylindrical inclusion (four times stiffer). The phantom was compressed with a plate in steps of 0.5 mm up to 3.0 mm (3% strain). Rf-data were acquired with a 3D matrix array transducer (X4, Philips Sonos 7500) in ECG-triggered 3D full volume mode. Preliminary in vivo validation was performed by acquiring 3D full volume data (frame rate = 19 Hz) of the left ventricle of a trained athlete. Both methods were able to produce high quality elastograms of the inclusion model up to an applied compression of 3% strain (resulting in 0.5% - 5% axial strain in the phantom). No significant difference in elastographic signal-to- noise ratio (SNRe) was found between the two methods. The iterative 2D algorithm is favored for the shorter computation time. The signal- and contrast-to-noise ratios (SNRe, CNRe) of the axial elastograms increased to 28 and 53 dB, respectively (compared to previously described BiPlane axial strain images). Lateral and elevational elastograms were also in accordance with finite element solutions of the phantom model. However, the SNRe and CNRe were considerably lower (16 and 33 dB), which is presumably caused by the lower in-plane spatial resolution of the 3D full volume data. Initial in vivo results revealed mean strain profiles in three orthogonal directions comparable with our previous studies, although, the maximum radial strain was lower than expected (20%). Hence, 3D cardiac strain imaging is feasible even at a relatively low frame rate.

A new methodology for frequency bandwidth enhancement using 3D layered reflectivity inversion is demonstrated in this paper. The inversion is used to obtain layered reflectivity and then band-limited to the outer bounds of the input seismic. The 3D reflectivity inversion starts with a thin layer model based on the geological picks following the structural and stratigraphic trends. The initial thickness of the thin layers is determined by the frequency bandwidth of the seismic data, and the inversion process orients them conformably to stratigraphic layers in the subsurface. The thickness and impedance of each layer (and hence interface reflection coefficient) is perturbed and the corresponding modelled synthetic is compared to the seismic data in a volumetric fashion to minimise the misfit (Walia, 2001). This contributes to the removal of tuning effects and provides a reflectivity sequence with a much higher frequency bandwidth than the seismic data. Subsequently, the reflection series is band-limited to obtain higher resolution seismic data. The limits of bandwidth enhancement is selected by correlating the enhanced seismic to well logs and VSPs. The comparison of seismic profiles attribute maps and correlation with well data presented in this paper confirm that this process has resolved real seismic events. This resulted in confident picking of top and base of individual sand units, consequently providing better maps of the reservoir sands. Application to Seismic Data Rigorous study of this technique on several 3D data volumes from the North Sea and Ecuador’s Oriente Basin have produced very encouraging results in terms of resolving Top/Base of producing thin sands, improving structural features and complex-fault definition. Fig. 1 shows a simple workflow of the methodology. An initial model, based on the main geological horizons and log data, is input to the inversion process. The inversion process describes the existing zero-phase 3D volume by a discrete number of thin (half of time-thickness of the input wavelet) layers (Ricker, 1953). Each layer consists of two parameters, the time thickness and initial impedance and hence reflection coefficient value at the interface of every bin location. During the inversion process both the parameters are perturbed within pre-defined constraints based on well log data. Convergence to the final solution is achieved using a full 3D algorithm based on simulated annealing. The resulting layered structure provides an effective resolution of typically an eighth of the wavelength (Gluck et al, 1997; Widess, 1973). The data reduction implicit in the layer definition makes this algorithm extremely robust to noise. This inverted reflectivity is then band-limited such that the enhanced seismic has a broader frequency bandwidth than the input seismic. Fig. 2 shows a comparison of the seismic data from the Oriente Basin before Hz dB

Manual landmark positioning in volumetric image data is a complex task and often results in erroneous landmark positions. The landmark positioning tool presented uses image curvature features to precompute suitable candidates for landmark positions on surface data of anatomical structures. A force-feedback I/O device is then used to haptically guide the user during the definition of the correct landmarks in the 3D data volume. Furthermore, existing landmarks in a time-point of a sequence of 3D volumes (4D data set) can iteratively be transferred to other time-points using a surface based registration technique. First results show significant time savings and small interobserver variability (IROV) compared to the IROV of manually defined landmark positions using orthogonal slices of the image data.

Exploded views and cutaway views have been demonstrated to be powerful visualization techniques to explore the complex internal structures in 3D volume data. Current approaches for creating them are mainly based on algorithms with very limited user controls. This paper proposes a novel set of whole-hand multitouch gestures for interactive creation and manipulation of these views on 3D volume data. Moreover, users can also interactively perform bimanual ‘V’-shape cut and volume exploding, remove unwanted volume partitions to effectively create cutaway views, and create threshold-based volume segmentation and exploded views. Our proposed interaction design was also tried out by eight participants in a preliminary pilot study.

"Patient specific dosimetry established during the last decade in modern radio-therapy. Usually, tracer kinetics in main compartments of observed metabolism is assessed from anterior and posterior whole body scans. The effective doses for each organ, derived by the MIRD scheme, provide evidence for following radiotherapeutic treatment and helps to meet vital dose limits for critical organs, e.g. kidneys. The calculation of individual dose in a three-dimensional context leads to more accurate dose estimates, as was proven by intensive research, but is still on the cusp to clinical application. In this work, a statistical approach, based on multi-modal image and feature data, is presented, to overcome manual segmentation, the most time consuming step, in 3D based dose calculation. 3D data volumes from a hybrid SPECT study, comprising SPECT and CT data, covering main compartments of metabolism, build the image features of a Gaussian classifier. From prior segmentations organspecific membership maps are derived, and substituted as additional feature into the segmentation procedure. Centroids, eccentricity and principal axes of organ models are registered to a rough thresholded image of the SPECT component, and define membership coefficients of the voxels. The new approach yields accurate results, even with real patient data. The new method needs minimal user interaction during selection of some sample regions, thus showing high potential for implementation in a clinical workflow."

High-energy diffraction microscopy (HEDM) in-situ mechanical testing experiments offer unique insight into the evolving deformation state within polycrystalline materials. These experiments rely on a sophisticated analysis of the diffraction data to instantiate a 3D reconstruction of grains and other microstructural features associated with the test volume. For microstructures of engineering alloys that are highly twinned and contain numerous features around the estimated spatial resolution of HEDM reconstructions, the accuracy of the reconstructed microstructure is not known. In this study, we address this uncertainty by characterizing the same HEDM sample volume using destructive serial sectioning (SS) that has higher spatial resolution. The SS experiment was performed on an Inconel 625 alloy sample that had undergone HEDM in-situ mechanical testing to a small amount of plastic strain (~ 0.7%), which was part of the Air Force Research Laboratory Additive Manufacturing (AM) Modeling Series. A custom-built automated multi-modal SS system was used to characterize the entire test volume, with a spatial resolution of approximately 1 µm. Epi-illumination optical microscopy images, backscattered electron images, and electron backscattered diffraction maps were collected on every section. All three data modes were utilized and custom data fusion protocols were developed for 3D reconstruction of the test volume. The grain data were homogenized and downsampled to 2 µm as input for Challenge 4 of the AM Modeling Series, which is available at the Materials Data Facility repository.

Purpose/Aims of the study: Bone’s hierarchical structure can be visualized using a variety of methods. Many techniques, such as light and electron microscopy generate two-dimensional (2D) images, while micro-computed tomography (µCT) allows a direct representation of the three-dimensional (3D) structure. In addition, different methods provide complementary structural information, such as the arrangement of organic or inorganic compounds. The overall aim of the present study is to answer bone research questions by linking information of different 2D and 3D imaging techniques. A great challenge in combining different methods arises from the fact that they usually reflect different characteristics of the real structure.Materials and methods: We investigated bone during healing by means of µCT and a couple of 2D methods. Backscattered electron images were used to qualitatively evaluate the tissue’s calcium content and served as a position map for other experimental data. Nanoindentation and X-ray scattering experiments were performed to visualize mechanical and structural properties.Results: We present an approach for the registration of 2D data in a 3D µCT reference frame, where scanning electron microscopies serve as a methodic link. Backscattered electron images are perfectly suited for registration into µCT reference frames, since both show structures based on the same physical principles. We introduce specific registration tools that have been developed to perform the registration process in a semi-automatic way.Conclusions: By applying this routine, we were able to exactly locate structural information (e.g. mineral particle properties) in the 3D bone volume. In bone healing studies this will help to better understand basic formation, remodeling and mineralization processes.

Selective laser melting (SLM) has been shown to be an attractive manufacturing route for the production of α/β titanium alloys. The relationship between the SLM process parameters and the microstructure of titanium alloys has been the object of several works, but the texture formation during the SLM process has yet to be understood. In the present study, the texture formation of Ti-6Al-4V components was investigated in order to clarify which microstructural features can be tailored during the SLM process. The microstructural characterization of the as-built components was carried out using various microscopy techniques. Phase and texture analysis were carried out using backscattered electron imaging and diffraction. It was found that as-built components consist exclusively of α′ martensitic phase precipitated from prior β columnar grains. The texture of the prior β phase was reconstructed and discussed in relation to the used SLM process parameters. It was found that the β grain solidification is influenced by the laser scan strategy and that the β phase has a strong 〈100〉 texture along its grain growth direction. The α′ martensitic laths that originate from the parent β grains precipitate according to the Burgers orientation relationship. It was observed that α′ laths clusters from the same β grain have a specific misorientation that minimizes the local shape strain. Texture inheritance across successive deposited layers was also observed and discussed in relation to various variant selection mechanisms.

Deep learning has huge potential for accurate disease prediction with neuroimaging data, but the prediction performance is often limited by training-dataset size and computing memory requirements. To address this, we propose a deep convolutional neural network model, Simple Fully Convolutional Network (SFCN), for accurate prediction of brain age using T1-weighted structural MRI data. Compared with other popular deep network architectures, SFCN has fewer parameters, so is more compatible with small dataset size and 3D volume data. The network architecture was combined with several techniques for boosting performance, including data augmentation, pre-training, model regularization, model ensemble and prediction bias correction. We compared our overall SFCN approach with several widely-used machine learning models. It achieved state-of-the-art performance in UK Biobank data (N = 14,503), with mean absolute error (MAE) = 2.14y in brain age prediction and 99.5% in sex classification. SFCN also won (both parts of) the 2019 Predictive Analysis Challenge for brain age prediction, involving 79 competing teams (N = 2,638, MAE = 2.90y). We describe here the details of our approach, and its optimisation and validation. Our approach can easily be generalised to other tasks using different image modalities, and is released on GitHub.

Generative adversarial networks(GANs) have been successfully applied for generating high quality natural images and have been extended to the generation of RGB videos and 3D volume data. In this paper we consider the task of generating RGB-D videos, which is less extensively studied and still challenging. We explore deep GAN architectures suitable for the task, and develop 4 GAN architectures based on existing video-based GANs. With a facial expression database, we experimentally find that an extended version of the motion and content decomposed GANs, known as MoCoGAN, provides the highest quality RGB-D videos. We discuss several applications of our GAN to content creation and data augmentation, and also discuss its potential applications in behavioral experiments.

Volume rendering is a powerful tool for visualizing sampled scalar values from 3D data without fitting geometric primitives to the data. However, the size of volume data is usually too big to handle in real time. Recently, various volume rendering algorithms have been proposed in order to reduce the rendering time. However, most of the proposed algorithms are not proper for large volume data, which is required in some applications such as the spine biopsy simulator. In this paper, we propose a block-based fast volume rendering algorithm using shear-warp factorization. The algorithm performs volume rendering by using information of organ segmentation as well as 3D volume data, which is a common environment in the spine biopsy simulator. The proposed block-based algorithm can reduce storage requirement and increase the rendering speed by treating the 3D data on a block-by-block basis. The proposed algorithm is evaluated by rendering 3D body images obtained from X-ray CT (computed tomograph) system.

3D Active Net, which is a 3D extension of Snakes, is an energy-minimizing surface model which can extract a volume of interest from 3D volume data. It is deformable and evolves in 3D space to be attracted to salient features, according to its internal and image energy. The net can be fitted to the contour of a target object by defining the image energy suitable for the contour property. We present testing results of the extraction of a muscle from the Visible Human Data by two methods: manual segmentation and the application of 3D Active Net. We apply principal component analysis, which utilizes the color information of the 3D volume data to emphasize an ill-defined contour of the muscle, and then apply 3D Active Net. We recognize that the extracted object has a smooth and natural contour in contrast with a comparable manual segmentation, proving an advantage of our approach.

Volume rendering is a technique that visualize 2D image of object from 3D volume data on image screen. Ray casting algorithm, one of popular volume rendering techniques, generate image with detail and high quality compared with other volume rendering algorithms but since this is a highly time consuming process given large number of voxels, many acceleration techniques have been developed. Here we introduce new acceleration technique, efficient space leaping method. Our new space leaping method traverse volume data and projects 3D location of voxel onto image screen to find pixels that have non-zero value in final volume image and locations of non-empty voxels that are closest to ray. During this process, adaptive run-length encoding and line drawing algorithm are used to traverse volume data and find pixels with non-zero value efficiently. Then we cast rays not through entire screen pixel but only through projected screen pixels and start rendering process from non-empty voxel location directly. This new method shows significant time savings applied to surface extraction without loss of image quality.