2D Image Sequences Research Articles

In-situ understanding of pore nucleation and growth in polyurethane foams by using real-time synchrotron X-ray tomography

The nucleation and growth of pores in PU foams have been visualized and quantified for the first time in 4D using in-situ tomographic microscopy. Acquiring sequences of 3D images in 156 ms each, we describe homogeneous and heterogeneous nucleation and subsequent bubble growth dynamics for materials with and without nucleating nanoparticles. Exploiting the time resolved tomographic sequences we obtained the bubble nucleation rate, bubble density and size evolution, based on which the diffusion coefficient of gas molecules in polymer can be calculated, evidencing the reduction of diffusivity provoked by the inclusion of nanoparticles. We found that nucleation and growth are in a large extend decoupled in time.

A Fast Fractal Based Compression for MRI Images

Magnetic resonance imaging (MRI), which assists doctors in determining clinical staging and expected surgical range, has high medical value. A large number of MRI images require a large amount of storage space and the transmission bandwidth of the PACS system in offline storage and remote diagnosis. Therefore, high-quality compression of MRI images is very research-oriented. Current compression methods for MRI images with high compression ratio cause loss of information on lesions, leading to misdiagnosis; compression methods for MRI images with low compression ratio does not achieve the desired effect. Therefore, a fast fractal-based compression algorithm for MRI images is proposed in this paper. First, three-dimensional (3D) MRI images are converted into a two-dimensional (2D) image sequence, which facilitates the image sequence based on the fractal compression method. Then, range and domain blocks are classified according to the inherent spatiotemporal similarity of 3D objects. By using self-similarity, the number of blocks in the matching pool is reduced to improve the matching speed of the proposed method. Finally, a residual compensation mechanism is introduced to achieve compression of MRI images with high decompression quality. The experimental results show that compression speed is improved by 2-3 times, and the PSNR is improved by nearly 10. It indicates the proposed algorithm is effective and solves the contradiction between high compression ratio and high quality of MRI medical images.

Hierarchical segmentation using equivalence test (HiSET): Application to DCE image sequences.

Dynamical contrast enhanced (DCE) imaging allows non invasive access to tissue micro-vascularization. It appears as a promising tool to build imaging biomarkers for diagnostic, prognosis or anti-angiogenesis treatment monitoring of cancer. However, quantitative analysis of DCE image sequences suffers from low signal to noise ratio (SNR). SNR may be improved by averaging functional information in a large region of interest when it is functionally homogeneous. We propose a novel method for automatic segmentation of DCE image sequences into functionally homogeneous regions, called DCE-HiSET. Using an observation model which depends on one parameter a and is justified a posteriori, DCE-HiSET is a hierarchical clustering algorithm. It uses the p-value of a multiple equivalence test as dissimilarity measure and consists of two steps. The first exploits the spatial neighborhood structure to reduce complexity and takes advantage of the regularity of anatomical features, while the second recovers (spatially) disconnected homogeneous structures at a larger (global) scale. Given a minimal expected homogeneity discrepancy for the multiple equivalence test, both steps stop automatically by controlling the Type I error. This provides an adaptive choice for the number of clusters. Assuming that the DCE image sequence is functionally piecewise constant with signals on each piece sufficiently separated, we prove that DCE-HiSET will retrieve the exact partition with high probability as soon as the number of images in the sequence is large enough. The minimal expected homogeneity discrepancy appears as the tuning parameter controlling the size of the segmentation. DCE-HiSET has been implemented in C++ for 2D and 3D image sequences with competitive speed.

Pincer‐type MRI morphology seen in over a third of asymptomatic healthy volunteers without femoroacetabular impingement

In daily routine, pincer femoroacetabular impingement (FAI) findings are often seen without a clinical diagnosis of pincer FAI. To assess the presence of pincer FAI MRI findings in asymptomatic volunteers with negative impingement test versus patients with clinically confirmed FAI. Case-control study. Sixty-three asymptomatic volunteers and 63 matched patients with FAI were included. A coronal T1 -weighted turbo spin-echo sequence as well as a 3D oblique transverse water-excitation true fast imaging sequence at 1.5T. The volunteers underwent standard MRI of the hip, and patients underwent MR arthrography of the symptomatic hip using the same MR sequences. Measurements of cranial acetabular version, acetabular depth, and lateral center-edge angle were performed independently by three fellowship-trained musculoskeletal radiologists. Descriptive statistics, Mann-Whitney U-test, Unpaired t-test, receiver operating characteristics (ROC), Intraclass correlation coefficient (ICC). Forty-one percent (26/63, reader 1), 32% and 37% (20 and 23/63, reader 2/3) of asymptomatic volunteers had at least one positive MR finding for pincer-FAI. Patients with pincer or mixed-type FAI had cranial retroversion of the acetabulum of -0.2° ± 7.1 (mean ± standard deviation) for reader 1 and -0.3° ± 5.5/-0.2° ± 4.8 for reader 2 / reader 3, while asymptomatic volunteers had an anteversion of 6.2° ± 6.4 (reader 1) and 3.2° ± 4.9/3.1° ± 6.5 (readers 2/3): This difference was statistically significant (P ≤ 0.002), but there was a large overlap between the groups. Acetabular depth measurements were very similar for patients with either pincer or mixed-type FAI (5.1-5.3 mm ± 3.1) and volunteers (5.2-6.1 mm ± 2.6), without a statistically significant difference (P ≤ 0.50). Lateral center-edge angle was also similar in patients with either pincer or mixed-type FAI (32.1-35.1° ± 9.1) and volunteers (30.7-33.2° ± 6.5), without a statistically significant difference (P ≤ 0.28). There is a large overlap in pincer-type MRI findings between patients with symptomatic FAI and asymptomatic volunteers. More than a third of volunteers exhibited at least one positive pincer-type MRI finding. 1 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2019;49:1296-1303.

Three‐dimensional image registration using distributed parallel computing

Three-dimensional (3D) images have become increasingly popular in practice. They are commonly used in medical imaging applications. In such applications, it is often critical to compare two 3D images, or monitor a sequence of 3D images. To make the image comparison or image monitoring valid, the related 3D images should be geometrically aligned first, which is called image registration (IR). However, IR for 3D images would take much computing time, especially when a flexible method is considered, which does not impose any parametric form on the underlying geometric transformation. Here, the authors explore a fast-computing environment for 3D IR based on the distributed parallel computing. The selected 3D IR method is based on the Taylor's expansion and 3D local kernel smoothing. It is flexible, but involves much computation. The authors demonstrate that this fast-computing environment can effectively handle the computing problem while keeping the good properties of the 3D IR method. The method discussed here is therefore useful for applications involving big data.

A hierarchical stochastic modelling approach for reconstructing lung tumour geometry from 2D CT images

ABSTRACTLung cancer is one of the deadliest cancers in both men and women. Nowadays, several methods are used to cure this cancer including surgery and radiotherapy. These methods require prior knowledge about the shape of tumours. This type of knowledge may also help physicians to determine the cancer type. In this paper we propose a novel approach for 3D reconstruction of tumour geometry from a sequence of 2D images. The proposed approach consists of two phases: tumour segmentation from computed tomography (CT) images and 3D shape reconstruction. Segmentation is conducted using snake optimisation and Gustafson–Kessel clustering. For 3D reconstruction, first, we propose a new approach to interpolate some intermediate slices between original slices. Then, the well-known marching cubes algorithm is used for surface reconstruction. Eventually, we smoothen the surface using an explicit fairing algorithm. Experiments show that our new approach can highly improve the quality and the accuracy of the reconstructed tumour shape.

A laboratory X-ray microscopy study of cracks in on-chip interconnect stacks of integrated circuits

Laboratory transmission X-ray microscopy with a spatial resolution of about 100 nm was used to image 3D interconnect structures and failures in microchips during mechanical loading, applied by a microDouble Cantilever Beam (micro-DCB) test. High-resolution 3D image sequences based on nano X-ray computed tomography (nano-XCT) are used to visualize crack opening and propagation in fully integrated multilevel on-chip interconnect structures of integrated circuits. The nondestructive investigation of sub-micron cracks during the in-situ micro-DCB test allows one to identify the weakest layers and interfaces, to image delamination along Cu/dielectric interfaces (adhesive failure) and fracture in dielectrics (cohesive failure), as well as to evaluate the robustness of Backend-of-Line stacks against process-induced thermomechanical stress.

Middle Cerebral Artery Atherosclerosis and Deep Subcortical Infarction: A 3T Magnetic Resonance Vessel Wall Imaging Study

BackgroundDeep subcortical infarction is a major subtype of stroke in middle cerebral artery (MCA) territory. This study aims to evaluate the relationship between characteristics of MCA plaque and features of deep subcortical infarction. MethodsPatients with recent acute ischemic stroke and deep subcortical infarction were prospectively enrolled. Both multicontrast brain sequences and 3D high-resolution vessel wall imaging (VWI) sequences were scanned for all patients. MCA plaque characteristics, including plaque presence, location, maximum vessel wall thickness (Max WT), signal intensity and luminal stenosis, and deep subcortical infarction features, including lowest infarct layer index (LILI), area, volume, maximum area, and infarct quantity were evaluated. Infarct feature differences were compared between MCA plaque+ group and MCA plaque− group. The correlations between MCA plaque characteristics and deep subcortical infarction features were analyzed. ResultsOf all 50 patients included in this study, 30 (60%) had MCA plaques. All deep subcortical infarction was single lesion for patients without MCA plaque. The average number of deep subcortical infarction for patients with MCA plaque was 3.10 ± 4.44. The LILI (P = .036) and infarct quantity (P = .030) showed significant differences between 2 groups. Max WT (P = .025) and stenosis degree (P = .023) were negatively correlated with LILI. Intraplaque hemorrhage was positively correlated with maximum area (P = .029) and infarct volume (P = .030). ConclusionsMCA plaque characteristics were correlated with deep subcortical infarct features. Magnetic resonance VWI may provide more information for etiological evidence of deep subcortical infarction.

Motion Management in PET/CT: Technological Solutions

Motion due to patient's breathing can introduce heavy bias in PET/CT, both in image quality and quantitation. This paper is a review of the main technical solutions available to manage movement in PET/CT studies: a) Respiratory Gated (RG), b) Motion Free (MF), c) End Expiration (EE), d) Banana Artefact Management (BAM) and e) Data Driven Gating (DDG). The most diffused solutions (RG, MF and EE) are based on LIST mode acquisition of a PET Field of View (4D FOV), centered on the anatomical region of interest; to link PET data not only to time and to spatial position but also to the corresponding breathing phase, the synchronized acquisition of the patient's breathing curve is performed by an external tracking device. Different commercial tools to track and to record patient breathing cycle are available to associate the internal organ motion with a measurable external parameter; for example these systems can measure the pressure on a chest elastic belt, the air flow trough patient nose, the breath-in and breath-out air temperature or the markers movement on the thorax/ abdominal region. Recently DDG techniques are developed to correct respiratory motion without the help of external motion tracking devices and to obtain a comparable result to that based on standard RG protocols. The final result of an RG or DDG protocol is a sequence of 3D images showing organs and lesions movement; using the other motion management options a single 3D motion-free image is obtained without motion artefacts and degradation. Compared to the previously described options the BAM solution is not a real motion management protocol but just a Banana Artefact correction technique obtained using an Attenuation Correction Map calculated merging the Whole Body Helical CT with a Cine CT on the diaphragm area. The motion management in PET/CT imaging shows benefits in terms of image quality, quantification and lesion detectability and it is useful both in diagnostic and radiotherapy planning.

Modelling farmland topography for suitable site selection of dam construction using unmanned aerial vehicle (UAV) photogrammetry

UAV photogrammetry and 3D mapping are gaining fast and wide applications around the world majorly due to the relatively low-cost advantage it offers in the acquisition of high resolution 3D topographic models when compared to LiDAR. This research seeks to demonstrate the applicability of UAV photogrammetry and Geographic Information System (GIS) in modelling the topography of a farm land located in Kwandere community, Lafia, Nasarawa state, Nigeria, for the selection of a suitable site for the construction of an earth-fill dam. The image data was acquired with the aid of a DJI Phantom 2 UAV at 120 m flying height, with a 3.61 mm camera focal length and at 65% overlap. The acquired images were processed using Agisoft Photoscan digital photogrammetric software and 20 GCPs were used for the georeferencing and check points for accuracy assessment. The Digital Elevation Model (DEM) of the area was generated from the acquired 2D image sequences using Structure from Motion (SfM) techniques, at a horizontal and vertical accuracy that falls within the desirable threshold according to NSSDA standard. Using the hydrologic tool of ArcGIS 10.3.1, the produced DEM together with the contour map of the area was finally used to determine the suitable location for the siting of dam using topographic constraints. The results obtained demonstrate the robustness of UAS products and their usefulness for planning purposes in construction and engineering applications.

Obtaining depth map from 2D non stereo images using deep neural networks

PurposeSingle image depth prediction allows to extract depth information from a usual 2D image without usage of special sensors such as laser sensors, stereo cameras, etc. The purpose of this paper is to solve the problem of obtaining depth information from 2D image by applying deep neural networks (DNNs).Design/methodology/approachSeveral experiments and topologies are presented: DNN that uses three inputs—sequence of 2D images from videostream and DNN that uses only one input. However, there is no data set, that contains videostream and corresponding depth maps for every frame. So technique of creating data sets using the Blender software is presented in this work.FindingsDespite the problem of an insufficient amount of available data sets, the problem of overfitting was encountered. Although created models work on the data sets, they are still overfitted and cannot predict correct depth map for the random images, that were included into the data sets.Originality/valueExisting techniques of depth images creation are tested, using DNN.

Determination of the Three-Dimensional Rate of Cancer Cell Rotation in an Optically-Induced Electrokinetics Chip Using an Optical Flow Algorithm.

Our group has reported that Melan-A cells and lymphocytes undergo self-rotation in a homogeneous AC electric field, and found that the rotation velocity of these cells is a key indicator to characterize their physical properties. However, the determination of the rotation properties of a cell by human eyes is both gruesome and time consuming, and not always accurate. In this paper, a method is presented to more accurately determine the 3D cell rotation velocity and axis from a 2D image sequence captured by a single camera. Using the optical flow method, we obtained the 2D motion field data from the image sequence and back-project it onto a 3D sphere model, and then the rotation axis and velocity of the cell were calculated. After testing the algorithm on animated image sequences, experiments were also performed on image sequences of real rotating cells. All of these results indicate that this method is accurate, practical, and useful. Furthermore, the method presented there can also be used to determine the 3D rotation velocity of other types of spherical objects that are commonly used in microfluidic applications, such as beads and microparticles.

High-resolution magnetic resonance vessel wall imaging in cerebrovascular diseases.

Most intracranial vascular disorders like atherosclerosis, vasculitis, and reversible cerebral vasoconstriction syndrome (RCVS) share similar lumenographic findings in traditional imaging modalities like computed tomography (CT), magnetic resonance imaging (MRI) and digital subtraction angiographic studies. Hence, there is a need for an advanced imaging modality like vessel wall imaging (VWI) to confirm the diagnosis so that appropriate clinical management could be done. Now, currently vessel wall imaging could be done in a high resolution manner with three dimensional (3D) imaging sequences. The aim of this article is to deal with the protocol as well as the current imaging applications of the high resolution VWI.

Comparison of 3D MR imaging sequences in knee articular cartilage at 1.5 T

Purpose: To compare 3D high-resolution MRI sequences in term of contrast, Signal to Noise Ratio (SNR), artefacts and cartilage thickness in knee articular cartilage at 1.5 T. Materials and Methods: Fourteen healthy volunteers and seven patients underwent articular knee cartilage exam. The 3D sequences were VOLUME Imaging with Body Enhancement (VIBE), True Fast Imaging with Steady-state Precession (TRUEFISP), Dual Echo Steady State (DESS), Multi Echo Data Image Combination (MEDIC) and Sampling Perfection with Application optimized Contrasts using different flip angle Evolutions (SPACE). For all 3D sequences, we measured contrast ratio between cartilage-bone and cartilage-meniscus, SNR and cartilage thickness. In addition, we quantified magnetic susceptibility and flow artifacts. Statistical analyses were performed using the t-test and P<0.005 was considered statistically significant. Results: Statistical results were significantly different (p>0.05). MEDIC and VIBE sequences showed the best contrast ratio between cartilage-meniscus (0.40 ± 0.14, p=0.000) and cartilage-bone (0.75 ± 0.4, p=0.022), respectively. TRUEFISP sequence showed the highest SNR (28.04 ± 14.05, p=0.000) whereas DESS sequence showed the lowest SNR (11.62 ± 3.99, p=0.000). VIBE sequence provided the highest cartilage thickness measurement compared to the remaining 3D sequences (2.0 ± 0.4 cm, IRR=75.33%, ICC=0.96). The quantification of magnetic susceptibility and flow artifacts revealed that TRUEFISP (IRR=59%, ICC=0.92) and SPACE (IRR=59%, ICC=0.96) sequences presented the highest artifacts compared to the others sequences. Conclusion: 3D high-resolution sequences provide knee articular cartilage imaging with high image quality in relatively short acquisition time. MEDIC and VIBE sequences showed the best contrast and the lowest magnetic susceptibility and flow artifacts, TRUEFISP and SPACE sequence showed the highest SNR but they are more sensitive to artifacts, DESS sequence showed medium contrast and VIBE sequence provided the highest cartilage thickness.

SPF-CellTracker: Tracking Multiple Cells with Strongly-Correlated Moves Using a Spatial Particle Filter.

Tracking many cells in time-lapse 3D image sequences is an important challenging task of bioimage informatics. Motivated by a study of brain-wide 4D imaging of neural activity in C. elegans, we present a new method of multi-cell tracking. Data types to which the method is applicable are characterized as follows: (i) cells are imaged as globular-like objects, (ii) it is difficult to distinguish cells on the basis of shape and size only, (iii) the number of imaged cells in the several-hundred range, (iv) movements of nearly-located cells are strongly correlated, and (v) cells do not divide. We developed a tracking software suite that we call SPF-CellTracker. Incorporating dependency on the cells' movements into the prediction model is the key for reducing the tracking errors: the cell switching and the coalescence of the tracked positions. We model the target cells' correlated movements as a Markov random field and we also derive a fast computation algorithm, which we call spatial particle filter. With the live-imaging data of the nuclei of C. elegans neurons in which approximately 120 nuclei of neurons were imaged, the proposed method demonstrated improved accuracy compared to the standard particle filter and the method developed by Tokunaga etal. (2014).

Highly Articulated Kinematic Structure Estimation Combining Motion and Skeleton Information.

In this paper, we present a novel framework for unsupervised kinematic structure learning of complex articulated objects from a single-view 2D image sequence. In contrast to prior motion-based methods, which estimate relatively simple articulations, our method can generate arbitrarily complex kinematic structures with skeletal topology via a successive iterative merging strategy. The iterative merge process is guided by a density weighted skeleton map which is generated from a novel object boundary generation method from sparse 2D feature points. Our main contributions can be summarised as follows: (i) An unsupervised complex articulated kinematic structure estimation method that combines motion segments with skeleton information. (ii) An iterative fine-to-coarse merging strategy for adaptive motion segmentation and structural topology embedding. (iii) A skeleton estimation method based on a novel silhouette boundary generation from sparse feature points using an adaptive model selection method. (iv) A new highly articulated object dataset with ground truth annotation. We have verified the effectiveness of our proposed method in terms of computational time and estimation accuracy through rigorous experiments with multiple datasets. Our experiments show that the proposed method outperforms state-of-the-art methods both quantitatively and qualitatively.

Three-dimensional surface microtopography recovery from a multifocus image sequence using an omnidirectional modified Laplacian operator with adaptive window size.

Optical microscopy enables the observation of highly magnified objects and material structures on microsurfaces, but it can only acquire 2D images. In order to observe areal features more accurately and intuitively, 3D surface microtopography recovery has been applied to form a 3D surface model of an object from its 2D image sequence. In the 3D reconstruction of the focus evaluation operator, we have the gray variance operator, the gray-scale difference absolute sum operator, the Roberts gradient operator, the Tenengrad gradient operator, the improved Laplace operator, etc. There are two problems with these operators: one is that there is no difference between (x,y) and the gray scale of the pixel in the diagonal direction in the field and the other is that the window size of the focus evaluation operator is fixed, e.g., 3×3, 5×5, etc. Thus, the size of the window for each pixel in the image is the same, and the small window may not cover enough field information while being vulnerable to noise. Large windows can cover more information, but they may result in a smoothing phenomenon, which affects the accuracy of the model. Different pixels around the field have different pixel colors when the size of the window is not the same. Therefore, this paper proposes a modified omnidirectional Laplacian operator with an adaptive window to automatically adjust the size of the window according to the color difference within the window. This also takes into consideration the pixels in the diagonal direction. In addition, very comprehensive verification experiments proved the conclusions.

MRI of the Hip: What the Surgeon Wants to Know

The purpose of this review is to explore novel MR imaging techniques focused on the hip as well as specific MR imaging findings of interest to the referring orthopedic surgeon. MR arthrography demonstrates high accuracy for the detection of chondrolabral lesions, which portend less favorable prognosis and may change surgical approach. The addition of leg traction has been shown to increase the accuracy of MR arthrography in detecting intra-articular pathology. 3D models of the hip generated from 3D imaging sequences can also be used for pre-surgical planning, including simulation of preoperative range of motion. Extra-articular pathology, including hamstring, gluteal, and iliopsoas tendon pathology can necessitate surgical intervention depending on the degree of injury. MR arthrography remains the imaging modality of choice in assessment of the hip when intra-articular pathology is clinically suspected, though conventional MR imaging at 3.0 T also demonstrates strong diagnostic performance. New techniques including direct traction and 3D imaging can provide valuable diagnostic and pre-surgical information in the assessment of the athlete’s hip.

Robust motion tracking in liver from 2D ultrasound images using supporters.

Effectiveness of image-guided radiation therapy with precise dose delivery depends highly on accurate target localization, which may involve motion during treatment due to, e.g., breathing and drift. Therefore, it is important to track the motion and adjust the radiation delivery accordingly. Tracking generally requires reliable target appearance and image features, whereas in ultrasound imaging acoustic shadowing and other artifacts may degrade the visibility of a target, leading to substantial tracking errors. To minimize such errors, we propose a method based on so-called supporters, a computer vision tracking technique. This allows us to leverage information from surrounding motion for improving robustness of motion tracking on 2D ultrasound image sequences of the liver. Image features, potentially useful for predicting the target positions, are individually tracked, and a supporter model capturing the coupling of motion between these features and the target is learned on-line. This model is then applied to predict the target position, when the target cannot be otherwise tracked reliably. The proposed method was evaluated using the Challenge on Liver Ultrasound Tracking (CLUST)-2015 dataset. Leave-one-out cross-validation was performed on the training set of 24 2D image sequences of each 1-5min. The method was then applied on the test set (24 2D sequences), where the results were evaluated by the challenge organizers, yielding 1.04mm mean and 2.26mm 95%ile tracking error for all targets. We also devised a simulation framework to emulate acoustic shadowing artifacts from the ribs, which showed effective tracking despite the shadows. Results support the feasibility and demonstrate the advantages of using supporters. The proposed method improves its baseline tracker, which uses optic flow and elliptic vessel models, and yields the state-of-the-art real-time tracking solution for the CLUST challenge.

Linear relationship found by magnetic resonance imaging between cerebrospinal fluid volume and body weight in dogs.

Despite numerous studies on cerebrospinal fluid (CSF) and its importance during hydrocephalus or myelography, no reliable values exist about its overall volume in dogs. In this study, our aim was to measure the intracranial (IC) volume of CSF in dogs and assess its possible relationship with body size and the symmetry of the lateral ventricles. We ran a 3D magnetic resonance imaging (MRI) sequence on the central nervous system of 12 healthy, male mongrel dogs between 3-5 years of age and 7.5-35.0 kg body weight. A validated semiautomatic segmentation protocol was implemented to segment the CSF and measure its volume. Values for the volume of the ventricular compartment were between 0.97 and 2.94 ml, with 62.1 ± 11.7% in the lateral ventricles, 17.6 ± 4.9% in the third ventricle, 4.9 ± 1.6% in the aqueductus mesencephali and 15.5 ± 6.6% in the fourth ventricle. In 11 cases a significant asymmetry was found between the lateral ventricles. The results suggest that it may be normal for a dog to have one of the lateral ventricles 1.5 times larger than the other. The correlation between body weight and CSF volume was linear, indicating that the current dosage protocols for myelography, based on a hypothetical proportional relationship with body weight, may have to be revised.