Shape Signature Research Articles

Origami: Single-cell 3D shape dynamics oriented along the apico-basal axis of folding epithelia from fluorescence microscopy data.

A common feature of morphogenesis is the formation of three-dimensional structures from the folding of two-dimensional epithelial sheets, aided by cell shape changes at the cellular-level. Changes in cell shape must be studied in the context of cell-polarised biomechanical processes within the epithelial sheet. In epithelia with highly curved surfaces, finding single-cell alignment along a biological axis can be difficult to automate in silico. We present 'Origami', a MATLAB-based image analysis pipeline to compute direction-variant cell shape features along the epithelial apico-basal axis. Our automated method accurately computed direction vectors denoting the apico-basal axis in regions with opposing curvature in synthetic epithelia and fluorescence images of zebrafish embryos. As proof of concept, we identified different cell shape signatures in the developing zebrafish inner ear, where the epithelium deforms in opposite orientations to form different structures. Origami is designed to be user-friendly and is generally applicable to fluorescence images of curved epithelia.

Multi-Modality Learning for Non-Rigid 3D Shape Retrieval via Structured Sparsity Regularizations

Big challenges are usually occurring in non-rigid 3D shape retrieval, for the shapes undergoing arbitrarily non-affine transformations. In this work, a novel design of feature learning approach is proposed for non-rigid 3D shape retrieval, dubbed Structured Sparsity Regularized Multi-Modality Method (SSR-MM). The shape signatures which capture the deformation-invariant characteristics are averaged and stacked for a multi-modality machine learning approach, and a transform matrix based on the structure sparsity regularization is utilized to map those signatures obtaining the discriminative features for retrieval. The proposed framework is evaluated on the publicly available non-rigid 3D human benchmarks, and the experimental results show the efficacy of our contributions and the advantages of our method over existing ones.

Cortical Surface Shape Analysis Based on Alexandrov Polyhedra.

Shape analysis has been playing an important role in early diagnosis and prognosis of neurodegenerative diseases such as Alzheimer's diseases (AD). However, obtaining effective shape representations remains challenging. This paper proposes to use the Alexandrov polyhedra as surface-based shape signatures for cortical morphometry analysis. Given a closed genus-0 surface, its Alexandrov polyhedron is a convex representation that encodes its intrinsic geometry information. We propose to compute the polyhedra via a novel spherical optimal transport (OT) computation. In our experiments, we observe that the Alexandrov polyhedra of cortical surfaces between pathology-confirmed AD and cognitively unimpaired individuals are significantly different. Moreover, we propose a visualization method by comparing local geometry differences across cortical surfaces. We show that the proposed method is effective in pinpointing regional cortical structural changes impacted by AD.

Revealing the formation histories of the first stars with the cosmic near-infrared background

ABSTRACT The cosmic near-infrared background (NIRB) offers a powerful integral probe of radiative processes at different cosmic epochs, including the pre-reionization era when metal-free, Population III (Pop III) stars first formed. While the radiation from metal-enriched, Population II (Pop II) stars likely dominates the contribution to the observed NIRB from the reionization era, Pop III stars – if formed efficiently – might leave characteristic imprints on the NIRB, thanks to their strong Lyα emission. Using a physically motivated model of first star formation, we provide an analysis of the NIRB mean spectrum and anisotropy contributed by stellar populations at z > 5. We find that in circumstances where massive Pop III stars persistently form in molecular cooling haloes at a rate of a few times $10^{-3}\, \mathrm{ M}_\odot \ \mathrm{yr}^{-1}$, before being suppressed towards the epoch of reionization (EoR) by the accumulated Lyman–Werner background, a unique spectral signature shows up redward of $1\, \mu$m in the observed NIRB spectrum sourced by galaxies at z > 5. While the detailed shape and amplitude of the spectral signature depend on various factors including the star formation histories, initial mass function, LyC escape fraction and so forth, the most interesting scenarios with efficient Pop III star formation are within the reach of forthcoming facilities, such as the Spectro-Photometer for the History of the Universe, Epoch of Reionization, and Ices Explorer. As a result, new constraints on the abundance and formation history of Pop III stars at high redshifts will be available through precise measurements of the NIRB in the next few years.

Coarse registration of point clouds with low overlap rate on feature regions

In order to solve the problems of long time consuming and easy failure of the existing coarse registration algorithms based on global registration on two 3D point clouds with low overlap rate, we proposed a coarse registration algorithm based on feature regions and the Super 4-Points Congruent Sets (SUPER4PCS) algorithm. Firstly, intrinsic shape signatures (ISS) algorithm was used to extract and describe the features of the down-sampled point clouds. Secondly, the feature point clouds were divided into regions and the initial overlapped sub-regions were extracted. Thirdly, the complete overlapping regions were grown from the overlapped sub-regions and gradually recovered. Finally, the SUPER4PCS was used for registration on the complete overlapping regions. The experimental results showed that the geometric accuracy, registration success rate, and robustness of the proposed algorithm were better than that of SUPER4PCS and its improved algorithms, and the time consumption was one order of magnitude lower than that of SUPER4PCS on point clouds with low overlap rate.

Statistical morphological analysis reveals characteristic paraspinal muscle asymmetry in unilateral lumbar disc herniation

Growing evidence suggests an association of lumbar paraspinal muscle morphology with low back pain (LBP) and lumbar pathologies. Unilateral spinal disorders provide unique models to study this association, with implications for diagnosis, prognosis, and management. Statistical shape analysis is a technique that can identify signature shape variations related to phenotypes but has never been employed in studying paraspinal muscle morphology. We present the first investigation using this technique to reveal disease-related paraspinal muscle asymmetry, using MRIs of patients with a single posterolateral disc herniation at the L5-S1 spinal level and unilateral leg pain. Statistical shape analysis was conducted to reveal disease- and phenotype-related morphological variations in the multifidus and erector spinae muscles at the level of herniation and the one below. With the analysis, shape variations associated with disc herniation were identified in the multifidus on the painful side at the level below the pathology while no pathology-related asymmetry in cross-sectional area (CSA) and fatty infiltration was found in either muscle. The results demonstrate higher sensitivity and spatial specificity for the technique than typical CSA and fatty infiltration measures. Statistical shape analysis holds promise in studying paraspinal muscle morphology to improve our understanding of LBP and various lumbar pathologies.

MFCIS: an automatic leaf-based identification pipeline for plant cultivars using deep learning and persistent homology

Recognizing plant cultivars reliably and efficiently can benefit plant breeders in terms of property rights protection and innovation of germplasm resources. Although leaf image-based methods have been widely adopted in plant species identification, they seldom have been applied in cultivar identification due to the high similarity of leaves among cultivars. Here, we propose an automatic leaf image-based cultivar identification pipeline called MFCIS (Multi-feature Combined Cultivar Identification System), which combines multiple leaf morphological features collected by persistent homology and a convolutional neural network (CNN). Persistent homology, a multiscale and robust method, was employed to extract the topological signatures of leaf shape, texture, and venation details. A CNN-based algorithm, the Xception network, was fine-tuned for extracting high-level leaf image features. For fruit species, we benchmarked the MFCIS pipeline on a sweet cherry (Prunus avium L.) leaf dataset with >5000 leaf images from 88 varieties or unreleased selections and achieved a mean accuracy of 83.52%. For annual crop species, we applied the MFCIS pipeline to a soybean (Glycine max L. Merr.) leaf dataset with 5000 leaf images of 100 cultivars or elite breeding lines collected at five growth periods. The identification models for each growth period were trained independently, and their results were combined using a score-level fusion strategy. The classification accuracy after score-level fusion was 91.4%, which is much higher than the accuracy when utilizing each growth period independently or mixing all growth periods. To facilitate the adoption of the proposed pipelines, we constructed a user-friendly web service, which is freely available at http://www.mfcis.online.

B-PO02-044 ELECTROGRAM FINGERPRINTS OF ATRIAL FIBRILLATION

It is unclear whether disorganized waves in atrial fibrillation (AF) produce distinct ‘signature shapes’ on the electrogram (EGM), reflecting specific patterns. We tested the hypothesis that deep learning (DL) could be used to identify specific EGM shapes pathognomic for AF, that could form the basis for future approaches to detect subtype and treat patients. We developed custom-built convolutional (CNN) and recurrent (RNN) neural networks using N=29,340 EGM segments (4 sec), from 86 patients with AF or AFL at ablation (25 female, 65±11 years), and using 10-fold cross-validation (80% patients for training, 20% for independent testing). We probed CNN by generating 29,880 EGM sequences from patient data, in which we successively controlled variables including cycle length (CL) and EGM shape. Trained CNN and RNN identified AF from other atrial arrhythmias with an AUC of 0.95 ± 0.05 and 0.97 ± 0.04 in the testing sets respectively (Fig. A). We identified EGM shapes that were classified as AF even if they had regular rhythm and slow rate (long CL). A systematic analysis of 101 such AF signature EGMs were classified as AF >80% by CNN regardless of rate and irregularity (fig B, orange), and also by the RNN on 30 of such EGM signatures. Signature AF electrograms were identical by both RNN and CNN classification for 15 shapes (fig C, red shapes). Deep learning identified signature AF electrograms. EGM signature shapes were more important to AF L classification than high rate or irregular timing. Extraction of complex electrophysiological information hidden to the expert eye may improve diagnosis and shed insights into mechanisms of EGM generation in AF.

Crystallinity characterization of white matter in the human brain

Human brain tissue is a heterogeneous material, consisting of soft outer grey matter tethered internally by stiffer cords of white matter. These white matter tracts conduct electrical impulses between grey matter regions, thereby underpinning neuronal communication. Understanding the material properties of white matter is thus crucial for understanding brain function generally. Efforts to assess white matter microstructure are currently hampered by the inherent limitations of reconstruction by diffusion imaging. Techniques typically represent white matter structures with single scalars that are often difficult to interpret. Here, we address these issues by introducing tools from materials physics for the characterization of white matter microstructure. We investigate structure on a mesoscopic scale by analyzing its homogeneity and determining which regions of the brain are structurally homogeneous, or ‘crystalline’ in the context of materials physics. We find that crystallinity provides novel information and varies across the brain along interpretable lines of anatomical difference, with highest homogeneity in regions adjacent to the corpus callosum, a large interhemispheric tract. Furthermore, crystallinity is markedly reliable across iterative measurement, yet also varies between individual human volunteers, making it potentially useful for examining individual differences in white matter along several dimensions including sex and age. We also parcellate white matter into ‘crystal grains’, or contiguous sets of voxels of high structural similarity, and find overlap with a common atlas of distinct white matter areas. Finally, we characterize the shapes of individual diffusion signatures through another tool from materials physics—bond-orientational order parameters—to locate fiber crossings and fascicles. Our results provide new means of assessing white matter microstructure on multiple length scales, and open multiple avenues of future inquiry involving soft matter physics and neuroscience.

Cell-free genomics reveal intrinsic, cooperative and competitive determinants of chromatin interactions.

Metazoan transcription factors distinguish their response elements from a large excess of similar sequences. We explored underlying principles of DNA shape read-out and factor cooperativity in chromatin using a unique experimental system. We reconstituted chromatin on Drosophila genomes in extracts of preblastoderm embryos, mimicking the naïve state of the zygotic genome prior to developmental transcription activation. We then compared the intrinsic binding specificities of three recombinant transcription factors, alone and in combination, with GA-rich recognition sequences genome-wide. For MSL2, all functional elements reside on the X chromosome, allowing to distinguish physiological elements from non-functional ‘decoy’ sites. The physiological binding profile of MSL2 is approximated through interaction with other factors: cooperativity with CLAMP and competition with GAF, which sculpts the profile by occluding non-functional sites. An extended DNA shape signature is differentially read out in chromatin. Our results reveal novel aspects of target selection in a complex chromatin environment.

Ventricle shape analysis using modified WKS for atrophy detection.

Brain ventricle is one of the biomarkers for detecting neurological disorders. Studying the shape of the ventricles will aid in the diagnosis process of atrophy and other CSF-related neurological disorders, as ventricles are filled with CSF. This paper introduces a spectral analysis algorithm based on wave kernel signature. This shape signature was used for studying the shape of segmented ventricles from the brain images. Based on the shape signature, the study groups were classified as normal subjects and atrophy subjects. The proposed algorithm is simple, effective, automated, and less time consuming. The proposed method performed better than the other methods heat kernel signature, scale invariant heat kernel signature, wave kernel signature, and spectral graph wavelet signature, which were used for validation purpose, by producing 94-95% classification accuracy by classifying normal and atrophy subjects correctly for CT, MR, and OASIS datasets.

Near-convex decomposition of 2D shape using visibility range

Part-based representation plays an important role in many shape related applications, including segmentation, recognition, editing and animation. An issue of particular interest in recent research is decomposing shapes into near-convex parts. However, it is usually challenging for existing methods to handle such heterogeneous real world shapes, especially when they possess long curved branches such as a lizard with a long curved tail. In this study, we propose a novel shape signature named visibility range, and a concavity measure based on this signature to describe the long curved branches. The visibility range reaches low values for points in concave regions and high values in convex regions, acting as the electrical charge distribution on the shape. Using these techniques, we present a coarse-to-fine approximate convex shape decomposition method, which separates the salient parts from the shape first and then refines the decomposition of the remaining main body of the shape by a visibility graph cut process. Qualitative and quantitative experiments have been conducted on shapes with various kinds of near-convex parts, demonstrating that our method captures the long curved branches as contiguous segments and outperforms the state-of-the-art methods that are based on other concave–convex features.

Low-complexity arrays of contour signatures for exact shape retrieval

We propose a framework for a fast exact shape retrieval called Low-complexity Arrays of Contour Signatures. The purposes are to match a shape against a database in constant time and to retrieve correct shapes very close to the query, while the latter may have undergone rigid transformations and noise. We present a shape signature based on prior works as well as a compact characterization of such signatures, a system of associative arrays allowing a short search time for retrieval and a technique of pairwise alignment. This method shows a good resilience to perturbations and is performed in constant computational time.

A Fast Point Clouds Registration Algorithm for Laser Scanners

Point clouds registration is an important step for laser scanner data processing, and there have been numerous methods. However, the existing methods often suffer from low accuracy and low speed when registering large point clouds. To meet this challenge, an improved iterative closest point (ICP) algorithm combining random sample consensus (RANSAC) algorithm, intrinsic shape signatures (ISS), and 3D shape context (3DSC) is proposed. The proposed method firstly uses voxel grid filter for down-sampling. Next, the feature points are extracted by the ISS algorithm and described by the 3DSC. Afterwards, the ISS-3DSC features are used for rough registration with the RANSAC algorithm. Finally, the ICP algorithm is used for accurate registration. The experimental results show that the proposed algorithm has faster registration speed than the compared algorithms, while maintaining high registration accuracy.

Climate and ice in the last glacial maximum explain patterns of isolation by distance inferred for alpine grasshoppers

Abstract Cold‐adapted species are likely to have had more widespread ranges and greater population connectivity during the last glacial period than is the case today. This contrasts with the trend in many species for range and population size to increase during interglacials. We examined the pattern of genetic and morphological variation within an endemic, wingless, alpine grasshopper Sigaus australis (Orthoptera: Acrididae) in the Southern Alps of New Zealand, testing for isolation by distance using geometric morphometric and mitochondrial ND2 sequences to document variation. Presence/absence data were analysed to estimate the environmental envelope (niche) of Sigaus australis and the resulting model used to infer the extent of available habitat for the species during the last glacial maximum. Estimates of past range size were modified using models of montane ice extent during the LGM. Clinal patterns of pronotum shape variation and signatures of isolation by distance support the hypothesis of a formerly more connected species. A north/south division was observed in pronotum shape, but the phenotypic variation was not diagnostic, as one would expect within a single species. Although the current habitat area occupied by Sigaus australis is much smaller than estimates for the LGM from our climate model, we show that realised area differed less due to the extension of valley glaciers. However, the current distribution of S. australis is more fragmented than in the past. This and other flightless alpine species currently restricted to fragmented high elevation habitat demonstrate genetic lag but are subject to loss of diversity as anthropogenic climate warming proceeds.

Loudness and short-term annoyance of sonic boom signatures at low levels.

Supersonic aircraft produce a sonic boom when flying faster than the speed of sound. To rule out detrimental effects for inhabitants of overflown areas, civil supersonic flights (like the Concorde) were only allowed to fly at supersonic speed when over water. Due to the progress in aircraft design, the supersonic boom may be reduced considerably in the future. In this study, listening tests were carried out with a variety of low boom simulations and conventional sonic boom signatures in a similar level range. Participants rated the loudness and the short-term annoyance of 24 sonic boom signals, which differed in terms of the signature shape and maximum pressure but were confined to a range of A-weighted sound exposure levels around 60 dB(A). The results showed main effects of signature and relative level variation as well as an interaction of the two. Correlation coefficients between the ratings and sound exposure levels were highest for A-weighted sound exposure levels compared to other frequency weightings. Contrary to our expectations, the provision of background information about the nature of the presented sound sources had no statistically significant influence on the ratings.

Topological radiation engineering in hyperbolic sonic semimetals

Hyperbolic dispersion enables unprecedented abilities for wave-field engineering which so far chiefly has been realized by man-made metamaterials. Recent classical explorations of topological media and semimetals suggest that these exotic structures may enable a novel route toward hyperbolic sound control. Here, we demonstrate that a three-dimensional acoustic semimetal implementation exhibits nodal lines of hyperbolic shape that are topologically protected due to translational symmetry. The structure comprises a cubic arrangement of crossed hollow channels whose hopping strengths are determined by their cross sections, which facilitate analytically exact prediction of tunable spin-Hall textures. Interestingly, thanks to these nodal signatures of hyperbolic shape, we are able to acquire an array of unusual emission features, spanning from directional collimation to either horizontally or vertically split radiation. We foresee that our findings will provide remarkable opportunities for advanced wave control with hyperbolic and topological attributes.

Next-generation osteometric sorting: Using 3D shape, elliptical Fourier analysis, and Hausdorff distance to optimize osteological pair-matching.

Determining which bilateral bones belong to the same person based on shape and size similarity is called pair-matching and it is instrumental for sorting commingled skeletons. To date, pair-matching has popularly been accomplished by visual inspection and/or linear caliper measurements; however, attention is turning increasingly to computational analysis. In this paper, we investigate a fast three-dimensional (3D) computerized shape-analysis method for whole-bone pair-matching using a test sample of 14 individuals (23 femora, 26 humeri, and 26 tibiae). Specifically, the method aims to find bilateral pairs using, as the shape signature criterion, a single 3D outline that snakes around each bone's perimeter as described by a 3D elliptical Fourier analysis function. This permits substantial 3D-point-cloud data reduction, that is, to 0.02% of the starting c.500,000 point cloud or just 100 points, while preserving key 3D shape information. The mean Hausdorff distance (Hd) was applied to measure the distance between each mirrored right-side outline to every left-side outline in pairwise fashion (132, 168 and 169 comparisons, respectively). Both thresholds and lowest Hd were investigated as pair-match criteria, with the lowest Hd producing the best performance results for searches jointly utilizing right-left and left-right directions for comparison: true positive rates of 1.00 (10/10), 1.00 (12/12), and 0.92 (11/12) for the femora, humeri, and tibiae, respectively. The computational time to calculate 469 pairwise 3D comparisons on a single stock-standard Intel® Core™ i7-4650U CPU @ 1.70GHz was 5s. This short data processing time makes the method viable for real-world application.

A SCALE AND ROTATION INVARIANT FAST IMAGE MINING FOR SHAPES

In recent times, fast content – based image retrieval is required in image mining for shapes, as image database is rising exponentially in size with time. In this paper, a sample, fast and efficient process through Distance Mapping for CBIR is proposed. In distance mapped domain, zero valued points represent the boundary of the object and the rest are assigned the value with the shortest distance to the boundary of the object. All the regional points within the object are defined with positive distance while the background points are given negative distances. The shape signature proposed is obtained from the statistical properties of the distance mapped functional. The features used are the number of grid points within pre-specified narrow band region across the object boundary and their algebraic relationship. It is found that, these features are invariant to sealing and rotations. T his increases the retrieval rate and also the speed of retrieval as no pre-scaling and rotation are required to register the shape. Conventional distance mapping is a time consuming process. This process can be used as coarse level in hierarchical CBIR that shrinks the database size from large set to a small one. This tiny database can further the database can be scrutinized meticulously using the above said methods like wavelets, curve lets etc. For feature extraction to improve the retriveval rate or even in the present work using accurate distance mapping.

Design, Implementation and Evaluation of a Support System for Educators and Therapists to Rate the Acquisition of Pre-Writing Skills

Assessing the acquisition of pre-writing skills in children with and without special educational needs is a time-consuming task for educators and therapists. It also involves a level of subjectivity, because the same set of strokes may receive different scores from different professionals. We present a system that automates the task by rating the execution of elementary figures (circle, square and triangle) according to the criteria of the Battelle guide for fine motor skills rating. The system uses a neural network trained with a collection of images drawn by 300 children and optimized through a systematic scan of hyperparameters, which revealed that shape signatures are better descriptors than Hu moments. Experiments carried out in collaboration with educators and therapists in Cuenca (Ecuador) provide evidence that the proposed system facilitates their work, automatically providing reliable assessments and in much shorter time than they would need for manual assessment, thus freeing their valuable time for education and therapy tasks.