Real Scanning Research Articles

Generative Adversarial Network-based Noncontrast CT Angiography for Aorta and Carotid Arteries.

Background Iodinated contrast agents (ICAs), which are widely used in CT angiography (CTA), may cause adverse effects in humans, and their use is time-consuming and costly. Purpose To develop an ICA-free deep learning imaging model for synthesizing CTA-like images and to assess quantitative and qualitative image quality as well as the diagnostic accuracy of synthetic CTA (Syn-CTA) images. Materials and Methods A generative adversarial network (GAN)-based CTA imaging model was trained, validated, and tested on retrospectively collected pairs of noncontrast CT and CTA images of the neck and abdomen from January 2017 to June 2022, and further validated on an external data set. Syn-CTA image quality was evaluated using quantitative metrics. In addition, two senior radiologists scored the visual quality on a three-point scale (3 = good) and determined the vascular diagnosis. The validity of Syn-CTA images was evaluated by comparing the visual quality scores and diagnostic accuracy of aortic and carotid artery disease between Syn-CTA and real CTA scans. Results CT scans from 1749 patients (median age, 60 years [IQR, 50-68 years]; 1057 male patients) were included in the internal data set: 1137 for training, 400 for validation, and 212 for testing. The external validation set comprised CT scans from 42 patients (median age, 67 years [IQR, 59-74 years]; 37 male patients). Syn-CTA images had high similarity to real CTA images (normalized mean absolute error, 0.011 and 0.013 for internal and external test set, respectively; peak signal-to-noise ratio, 32.07 dB and 31.58 dB; structural similarity, 0.919 and 0.906). The visual quality of Syn-CTA and real CTA images was comparable (internal test set, P = .35; external validation set, P > .99). Syn-CTA showed reasonable to good diagnostic accuracy for vascular diseases (internal test set: accuracy = 94%, macro F1 score = 91%; external validation set: accuracy = 86%, macro F1 score = 83%). Conclusion A GAN-based model that synthesizes neck and abdominal CTA-like images without the use of ICAs shows promise in vascular diagnosis compared with real CTA images. Clinical trial registration no. NCT05471869 © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Zhang and Turkbey in this issue.

PROTECTION OF POLIGRAPHIC PRODUCTS USING DIGITAL WATERMARKS

The article proposes algorithms for steganographic embedding and extraction of digital watermarks (DWM)
 embedded in the digital original of printed products in order to protect copyrights to them. The built-in digital
 image is resistant to distortions introduced into images during printing and scanning and is not visually noticeable.
 The resistance of the extraction algorithm to geometric distortions is achieved through the use of an algorithm
 for restoring the original geometry of the image by comparing the special points of the distorted copy of the container
 image and the original image. The results of testing the robustness of the algorithm both to simulated distortions
 of the printing and scanning channel, and to distortions introduced into images when using real printing and scanning
 devices are presented. The impact of the embedding process on image quality is also considered.

Reservoir tortuosity prediction: Coupling stochastic generation of porous media and machine learning

Accurate reservoir tortuosity prediction is the foundation of the high-quality evaluation of reservoir petrophysical properties. However, conventional empirical equations as a form of experimental data fitting lacks universality because the data are usually from a single horizon or block. We developed a model combining the stochastic generation of porous media with machine learning (ML) to predict reservoir tortuosity based on pore structure parameters. Real core scanning images from public databases were employed in stochastic generation as reference, which is an economic and accurate method of meeting the dataset quality and scale requirements of ML. The particle swarm optimization algorithm, an efficient method of obtaining the best hyperparameter combination, was introduced for the hyperparameter tuning of six commonly used ML algorithms to determine the optional model for tortuosity prediction. Our trained ML models demonstrated superior tortuosity prediction accuracy over deterministic linear and exponential empirical equations with porosity as the only variable, which effectively demonstrates the potential of tortuosity prediction using pore structure parameters. The proposed ML model enables precise tortuosity predictions based on a few measurable pore structural features, which can be obtained from well logging data and CT scanning; thus, it can be widely used in petroleum and logging fields.

As-built BIM reconstruction of piping systems using smartphone videogrammetry and terrestrial laser scanning

The increasing adoption of Building Information Modeling (BIM) in the construction industry has created a demand for methods to update existing facilities with accurate as-built BIM models. While terrestrial laser scanners (TLS) are commonly used for indoor surveys due to their speed and precision, they face limitations in areas with physical constraints like confined spaces and occlusions, with pipes being particularly affected. This paper presents a smartphone videogrammetry approach designed to enhance the accuracy and completeness of terrestrial laser scanner (TLS) point clouds for as-built BIM reconstruction. The method involves capturing missing areas within the TLS point cloud with smartphones and then seamlessly integrating these data into a more comprehensive multi-modal point cloud. This approach is validated using real scan data and offers convenience through the use of readily available and cost-effective smartphones, while its iterative nature facilitates progressive BIM reconstruction of pipe structures. Additionally, a compatible progressive BIM reconstruction framework is outlined, aiming to modernize scan-to-BIM practices in the construction industry, with a case study demonstrating its feasibility.

Visual programming simulator for producing realistic labeled point clouds from digital infrastructure models

The increasing availability of point clouds has led to intensive research into automating point cloud processing using machine learning. While supervised systems require large and diverse labeled datasets, the cost and time of manual data creation can be overcome with synthetic data. This paper introduces DynamoPCSim, a versatile scanning simulator based on visual programming, implementing ray tracing, and operating on BIM models. The simulator collects measurements of digital models and transfers the model semantic data to generated point clouds, enabling automated labeling. Customizable scanning parameters allow for the reflection of real scanners (including imperfections) and the transformation of synthetic point clouds, making the data more realistic. The evaluation of generated point clouds against real-world data through a neural network segmentation experiment provides a foundation for the effective utilization of DynamoPCSim synthetic point clouds in machine learning training.

Groupwise Shape Correspondence Refinement with a Region of Interest Focus

AbstractWhile collections of scan shapes are becoming more prevalent in many real‐world applications, finding accurate and dense correspondences across multiple shapes remains a challenging task. In this work, we introduce a new approach for refining non‐rigid correspondences among a collection of 3D shapes undergoing non‐rigid deformation. Our approach incorporates a Region Of Interest (ROI) into the refinement process, which is specified by the user on one shape within the collection. Based on the functional map framework and more specifically on the notion of cycle‐consistency, our formulation improves the overall matching consistency while prioritizing that of the region of interest. Specifically, the initial pairwise correspondences are refined by first defining the localized harmonics that are confined within the transferred ROI on each shape, and subsequently applying the CCLB (Canonical Consistent Latent Basis) framework both on the global and the localized harmonics. This leads to an enhanced matching accuracy for both the ROIs and the overall shapes across the collection. We evaluate our method on various synthetic and real scan datasets, in comparison with the state‐of‐the‐art techniques.

Gaze-probe joint guidance with multi-task learning in obstetric ultrasound scanning

In this work, we exploit multi-task learning to jointly predict the two decision-making processes of gaze movement and probe manipulation that an experienced sonographer would perform in routine obstetric scanning. A multimodal guidance framework, Multimodal-GuideNet, is proposed to detect the causal relationship between a real-world ultrasound video signal, synchronized gaze, and probe motion. The association between the multi-modality inputs is learned and shared through a modality-aware spatial graph that leverages useful cross-modal dependencies. By estimating the probability distribution of probe and gaze movements in real scans, the predicted guidance signals also allow inter- and intra-sonographer variations and avoid a fixed scanning path. We validate the new multi-modality approach on three types of obstetric scanning examinations, and the result consistently outperforms single-task learning under various guidance policies. To simulate sonographer’s attention on multi-structure images, we also explore multi-step estimation in gaze guidance, and its visual results show that the prediction allows multiple gaze centers that are substantially aligned with underlying anatomical structures.

Meta-evaluation for 3D Face Reconstruction Via Synthetic Data.

The standard benchmark metric for 3D face reconstruction is the geometric error between reconstructed meshes and the ground truth. Nearly all recent reconstruction methods are validated on real ground truth scans, in which case one needs to establish point correspondence prior to error computation, which is typically done with the Chamfer (i.e., nearest neighbor) criterion. However, a simple yet fundamental question have not been asked: Is the Chamfer error an appropriate and fair benchmark metric for 3D face reconstruction? More generally, how can we determine which error estimator is a better benchmark metric? We present a meta-evaluation framework that uses synthetic data to evaluate the quality of a geometric error estimator as a benchmark metric for face reconstruction. Further, we use this framework to experimentally compare four geometric error estimators. Results show that the standard approach not only severely underestimates the error, but also does so inconsistently across reconstruction methods, to the point of even altering the ranking of the compared methods. Moreover, although non-rigid ICP leads to a metric with smaller estimation bias, it could still not correctly rank all compared reconstruction methods, and is significantly more time consuming than Chamfer. In sum, we show several issues present in the current benchmarking and propose a procedure using synthetic data to address these issues.

Machine learning-based early diagnosis of autism according to eye movements of real and artificial faces scanning.

Studies on eye movements found that children with autism spectrum disorder (ASD) had abnormal gaze behavior to social stimuli. The current study aimed to investigate whether their eye movement patterns in relation to cartoon characters or real people could be useful in identifying ASD children. Eye-tracking tests based on videos of cartoon characters and real people were performed for ASD and typically developing (TD) children aged between 12 and 60 months. A three-level hierarchical structure including participants, events, and areas of interest was used to arrange the data obtained from eye-tracking tests. Random forest was adopted as the feature selection tool and classifier, and the flattened vectors and diagnostic information were used as features and labels. A logistic regression was used to evaluate the impact of the most important features. A total of 161 children (117 ASD and 44 TD) with a mean age of 39.70 ± 12.27 months were recruited. The overall accuracy, precision, and recall of the model were 0.73, 0.73, and 0.75, respectively. Attention to human-related elements was positively related to the diagnosis of ASD, while fixation time for cartoons was negatively related to the diagnosis. Using eye-tracking techniques with machine learning algorithms might be promising for identifying ASD. The value of artificial faces, such as cartoon characters, in the field of ASD diagnosis and intervention is worth further exploring.

Illustrated argument for CT-scanning a whole car for the forensic investigation of projectile holes, defects, fragments and possible trajectories

Contemporary documentation of a car with bullet defects after a shooting incident can secure the usual tracks and gunshot residue, take photographs, and use trajectory rods and probes. Since the advent of the “XXL-CT -Scanner” (Fraunhofer Institute, Germany), we have wondered if the advantages of volume scanning CT, already noted for forensic pathology, could be applied to cars. To this end, we damaged a small 3D-printed car model with an electric drill and added CT -dense material with a soldering iron, simulating linearly configured defect morphologies with metal particles. This model was CT -scanned and the resulting data visualized to illustrate how these visualizations can support reconstructive visualization of trajectories. Performing a real XXL-CT scan of a bullet-riddled car requires extensive preparation, transportation, and other logistical measures that are costly and time-consuming. Nonetheless, we suggest that this is a worthwhile research direction to explore.

Simulation of Laser Profilometer Measurements in the Presence of Speckle Using Perlin Noise.

In the manufacturing industry, inspection systems play a crucial role in ensuring product quality. High-resolution profilometric sensors have become increasingly popular for inspection due to their ability to provide detailed surface information. However, the development and testing of inspection systems can be costly and time-consuming. This paper presents the development of a simulation of an inspection system using a high-resolution profilometric sensor. A geometrical and noise model is proposed to simulate the readings of any actual profilometric sensor. The model replicates the sensor's movement on the CAD model of the inspected part. The model incorporates the physical properties of the sensor and combines noise sources from sensor uncertainty and speckle noise induced by the roughness of the material. Our contribution lies in noise modeling. This work proposes a combination of Perlin noise to simulate the speckle noise and Gaussian noise for the uncertainty-related noise. Perlin noise is generated based on the surface roughness parameters of the inspected part. The accuracy of the simulation system is evaluated by comparing the simulated scans with real scans. The results highlight the ability to simulate real scans of different parts, using commercial sensor specifications and the CAD model of the inspected part.

A customized anthropomorphic 3D-printed phantom to reproducibility assessment in computed tomography: an oncological case study.

Studies on computed tomography (CT) reproducibility at different acquisition parameters have to take into account radiation dose administered and related ethical issues. 3D-printed phantoms provide the possibility to investigate these features deeply and to foster CT research, also taking advantage by outperforming new generation scanners. The aim of this study is to propose a new anthropomorphic 3D-printed phantom for chest lesions, tailored on a real patient CT scan, to investigate the variability of volume and Hounsfield Unit (HU) measurements at different CT acquisition parameters. The chest CT of a 75-year-old patient with a paramediastinal lung lesion was segmented based on an eight-compartment approach related to HU ranges (air lung, lung interstitium, fat, muscle, vascular, skin, bone, and lesion). From each mask produced, the 3D.stl model was exported and linked to a different printing infill value, based on a preliminary test and HU ratios derived from the patient scan. Fused deposition modeling (FDM) technology printing was chosen with filament materials in polylactic acid (PLA). Phantom was acquired at 50 mAs and three different tube voltages of 80, 100, and 120 kVp on two different scanners, namely, Siemens Somatom Force (Siemens Healthineers, Erlangen, Germany; same setting of real patient for 80 kVp acquisition) and GE 750 HD CT (GE Healthcare, Chicago, IL). The same segmentation workflow was then applied on each phantom acquisition after coregistration pipeline, and Dice Similarity Coefficient (DSC) and HU averages were extracted and compared for each compartment. DSC comparison among real patient versus phantom scans at different kVp, and on both CT scanners, demonstrated a good overlap of different compartments and lesion vascularization with a higher similarity for lung and lesion masks for each setting (about 0.9 and 0.8, respectively). Although mean HU was not comparable with real data, due to the PLA material, the proportion of intensity values for each compartment remains respected. The proposed approach demonstrated the reliability of 3D-printed technology for personalized approaches in CT research, opening to the application of the same workflow to other oncological fields.

Efficiently Training Vision Transformers on Structural MRI Scans for Alzheimer's Disease Detection.

Neuroimaging of large populations is valuable to identify factors that promote or resist brain disease, and to assist diagnosis, subtyping, and prognosis. Data-driven models such as convolutional neural networks (CNNs) have increasingly been applied to brain images to perform diagnostic and prognostic tasks by learning robust features. Vision transformers (ViT) - a new class of deep learning architectures - have emerged in recent years as an alternative to CNNs for several computer vision applications. Here we tested variants of the ViT architecture for a range of desired neuroimaging downstream tasks based on difficulty, in this case for sex and Alzheimer's disease (AD) classification based on 3D brain MRI. In our experiments, two vision transformer architecture variants achieved an AUC of 0.987 for sex and 0.892 for AD classification, respectively. We independently evaluated our models on data from two benchmark AD datasets. We achieved a performance boost of 5% and 9-10% upon fine-tuning vision transformer models pre-trained on synthetic (generated by a latent diffusion model) and real MRI scans, respectively. Our main contributions include testing the effects of different ViT training strategies including pre-training, data augmentation and learning rate warm-ups followed by annealing, as pertaining to the neuroimaging domain. These techniques are essential for training ViT-like models for neuroimaging applications where training data is usually limited. We also analyzed the effect of the amount of training data utilized on the test-time performance of the ViT via data-model scaling curves.Clinical Relevance- The models evaluated in this work could be trained on neuroimaging data to assist in diagnosis, subtyping and prognosis of Alzheimer's disease.

Vector method of strain estimation in OCT-elastography with adaptive choice of scale for estimating interframe phase-variation gradients

In this paper we present a method which significantly improves strain-estimation quality in phase-sensitive optical coherence elastography (OCE). Specifically, we consider the realization of phase-sensitive OCE termed ‘vector method’, among main advantages of which are its high computational efficiency and high robustness with respect to measurement noises. The method does not require any search operations, but the quality of local strain estimation heavily depends on the quality of numerical spatial differentiation of inter-frame phase variations. This differentiation does not require phase unwrapping even for supra-wavelength displacements of scatterers, but the differentiation results depend on the chosen scale for phase-gradient estimation. Either too small or too large scale may strongly degrade the results of elastographic visualization, especially in the presence of pronounced spatio-temporal strain inhomogeneity. The proposed adaptive automatic choice of the differentiation scale complements earlier proposed vector averaging and significantly improves strain-mapping quality in OCE. The method efficiency is demonstrated using both numerically simulated and real OCT scans.

Robust pose estimation for ship block assembly feature based on large-scale scanning

PurposeHull block assembly is a vital task in ship construction. It is necessary to obtain the actual poses of the assembly features to guide further block alignment. Traditional methods use single-point measurement, which is time-consuming and may lead to loss of key information. Thus, large-scale scanning is introduced for data acquisition, and this paper aims to provide a precise and robust method for retrieving poses based on point set registration.Design/methodology/approachThe main problem of point registration is to find the correct transformation between the model and the scene. In this paper, a vote framework based on a new point pair feature is used to calculate the transformation. First, a special edge indicator for multiplate objects is proposed to determine the edges. Subsequently, pair features with an edge description are noted for every point. Finally, a voting scheme based on agglomerative clustering is implemented to determine the optimal transformation.FindingsThe proposed method not only improves registration efficiency but also maintains high accuracy compared to several commonly used approaches. In particular, for objects composed of plates, the results of pose estimation are more promising because of the compact pair feature. The multiple ship longitudinal localization experiment validates the effectiveness in real scan applications.Originality/valueThe proposed edge description performs a better detection for the edges of multiplate objects. The pair feature incorporating the edge indicator is more discriminative than the original template, resulting in better robustness to outliers, noise and occlusions.

“I think this could be a big success” – A mixed methods study on practitioner perspectives on the acceptance of a virtual reality tool for preparation in MRI

IntroductionA key part of a radiographer's role within MRI is providing the required emotional support to help patients succeed with a scan. Being informed is important; whilst information leaflets and videos are commonly used, these can be limited in their representation of the experience. Virtual reality tools are being shown to reasonably replicate a scan experience, having a positive impact on patient satisfaction and anxiety. The aim was to obtain the views of practitioners on the use and implementation of such a tool in practice. MethodsA mixed methods study was conducted looking at the use of a virtual scan experience for patients prior to MRI. Nine radiographers attended two focus group sessions to see the tool and undergo a virtual experience. Following this, a survey based on the technology acceptance model was completed along with a semi-structured discussion about its use. ResultsPerceived usefulness, ease of use, attitude and intention to use were all positive towards the virtual scan tool. All practitioners saw value in such a tool and how it could be implemented within practice, highlighting areas for improvement and development. ConclusionThe practitioner's perspective was that access to such a virtual scan experience could be of use to better prepare and support those patients needing extra support before a real scan. Acknowledgement of having time to discuss patient concerns was noted and this could provide a means of doing so away from busy scanning lists whilst not taking up additional time. Implications for practiceUse of VR tools could be a conduit through which trust and rapport are built in advance away from busy scanning lists, thereby not impacting on operational throughput and hindering efficiency.

A robust workflow for b-rep generation from image masks

A novel approach to generating watertight, manifold boundary representations from noisy binary image masks of MRI or CT scans is presented. The method samples an input segmented image and locally approximates the material boundary. Geometric error metrics between the voxelated boundary and an approximating template surface are minimized, and boundary point/normals are correspondingly generated. Voronoi partitioning is employed to perform surface reconstruction on the resulting oriented point cloud. The method performs competitively against other approaches, both in comparisons of shape and volume error metrics to a canonical image mask, and in qualitative comparisons using noisy image masks from real scans. The framework readily admits enhancements for capturing sharp edges and corners. The approach robustly produces high-quality b-reps that may be inserted into an image-based meshing pipeline for purposes of physics-based simulation.

Dynamic low-count PET image reconstruction using spatio-temporal primal dual network

Objective. Dynamic positron emission tomography (PET) imaging, which can provide information on dynamic changes in physiological metabolism, is now widely used in clinical diagnosis and cancer treatment. However, the reconstruction from dynamic data is extremely challenging due to the limited counts received in individual frame, especially in ultra short frames. Recently, the unrolled model-based deep learning methods have shown inspiring results for low-count PET image reconstruction with good interpretability. Nevertheless, the existing model-based deep learning methods mainly focus on the spatial correlations while ignore the temporal domain. Approach. In this paper, inspired by the learned primal dual (LPD) algorithm, we propose the spatio-temporal primal dual network (STPDnet) for dynamic low-count PET image reconstruction. Both spatial and temporal correlations are encoded by 3D convolution operators. The physical projection of PET is embedded in the iterative learning process of the network, which provides the physical constraints and enhances interpretability. Main results. The experiments of both simulation data and real rat scan data have shown that the proposed method can achieve substantial noise reduction in both temporal and spatial domains and outperform the maximum likelihood expectation maximization, spatio-temporal kernel method, LPD and FBPnet. Significance. Experimental results show STPDnet better reconstruction performance in the low count situation, which makes the proposed method particularly suitable in whole-body dynamic imaging and parametric PET imaging that require extreme short frames and usually suffer from high level of noise.

NeAF: Learning Neural Angle Fields for Point Normal Estimation

Normal estimation for unstructured point clouds is an important task in 3D computer vision. Current methods achieve encouraging results by mapping local patches to normal vectors or learning local surface fitting using neural networks. However, these methods are not generalized well to unseen scenarios and are sensitive to parameter settings. To resolve these issues, we propose an implicit function to learn an angle field around the normal of each point in the spherical coordinate system, which is dubbed as Neural Angle Fields (NeAF). Instead of directly predicting the normal of an input point, we predict the angle offset between the ground truth normal and a randomly sampled query normal. This strategy pushes the network to observe more diverse samples, which leads to higher prediction accuracy in a more robust manner. To predict normals from the learned angle fields at inference time, we randomly sample query vectors in a unit spherical space and take the vectors with minimal angle values as the predicted normals. To further leverage the prior learned by NeAF, we propose to refine the predicted normal vectors by minimizing the angle offsets. The experimental results with synthetic data and real scans show significant improvements over the state-of-the-art under widely used benchmarks. Project page: https://lisj575.github.io/NeAF/.

A method for quantification of noise non-uniformity in computed tomography images: A computational study

In computed tomography (CT), the noise is sometimes non-uniform, i.e. the noise magnitude may vary with the gradient level within the image. The purpose of this study was to quantify the noise non-uniformity in CT images using appropriate 1D and 2D computational phantoms, and to validate the effectiveness of the proposed concept in images filtered by the bilateral filter (BF), as an example of a non-linear filter. We first developed 1D and 2D computational phantoms, and Gaussian noises with several noise levels were then added to the phantoms. In addition, to simulate the real form of noise from images obtained in a real CT scanner, a homogeneous water phantom image was used. These noise levels were referred to as ground truth noise (σG). The phantoms were then filtered by the bilateral filter with various pixel value spreads (σ) to produce non-uniform noise. The original gradient phantoms (G) were subtracted from both the noisy phantoms (IN) and the filtered noisy phantoms (IBF), and the magnitudes of the resulting noise for each gradient were computed. The noise-gradient dependency (NGD) curve was used to display the dependency of noise magnitude on image gradient in the non-uniform noise. It is found that for uniform noise, the magnitude of noise was constant for all gradients. However, for non-uniform noise, the measured noise was dependent on the gradient levels and on the strength of the BF for every ground truth noise (σG). It was found that the noise magnitude was large for the large gradients and decreased with the magnitude of the image gradient.