Oblique Images Research Articles

Structure-aware completion of photogrammetric meshes in urban road environment

Photogrammetric mesh models obtained from aerial oblique images have been widely used for urban reconstruction. However, photogrammetric meshes suffer from severe texture problems, particularly in typical road areas, owing to occlusion. This paper proposes a structure-aware completion approach to improve mesh quality by seamlessly removing undesired vehicles. Specifically, a discontinuous texture atlas is first integrated into a continuous screen space by rendering trough a graphics pipeline. The rendering also records the necessary mapping for deintegration to the original texture atlas after editing. Vehicle regions are masked by a standard object detection approach, namely, Faster RCNN. Subsequently, the masked regions are completed, guided by the linear structures and regularities in the road region; this is implemented based on PatchMatch. Finally, the completed rendered image is deintegrated to the original texture atlas, and the triangles for the vehicles are also flattened so that improved meshes can be obtained. Experimental evaluation and analysis are conducted on three datasets, which were captured with different sensors and ground sample distances. The results demonstrate that the proposed method can produce quite realistic meshes after removing the vehicles. The structure-aware completion approach for road regions outperforms popular image completion methods, and an ablation study further confirms the effectiveness of the linear guidance. It should be noted that the proposed method can also handle tiled mesh models for large-scale scenes. Code and datasets are available at the project website11https://vrlab.org.cn/hanhu/projects/mesh.

Prevalence and site of predilection of carotid webs focusing on symptomatic and asymptomatic Japanese patients.

Carotid webs (CWs) have increasingly been recognized as a cause of recurrent ischemic stroke. However, the natural history and clinical course of CWs remain unclear. The authors aimed to clarify the prevalence, imaging features, and optimal treatment of CWs in a Japanese cohort study. A series of 444 consecutive Japanese patients who had undergone CTA of the head and neck between April 2011 and October 2016 was retrospectively reviewed. CW was diagnosed on CT angiograms as a membrane-like intraluminal filling defect along the posterior wall of the carotid bulb or the origin of the internal carotid artery (ICA) on oblique sagittal images and a corresponding thin septum on axial images. Two patients with CWs were identified among 132 patients with suspected stroke. The prevalence of CWs among symptomatic patients with suspected stroke was 1.5%. The prevalence of asymptomatic CWs was 2.2% (7 of 312 cases). The CWs were located in the posterior wall of the carotid bulb in 7 patients and just distal to the ICA origin in 2 patients. There were no apparent differences in the location or lesion length between symptomatic and asymptomatic CWs. Four of the 7 asymptomatic CWs remained asymptomatic for at least 2 years of follow-up. Two patients with symptomatic CWs developed recurrent cerebral infarction and transient ischemic attack despite being on a regimen of oral antiplatelet agents, and carotid endarterectomy was performed as radical treatment. Patients with CWs were younger than controls (median age 55 vs 69 years, p = 0.003) and were less frequently male than controls (33% vs 72%, p = 0.025). CW cases showed significantly fewer common atherosclerosis risk factors than the control group (p < 0.05). Although limited to CTA patients, this study reported on the prevalence and common site of CWs, focusing on symptomatic and asymptomatic Japanese patients. Extensive cross-sectional and prospective observational studies are warranted to elucidate the overall prevalence and natural history of CWs.

Oblique view individual tree crown delineation

Individual tree crown (ITC) segmentation supports numerous applications in forest management and ecology. In the latter context, special attention is dedicated to the study of angular reflection effects, caused by the interaction of incident sunlight with a canopy. High precision airborne analysis of these effects requires multi-view sensor systems and multi-view ITC segmentation. In particular oblique view image segmentation is difficult and has been addressed by numerous template based methods. This contribution identifies persistent shortcomings in the state of the art and tackles the problem by a multi-step workflow utilizing the digital surface model (DSM), derived from multi-view stereo data. A slightly revised version of the previously published levelset-watershed segmentation of the DSM is presented as the first step. In the second step, the contour of the visible part of a candidate tree in images with known orientation is obtained by means of ray casting and concave hull calculation. The method was tested on a deciduous, mixed and coniferous plot, whose aerial images were acquired using the 3K camera system in 2018 at Kranzberg Forest, Bavaria, Germany. Accuracies were assessed at hand of human operator generated groundtruth tree tops for the DSM as well as images with zenit angles of approximately 0,45 and 52 degrees. The resulting F1-scores, averaged over the plots, are 0.909/0.902/0.886/0.876 for the DSM/near-nadir-/oblique-/maximum-oblique-images, respectively.

Accuracy of omni-planar and surface casting of epileptiform activity for intracranial seizure localization.

Intracranial electroencephalography (ICEEG) recordings are performed for seizure localization in medically refractory epilepsy. Signal quantifications such as frequency power can be projected as heatmaps on personalized three-dimensional (3D) reconstructed cortical surfaces to distill these complex recordings into intuitive cinematic visualizations. However, simultaneously reconciling deep recording locations and reliably tracking evolving ictal patterns remain significant challenges. We fused oblique magnetic resonance imaging (MRI) slices along depth probe trajectories with cortical surface reconstructions and projected dynamic heatmaps using a simple mathematical metric of epileptiform activity (line-length). This omni-planar and surface casting of epileptiform activity approach (OPSCEA) thus illustrated seizure onset and spread among both deep and superficial locations simultaneously with minimal need for signal processing supervision. We utilized the approach on 41 patients at our center implanted with grid, strip, and/or depth electrodes for localizing medically refractory seizures. Peri-ictal data were converted into OPSCEA videos with multiple 3D brain views illustrating all electrode locations. Five people of varying expertise in epilepsy (medical student through epilepsy attending level) attempted to localize the seizure-onset zones. We retrospectively compared this approach with the original ICEEG study reports for validation. Accuracy ranged from 73.2% to 97.6% for complete or overlapping onset lobe(s), respectively, and ~56.1% to 95.1% for the specific focus (or foci). Higher answer certainty for a given case predicted better accuracy, and scorers had similar accuracy across different training levels. In an era of increasing stereo-EEG use, cinematic visualizations fusing omni-planar and surface functional projections appear to provide a useful adjunct for interpreting complex intracranial recordings and subsequent surgery planning.

Depth-enhanced feature pyramid network for occlusion-aware verification of buildings from oblique images

Detecting the changes of buildings in urban environments is essential. Existing methods that use only nadir images suffer from severe problems of ambiguous features and occlusions between buildings and other regions. Furthermore, buildings in urban environments vary significantly in scale, which leads to performance issues when using single-scale features. To solve these issues, this paper proposes a fused feature pyramid network, which utilizes both color and depth data for the 3D verification of existing buildings 2D footprints from oblique images. First, the color data of oblique images are enriched with the depth information rendered from 3D mesh models. Second, multiscale features are fused in the feature pyramid network to convolve both the color and depth data. Finally, multi-view information from both the nadir and oblique images is used in a robust voting procedure to label changes in existing buildings. Experimental evaluations using both the ISPRS benchmark datasets and Shenzhen datasets reveal that the proposed method outperforms the ResNet and EfficientNet networks by 5% and 2%, respectively, in terms of recall rate and precision. We demonstrate that the proposed method can successfully detect all changed buildings; therefore, only those marked as changed need to be manually checked during the pipeline updating procedure; this significantly reduces the manual quality control requirements. Moreover, ablation studies indicate that using depth data, feature pyramid modules, and multi-view voting strategies can lead to clear and progressive improvements.

Increased crystalline lens coverage in optical coherence tomography with oblique scanning and volume stitching

A three-dimensional optical coherence tomography (OCT) crystalline lens imaging method based on oblique scanning and image stitching is presented. The method was designed to increase OCT imaging volume of crystalline lens in vivo. A long-range swept-source (SS)-OCT imaging system, which can measure the entire anterior segment of eye in a single acquisition, is used to acquire one central volume and 4 extra volumes with different angles between optical axis of OCT instrument and the pupillary axis. The volumes are then stitched automatically by developed software. To show its effectiveness and verify its validity, we scanned the subjects before and after pupil dilation drops and compared the experimental results. By determining the number of voxels representing the signal from the crystalline lens in 3-D OCT images, our method can provide around 17% additional volumetric lens coverage compared with a regular imaging procedure. The proposed approach could be used clinically in early diagnosis of cortical cataract. Wider field of view offered by this method may facilitate more accurate lens biometry in its peripheral zones, which potentially contributes to understanding of lens shape modifications of the accommodating eye.

A hybrid global structure from motion method for synchronously estimating global rotations and global translations

Over the last few decades, the methods of global image orientation, which is also called global SfM, have attracted a lot of attention from researchers, mainly thanks to its advantage of time efficiency. Based on the input of relative orientation results, most conventional global SfM methods employ a two-step strategy consisting of global rotation estimation and global translation estimation. This paper, on the contrary, introduces a hybrid global approach that intends to solve global rotations and translations synchronously, but hierarchically. To improve the robustness and time efficiency, we first propose a novel efficient method that is much faster than the previous approaches for extracting an optimal minimum cover of a connected image triplet set (OMCTS). The OMCTS makes all the available images contained in a minimum number of connected image triplets, as well as all of those selected triplets, satisfy the constraint that the three corresponding relative orientations are as compatible as possible to each other. In order to solve non-collinear triplets in the OMCTS, some fundamental characterizations of essential matrices in the multiple-image setting are used, and image pose parameters are then estimated via averaging the constrained essential matrices. For the collinear triplets, the above approach is invalid and the image pose parameters are then alternatively determined from the relative orientations using the depth of tie points from each individual local spatial intersection. Finally, all image orientations are moved to a common coordinate system by traversing the solved connected triplets using similarity transformations. Compared to the state-of-the-art global SfM methods, the performance and capability of the proposed hybrid approach are thoroughly demonstrated on various public datasets (mainly including ordered and unordered internet images, oblique aerial images, hard and complex datasets, etc.).

Ultra-High Field Magnetic Resonance Imaging of the Retrobulbar Optic Nerve, Subarachnoid Space, and Optic Nerve Sheath in Emmetropic and Myopic Eyes.

PurposeWe aimed to image the optic nerve, subarachnoid space and optic nerve sheath in emmetropes and myopes ultra-high field (7-Tesla) magnetic resonance imaging (MRI). We targeted the retrobulbar distance of approximately 3 mm behind the eyeball, an area of clinical interest because of optic nerve sheath distensibility and pressure-related enlargement.MethodsEleven emmetropes (+0.75 to −0.50D, aged 20–41 years) and 10 myopes (−4.5 to −12D, aged 21–37 years) participated. Cross-sectional area of the optic nerve, subarachnoid space and optic nerve sheath at approximately 3 mm behind the eye were measured from two-dimensional T2-weighted coronal oblique MRI images obtained through the left optic nerve. Axial length of the left eye was measured from T2-weighted axial MRI images. In nine emmetropes and seven myopes, the optic nerve head was imaged with optical coherence tomography to compare retrobulbar and intraocular measures.ResultsRetrobulbar optic nerve, subarachnoid space and optic nerve sheath dimensions differed between myopes and emmetropes. Myopes tended to have smaller optic nerve and subarachnoid space. Longer MRI-derived axial length was associated with smaller optic nerve area (P = 0.03). Bruch's membrane opening area did not predict retrobulbar optic nerve area (P = 0.48).ConclusionsThis study demonstrates the feasibility of using 7-Tesla MRI to measure optic nerve, subarachnoid space, and optic nerve sheath dimensions behind the eye. In healthy adults, the retrobulbar optic nerve and subarachnoid space size are influenced by the degree of myopia.Translational Relevanceultra-high field MRI is a practical tool for assessing the morphometry of the optic nerve and surrounding anatomy behind the eye.

Serial right ventricular assessment in patients with hypoplastic left heart syndrome patients (HLHS): a cardiovascular magnetic resonance study

Abstract Funding Acknowledgements Type of funding sources: None. Background Patients with hypoplastic left heart syndrome (HLHS) are at risk for right ventricular (RV) dysfunction over the course of the three-stage surgical palliation with the final step being the completion of the total cavopulmonary connection (TCPC). However, less is known about RV function during follow-up after TCPC completion. We assessed RV function by analysing serial cardiovascular magnetic resonance (CMR) studies in a large cohort of HLHS patients. Methods CMR studies from 95 HLHS patients (67 males) were retrospectively analysed. Short axis cine images were used to measure RV end systolic and end diastolic volumes and ejection fraction (RVEF). Oblique cine images showing the atria and right ventricle in a similar manner like a standard "4-chamber view" were used to measure tricuspid annular plane systolic excursion (TAPSE) and long axis strain (LAS). Results From the 95 patients, all had at least two and 32 patients had three CMR scans. The first scan was performed at a mean age of 4.9 ± 2.8 years, the second scan at a mean age of 9.3 ± 4 years and the third at a mean age of 14.3 ± 3.7 years. The mean values of RV end diastolic and end systolic volume indexed to body surface area (REDVi, RVESVi) as well as RV ejection fraction (RVEF) at the three time points were: 1) REDVi 92.6 ± 21.9 ml/m2, RVESVi 43 ± 15.1 ml/m2, RVEF 54.2 ± 7.1%; 2) REDVi 93.9 ± 25.6 ml/m2, RVESVi 44.6 ± 18.3 ml/m2, RVEF 53.6 ± 7.8%; 3) REDVi 110.9 ± 41.9 ml/m2, RVESVi 58.1 ± 35 ml/m2, RVEF 50.1 ± 10.1%. There was a statistically significant increase in RVEDVi and RVESVi from the first and the second scan to the third scan (p &amp;lt; 0.01). RVEF was lower at the time of the third scan compared to the first and second scan, but this difference was not statistically significant. TAPSE increased slightly from the first to the third scan (p &amp;lt; 0.05). There was no change in stroke volume and LAS from the first to the third scan. Strong correlations were found between RVEF and LAS as well as between RVEF and TAPSE (r = 0.49 and r=-0.50; p &amp;lt; 0.001, respectively). Conclusion Serial assessment of CMR studies in HLHS patients after TCPC completion could demonstrate an increase in indexed RV volumes over time, whereas RV stroke volume, RVEF and LAS largely remain stable.

Arctic coastal erosion: UAV-SfM data collection strategies for planimetric and volumetric measurements

Above average warming in the Arctic is leading to increasing permafrost temperatures and a reduction in sea ice cover, which are expected to contribute to increasing rates of Arctic coastal erosion and sediment release. We studied a 1.5 km stretch of coastline off Richard’s Island, Northwest Territories, Canada, consisting of multiple retrogressive thaw slumps (RTSs) with varying degrees of activity over a one-year period. Multi-temporal 2D and 3D geomorphic analysis was based on unmanned aerial vehicle-Structure-from-Motion (UAV-SfM) data sets collected in 2018 and 2019. Over the observation period, −3.9 m and −1.1 m of planimetric cliff edge and toe retreat occurred, respectively, and corresponded to an average volumetric change of 8.1 m3 m−1. The accuracy of UAV-SfM-derived digital elevation models was tested using 12 data collection and processing scenarios, testing the influence of off-nadir camera angle, flight pattern, and georeferencing strategy. We found that oblique imaging and georeferencing strategy had a large influence on vertical accuracy and variability across the study site and has implications for studying volumetric changes in RTSs. This study furthers the geomorphological understanding of RTS processes by highlighting the complex relationship between planimetric and volumetric change along rapidly retreating Arctic coasts, and demonstrates advancements in measurement practices for UAV-SfM data sets.

Scanned versus Fused-Reconstructed Oblique MR-Images for Assessment of the Tibiofibular Syndesmosis—Diagnostic PerFormance and Reader Agreement

To evaluate the diagnostic performance and reader agreement of a novel MRI image fusion method enabling the reconstruction of oblique images for the assessment of the tibiofibular syndesmosis. We evaluated 40 magnetic resonance imaging examinations of patients with ankle sprains (16 with ruptures and 24 without) for the presence of anteroinferior tibiofibular ligament rupture. For all patients, we performed a fusion of standard two-dimensional transversal and coronal 3 mm PDw TSE images into an oblique-fusion reconstruction (OFR) and compared these against conventionally scanned oblique sequence for the evaluation of the tibiofibular syndesmosis. To evaluate diagnostic performance, two expert readers independently read the OFR images twice. We analyzed sensitivity, specificity, negative and positive predictive values, accuracy, and agreement. Reader 1 misinterpreted one OFR as a false negative, demonstrating a sensitivity of 0.94 and specificity of 1.00, reader 2 demonstrated perfect accuracy. Intrareader agreement was almost perfect for reader 1 (α = 0.95) and was perfect for reader 2 (α = 1.00). Additionally, interreader agreement between all fusion sequence reads was almost perfect (α = 0.97). The proposed OFR enables reliable detection of anteroinferior tibiofibular ligament rupture with excellent inter- and intrareader agreement, making conventional scanning of oblique images redundant and supplies a method to retroactively create oblique images, e.g., from external examinations.

Enhanced Region of Interest Extraction method for Finger Vein Images

Extracting Region of Interest (ROI) is an important step for finger vein recognition system. The purpose of this process is to determine the part of the image that we need for extracting features. In this paper we present an ROI extraction method that overcome the problems of finger rotation and displacement. We first locate the finger midline to be used in correcting the oblique images. We then use a sliding window to determine the Proximal inter phalangeal joint and to further identify the ROI height. Finally, from the corrected image of a certain height, the ROI is obtained through the use of finger edges internal tangents as ROI boundaries. The results prove that our method in a more accurate and effective manner in comparison with the method of [1], and thus enhance the performance of the system.

Femoral Torsion in CT: Why We Should Add Some Trigonometry

Purpose: In our department, we routinely perform Computed Tomography (CT) or Magnetic Resonance Imaging (MRI) of the lower extremity to assess femoral torsion for preoperative planning prior correcting osteotomies. It might be difficult to assess the FA, because on axial images the depicted part of the femoral neck is too short to allow correct alignment of the axis of the femoral neck. Measurements can also be performed on oblique axial images with improvement of depiction of the femoral neck but are smaller than those of measurements on axial images, due to the fact that images depend on rules of trigonometry. The aim is to provide a trigonometrical formula to calculate the value for FA, allowing precise assessment of femoral anteversion on oblique axial images similar to those on axial images. Materials and Methods: Trigonometrical transformation is performed in three steps. Initial measurement of femoral anteversion on oblique images is transformed via a corresponding rectangular triangle from coronal images to calculated femoral anteversion on axial images. Results: The first triangle is labelled with a1, b1, c1 for the sides and alpha1, beta1for the angles. Second and third triangle is labelled correspondingly. Length of both cathetus a1 and b1 are calculated as follows: cathetus a1=sin alpha1*c1 and cathetus b1=cos alpha 1*c1. Cathetus b2 is calculated as follows: b2=cos alpha2*c2. alpha3=tan-1 (a3/b3). Initial calculation of angle alpha3=tan-1 (a3/b3)=tan-1 (a1/b2), with a3=a1=sin alpha1*c1 and b3=b2=cos alpha2*b1*c, with b1=cos alpha1*c1 and b3=cos alpha2*cos alpha1*c1². The final formula is then: alpha3=tan-1 ((sin alpha1*c1)/(cos alpha²*cos alpha1*c12)) = tan-1 ((sin alpha1/(alpha2*cos alpha1*c1)) Conclusion: In this study we can provide a formula: alpha3=tan-1 ((sin alpha1/(cos alpha2*cos alpha1*c1)), which allows to calculate the femoral anteversion for true axial reconstructed images with the increased accuracy of measurements on oblique images and the ability to use the already known reference values from the literature.

Matching Large Baseline Oblique Stereo Images Using an End-to-End Convolutional Neural Network

The available stereo matching algorithms produce large number of false positive matches or only produce a few true-positives across oblique stereo images with large baseline. This undesired result happens due to the complex perspective deformation and radiometric distortion across the images. To address this problem, we propose a novel affine invariant feature matching algorithm with subpixel accuracy based on an end-to-end convolutional neural network (CNN). In our method, we adopt and modify a Hessian affine network, which we refer to as IHesAffNet, to obtain affine invariant Hessian regions using deep learning framework. To improve the correlation between corresponding features, we introduce an empirical weighted loss function (EWLF) based on the negative samples using K nearest neighbors, and then generate deep learning-based descriptors with high discrimination that is realized with our multiple hard network structure (MTHardNets). Following this step, the conjugate features are produced by using the Euclidean distance ratio as the matching metric, and the accuracy of matches are optimized through the deep learning transform based least square matching (DLT-LSM). Finally, experiments on Large baseline oblique stereo images acquired by ground close-range and unmanned aerial vehicle (UAV) verify the effectiveness of the proposed approach, and comprehensive comparisons demonstrate that our matching algorithm outperforms the state-of-art methods in terms of accuracy, distribution and correct ratio. The main contributions of this article are: (i) our proposed MTHardNets can generate high quality descriptors; and (ii) the IHesAffNet can produce substantial affine invariant corresponding features with reliable transform parameters.

Systematic review of radiological cervical foraminal grading systems.

The study design of this paper is systematic review. The purpose of this review is to evaluate the existing radiological grading systems that are used to assess cervical foraminal stenosis. The importance of imaging the cervical spine using CT or MRI in evaluating cervical foraminal stenosis is widely accepted; however, there is no consensus for standardized methodology to assess the compression of the cervical nerve roots. A systematic search of Ovid Medline databases, Embase 1947 to present, Cinahl, Web of Science, Cochrane Library, ISRCTN and WHO international clinical trials was performed for reports of cervical foraminal stenosis published before 01 February 2020. In collaboration with the University of Leeds, a search strategy was developed. A total of 6952 articles were identified with 59 included. Most of the reports involved multiple imaging modalities with standard axial and sagittal imaging used most. The grading themes that came from this systematic review show that the most mature for cervical foraminal stenosis is described by (Kim et al. Korean J Radiol 16:1294, 2015) and (Park et al. Br J Radiol 86:20120515, 2013). Imaging of the cervical nerve root canals is mostly performed using MRI and is reported using subjective terminology. The Park, Kim and Modified Kim systems for classifying the degree of stenosis of the nerve root canal have been described. Clinical application of these scoring systems is limited by their reliance on nonstandard imaging (Park), limited validation against clinical symptoms and surgical outcome data. Oblique fine cut images derived from three dimensional MRI datasets may yield more consistency, better clinical correlation, enhanced surgical decision-making and outcomes.

Drone Mapping Versus Terrestrial Laser Scanning for Building Mapping andAssesement

Drone mapping and terrestrial laser scanning are two advanced techniques that have been used around the world in recent years for documentation and mapping of buildings with historic value, construction project/infrastructure monitoring, mapping of cities, and a lot more survey. A well-defined methodology for data extraction, modelling, and monitoring is yet to evolve. This study compares and demonstrates terrestrial laser scanning and drone survey for building mapping with validation and accuracy checks. A terrestrial scanner was used to perform three scans of a building and drone was used to capture high-performance vertical images and oblique images. A three dimensional (3D) model was generated using advanced software techniques. Building facade elevation and plans were extracted from the 3D model after validation and accuracy check. The drone model produced high-quality visualization while terrestrial scanner output data allowed to extracting minute details after removing noise. Drone mapping and terrestrial laser scanning are two unique stand-alone technologies for serving different purposes individually, but they greatly complement each other when used together. Visualization of digital real-time 3D model helps in scrolling around at different nooks and corners of the building for a survey which is not possible through conventional mapping techniques; this is one of the greatest advantages of these techniques.

Single shot zonal oblique multislice SE-EPI diffusion-weighted imaging with low to ultra-high b-values for the differentiation of benign and malignant vertebral spinal fractures

PurposeTo investigate the diagnostic yield of low to ultra-high b-values for the differentiation of benign from malignant vertebral fractures using a state-of-the-art single-shot zonal-oblique-multislice spin-echo echo-planar diffusion-weighted imaging sequence (SShot ZOOM SE-EPI DWI). Materials and Methods66 patients (34 malignant, 32 benign) were examined on 1.5 T MR scanners. ADC maps were generated from b-values of 0,400; 0,1000 and 0,2000s/mm2. ROIs were placed into the fracture of interest on ADC maps and trace images and into adjacent normal vertebral bodies on trace images. The ADC of fractures and the Signal-Intensity-Ratio (SIR) of fractures relative to normal vertebral bodies on trace images were considered quantitative metrics. The appearance of the fracture of interest was graded qualitatively as iso-, hypo-, or hyperintense relative to normal vertebrae. ResultsADC achieved an area under the curve (AUC) of 0.785/0.698/0.592 for b = 0,400/0,1000/0,2000s/mm2 ADC maps respectively. SIR achieved an AUC of 0.841/0.919/0.917 for b = 400/1000/2000s/mm2 trace images respectively. In qualitative analyses, only b = 2000s/mm2 trace images were diagnostically valuable (sensitivity:1, specificity:0.794). Machine learning models incorporating all qualitative and quantitative metrics achieved an AUC of 0.95/0.98/0.98 for b-values of 400/1000/2000s/mm2 respectively. The model incorporating only qualitative metrics from b = 2000s/mm2 achieved an AUC of 0.97. ConclusionBy using quantitative and qualitative metrics from SShot ZOOM SE-EPI DWI, benign and malignant vertebral fractures can be differentiated with high diagnostic accuracy. Importantly qualitative analysis of ultra-high b-value images may suffice for differentiation as well.

Ultrasonographic Characteristics of Cervical Nerve Roots in Patients with the Degeneration of Cervical Vertebra

Background: Brachial plexus injury is a complex nerve injury. This may be due to the anatomical complexity of the brachial plexus. Purpose: To obtain sonographic characteristics and oblique sagittal diameter of cervical nerve root by high frequency ultrasound, and the morphology of brachial plexus was described. Methods: Two hundred and twenty-one subjects were recruited in our ultrasonography laboratory between March and December 2016. Their ultrasonographic characteristics of bilateral brachial plexus and cervical nerve roots were obtained by ultrasound, and they were measured with high echo sites on the bone cortex of anterior or posterior tubercles as positioning points in oblique sagittal. Results: 1. Cross-sectional sonogram of brachial plexus showed round and (or) oval “uniform” hypoechoic masses in the gap between anterior scalene muscle with middle scalene muscle, which were arranged in an arc shape; uniform low echo detected in short axis section and long axis section imaging of nerve bundles, which were wrapped by slightly hyper echoic epineuria. 2. Ultrasonographic imaging demonstrated that cervical nerve roots arranged between anterior tubercle and posterior tubercle, and the displaying rate of nerve roots of C4 to C7 was 100%; the displaying rate of nerve root of C8 was 78%. 3. Normal values of the diameters of cervicalr root nerves at intervertebral foramen were: C4, 2.65±0.27 mm; C5, 3.33±0.33 mm; C6, 3.76±0.36 mm; C7, 4.84±0.30 mm; C8, 3.48±0.34 mm. Conclusion: 1. Oblique sagittal imaging, nerve root of C7 is the thickest among all nerve roots of all examinee; 2. The diameter that of cervical nerve root of C4 may be positively correlated with the height.

Perceived Realism of High-Resolution Generative Adversarial Network-derived Synthetic Mammograms.

To explore whether generative adversarial networks (GANs) can enable synthesis of realistic medical images that are indiscernible from real images, even by domain experts. In this retrospective study, progressive growing GANs were used to synthesize mammograms at a resolution of 1280 × 1024 pixels by using images from 90 000 patients (average age, 56 years ± 9) collected between 2009 and 2019. To evaluate the results, a method to assess distributional alignment for ultra-high-dimensional pixel distributions was used, which was based on moment plots. This method was able to reveal potential sources of misalignment. A total of 117 volunteer participants (55 radiologists and 62 nonradiologists) took part in a study to assess the realism of synthetic images from GANs. A quantitative evaluation of distributional alignment shows 60%-78% mutual-information score between the real and synthetic image distributions, and 80%-91% overlap in their support, which are strong indications against mode collapse. It also reveals shape misalignment as the main difference between the two distributions. Obvious artifacts were found by an untrained observer in 13.6% and 6.4% of the synthetic mediolateral oblique and craniocaudal images, respectively. A reader study demonstrated that real and synthetic images are perceptually inseparable by the majority of participants, even by trained breast radiologists. Only one out of the 117 participants was able to reliably distinguish real from synthetic images, and this study discusses the cues they used to do so. On the basis of these findings, it appears possible to generate realistic synthetic full-field digital mammograms by using a progressive GAN architecture up to a resolution of 1280 × 1024 pixels.Supplemental material is available for this article.© RSNA, 2020.

Cattle Detection Using Oblique UAV Images

The evolution in imaging technologies and artificial intelligence algorithms, coupled with improvements in UAV technology, has enabled the use of unmanned aircraft in a wide range of applications. The feasibility of this kind of approach for cattle monitoring has been demonstrated by several studies, but practical use is still challenging due to the particular characteristics of this application, such as the need to track mobile targets and the extensive areas that need to be covered in most cases. The objective of this study was to investigate the feasibility of using a tilted angle to increase the area covered by each image. Deep Convolutional Neural Networks (Xception architecture) were used to generate the models for animal detection. Three experiments were carried out: (1) five different sizes for the input images were tested to determine which yields the highest accuracies; (2) detection accuracies were calculated for different distances between animals and sensor, in order to determine how distance influences detectability; and (3) animals that were completely missed by the detection process were individually identified and the cause for those errors were determined, revealing some potential topics for further research. Experimental results indicate that oblique images can be successfully used under certain conditions, but some practical limitations need to be addressed in order to make this approach appealing.