Tomography Datasets Research Articles

Five-Year Experience With the Zenith Fenestrated Endovascular Graft for Treatment of Juxtarenal Abdominal Aortic Aneurysms at a Medium Volume Center

The use of fenestrated endovascular aortic repair (FEVAR) to treat abdominal aortic aneurysms involving the visceral arteries has increased worldwide with excellent results. The aim of this study was to evaluate the safety and effectiveness of the Zenith fenestrated endovascular graft (Cook Medical, Bloomington, IN) for the treatment of juxtarenal abdominal aortic aneurysms (AAAs) at a medium volume center. We performed a 5-year (2015-2020) retrospective analysis of all patients who had presented to our institution with juxtarenal AAAs and undergone FEVAR. Custom-made fenestrated stent grafts were designed with one to three fenestrations after analysis of computed tomography data sets. Renal alignment was performed with balloon-expandable stents. Follow-up included clinical examination, laboratory studies, mesenteric-renal duplex ultrasound, abdominal radiography, and computed tomography at hospital discharge, at 1, 6, and 12 months, and annually thereafter for ≤5 years. A total of 25 patients with juxtarenal AAAs had undergone FEVAR during the study period and were included in our analysis. Of the 25 patients, 18 were men and 7 were women, with a mean age of 71 ± 10 years. The mean aneurysm diameter was 60 ± 5 mm. A total of 53 visceral arteries required incorporation with fenestrations. Of these, 51 (96%) fenestrations were aligned by stents. Technical success was 100%. One intraoperative complication occurred. The patient developed bilateral common femoral artery plaque dissection and required common femoral artery endarterectomy. The mean length of hospital stay was 2.8 ± 1.5 days. No aneurysm ruptures or conversions were noted during a mean follow-up of 41 ± 15 months (range, 6-67 months). Of a total of 50 renal arteries targeted by fenestration, one renal artery occlusion (2%) and five cases of stenosis (10%) developed. Five patients (20%) required secondary interventions for renal artery stenosis or occlusion. One type I endoleak developed in 2 patients and one type II endoleak developed in 11 patients. One patient with a type II endoleak had experienced an increase in sac size at 1 year and underwent inferior mesenteric artery ligation. None of the patients died during the follow-up period. Our study has demonstrated that endovascular repair of juxtarenal AAAs with the Zenith fenestrated AAA stent graft at a medium volume center is safe and effective. Mortality and morbidity are low for properly selected patients treated at centers with experience in these procedures.

Retrospective Investigation of Cranial Volume and Cephalic Index in Patients with Nonsyndromic Sagittal Synostosis Operated by Total Vault Remodeling

Background: Premature fusion of cranial sutures affects skull development and leads to head deformity. Intracranial pressure increase and brain growth restriction can occur in untreated craniosynostosis. Operative treatment aims to achieve an immediate and long-lasting correction of skull shape that is close to the average and to prevent or release possible increased intracranial pressure by increasing the intracranial volume (ICV) or normalizing the ICV if it is already below the standards. This study was designed to evaluate the effect of a total calvarial reconstruction on skull development in patients with nonsyndromic sagittal synostosis. Material and Methods: The study population included 19 male and 5 female patients with isolated nonsyndromic sagittal suture synostosis. During the operation, temporarily fixed prebent metal plates provided an intraoperative reference for the desired cranial expansion gain of height and shortening. Preoperative and postoperative ICVs and cephalic indices were measured on computed tomography datasets using the software program ImageJ and were compared with one another and with normative data. Results: The male population presented with a preoperative mean ICV of 863.3 cm³. A postoperative mean ICV increase of 243.5 cm³ (p < 0.001) and a further ICV enlargement (p < 0.001) was measured. The mean CI changed from 71.0% preoperatively to 75.4% postoperatively (p = 0.002) and decreased insignificantly in the follow-up (p = 0.546). The female population had a preoperative mean ICV of 804.9 cm³. Postoperatively, the mean ICV increased by 211.1 cm³ (p = 0.043) and also increased in the follow-up (p = 0.043). Their mean CI values increased from 66.5% preoperatively to 72.8% (p = 0.043) postoperatively and decreased insignificantly in the follow-up (p = 0.345). Conclusion: This method of total vault remodeling provides reliable ICV increase and improvement in length and width of skull proportions beyond the immediate postoperative period together with an ICV increase.

Impact of split X-jaw technique on target volume coverage and organ at risk sparing in prostate cancer: a comparative dosimetric study

Abstract Introduction: The Varian Trilogy linear accelerator’s multi-leaf collimator moves on a carriage with a maximum leaf span of 15 cm. The traditional open and limited X-jaw technique of volumetric-modulated arc radiotherapy (VMAT) yields a relatively compromised dose distribution within the planning target volume (PTV) region. This study aimed to determine whether the split X-jaw planning technique for VMAT improves plan quality regarding target dose coverage and organs at risk (OAR) sparing for PTVs that require a field size of more than 15 cm in the X-jaw direction in prostate cancer patients. Method: Computed tomography data sets from 15 patients with prostate cancer were enrolled in the study. Only the PTVs requiring a field size larger than 18 cm in the X-jaw position were considered, and a dose of 4500 cGy in 25 fractions was prescribed. For each case, three separate treatment plans were generated: open, limited and split X-jaw planning techniques with similar planning objectives Results: The split X-jaw technique resulted in statistically significant superior coverage of PTV when compared with the open (P &lt; 0·0001) and limited methods (P &lt; 0·001). The split technique delivered a lower dose to the OARs, although statistical significance could not be achieved. D2% (cGy) was lowest for the PTV in the split technique (4684·8 ± 18·16) and highest for the open technique (4710 ± 18·75), P &lt; 0·001. Conclusion: The x-split jaw technique can replace the traditional open X-jaw practice of VMAT for PTVs requiring an X-jaw width of more than 15 cm in the Varian linear accelerator.

Convolutional neural network for automatic maxillary sinus segmentation on cone-beam computed tomographic images

An accurate three-dimensional (3D) segmentation of the maxillary sinus is crucial for multiple diagnostic and treatment applications. Yet, it is challenging and time-consuming when manually performed on a cone-beam computed tomography (CBCT) dataset. Recently, convolutional neural networks (CNNs) have proven to provide excellent performance in the field of 3D image analysis. Hence, this study developed and validated a novel automated CNN-based methodology for the segmentation of maxillary sinus using CBCT images. A dataset of 264 sinuses were acquired from 2 CBCT devices and randomly divided into 3 subsets: training, validation, and testing. A 3D U-Net architecture CNN model was developed and compared to semi-automatic segmentation in terms of time, accuracy, and consistency. The average time was significantly reduced (p-value < 2.2e−16) by automatic segmentation (0.4 min) compared to semi-automatic segmentation (60.8 min). The model accurately identified the segmented region with a dice similarity co-efficient (DSC) of 98.4%. The inter-observer reliability for minor refinement of automatic segmentation showed an excellent DSC of 99.6%. The proposed CNN model provided a time-efficient, precise, and consistent automatic segmentation which could allow an accurate generation of 3D models for diagnosis and virtual treatment planning.

The arteries of the musculoskeletal system of siamangs, and a comparison with other hylobatids, greater apes, and humans.

Limited research on the gross anatomy of the blood vessels has been conducted on hylobatids, or lesser apes, so far. Here, we present a detailed study of the arteries of siamangs (Symphalangus) and compare our findings with data compiled from our previous studies as well as from the literature about other hylobatids, greater apes, and humans. In particular, a three-dimensionalfull-body computed tomography data set of a siamang neonate was analyzed in detail for this study, with notable findings in the head and neck, thorax, upper limb, abdomen and pelvis, and lower limb. Of the 62 arteries that we studied in detail, a total of 20 arteries that have never been described in detail in hylobatids are reported in this study. Key similarities to other apes differing from humans include the existence of a humeral common circumflex trunk and the origination of the dorsalis pedis from the posterior tibial artery or saphenous artery instead of the anterior tibial artery. Similarities to humans differing from other apes include the separation of the lingual and facial arteries and the origination of the profunda brachii from the brachial artery instead of the axillary artery. Our research and broader comparisons, therefore, contribute to knowledge about the soft tissues of hylobatids, other apes, and primates in general and facilitate a better understanding of the anatomical evolution and key differences and similarities among these taxa.

Using 4D dose accumulation to calculate organ-at-risk dose deviations from motion-synchronized liver and lung tomotherapy treatments.

Tracking systems such as Radixact Synchrony change the planned delivery of radiation during treatment to follow the target. This is typically achieved without considering the location changes of organs at risk (OARs). The goal of this work was to develop a novel 4D dose accumulation framework to quantify OAR dose deviations due to the motion and tracked treatment. The framework obtains deformation information and the target motion pattern from a four‐dimensional computed tomography dataset. The helical tomotherapy treatment plan is split into 10 plans and motion correction is applied separately to the jaw pattern and multi‐leaf collimator (MLC) sinogram for each phase based on the location of the target in each phase. Deformable image registration (DIR) is calculated from each phase to the references phase using a commercial algorithm, and doses are accumulated according to the DIR. The effect of motion synchronization on OAR dose was analyzed for five lung and five liver subjects by comparing planned versus synchrony‐accumulated dose. The motion was compensated by an average of 1.6 cm of jaw sway and by an average of 5.7% of leaf openings modified, indicating that most of the motion compensation was from jaw sway and not MLC changes. OAR dose deviations as large as 19 Gy were observed, and for all 10 cases, dose deviations greater than 7 Gy were observed. Target dose remained relatively constant (D95% within 3 Gy), confirming that motion‐synchronization achieved the goal of maintaining target dose. Dose deviations provided by the framework can be leveraged during the treatment planning process by identifying cases where OAR doses may change significantly from their planned values with respect to the critical constraints. The framework is specific to synchronized helical tomotherapy treatments, but the OAR dose deviations apply to any real‐time tracking technique that does not consider location changes of OARs.

Sub-micrometer morphology of human atherosclerotic plaque revealed by synchrotron radiation-based μCT—A comparison with histology

Histology is a long standing and well-established gold standard for pathological characterizations. In recent years however, synchrotron radiation-based micro-computed tomography (SRμCT) has become a tool for extending the imaging of two-dimensional thin sections into three-dimensional imaging of tissue blocks, enabling so-called virtual histology with arbitrary clipping planes, volumetric rendering and automatic segmentation. In this study, we present a thorough characterization of human carotid plaques after endarterectomy of patients with stroke or transient ischemic attack (TIA), investigating several different pathologic structures using both SRμCT and histology. Phase-contrast SRμCT was performed with two different magnifications (voxel sizes 6.5 μm and 0.65 μm, respectively), and histology was performed with multiple different stainings (Alpha-actin, Glycophorin A, von Kossa, Movat, CD68). The 0.65 μm high-resolution SRμCT was performed on selected areas with plaque typical relevant morphology, identified on the 6.5 μm low-resolution SRμCT. The tomography datasets were reconstructed with additional 3D volume rendering and compared to histology. In total, nine different regions with typical pathologic structures were identified and imaged with high-resolution SRμCT. The results show many characteristics typical for advanced atherosclerotic plaques, clinically relevant, namely ruptures with thrombosis, neo-vascularization, inflammatory infiltrates in shoulder regions, lipid rich necrotic cores (LRNC), thin fibrous cap, calcifications, lumen irregularities, and changes in vessel wall structures such as the internal elastic membrane. This method’s non-destructive nature renders details of micro-structures with an excellent visual likeness to histology, with the additional strength of multiplanar and 3D visualization and the possibility of multiple re-scans.

Probing the Role of Multi-scale Heterogeneity in Graphite Electrodes for Extreme Fast Charging.

Electrode-scale heterogeneity can combine with complex electrochemical interactions to impede lithium-ion battery performance, particularly during fast charging. This study investigates the influence of electrode heterogeneity at different scales on the lithium-ion battery electrochemical performance under operational extremes. We employ image-based mesoscale simulation in conjunction with a three-dimensional electrochemical model to predict performance variability in 14 graphite electrode X-ray computed tomography data sets. Our analysis reveals that the tortuous anisotropy stemming from the variable particle morphology has a dominating influence on the overall cell performance. Cells with platelet morphology achieve lower capacity, higher heat generation rates, and severe plating under extreme fast charge conditions. On the contrary, the heterogeneity due to the active material clustering alone has minimal impact. Our work suggests that manufacturing electrodes with more homogeneous and isotropic particle morphology will improve electrochemical performance and improve safety, enabling electromobility.

Efficient Quality Control with Mixed CT and CTA Datasets.

Deep learning promises the extraction of valuable information from traumatic brain injury (TBI) datasets and depends on efficient navigation when using large-scale mixed computed tomography (CT) datasets from clinical systems. To ensure a cleaner signal while training deep learning models, removal of computed tomography angiography (CTA) and scans with streaking artifacts is sensible. On massive datasets of heterogeneously sized scans, time-consuming manual quality assurance (QA) by visual inspection is still often necessary, despite the expectation of CTA annotation (artifact annotation is not expected). We propose an automatic QA approach for retrieving CT scans without artifacts by representing 3D scans as 2D axial slice montages and using a multi-headed convolutional neural network to detect CT vs CTA and artifact vs no artifact. We sampled 848 scans from a mixed CT dataset of TBI patients and performed 4-fold stratified cross-validation on 698 montages followed by an ablation experiment-150 stratified montages were withheld for external validation evaluation. Aggregate AUC for our main model was 0.978 for CT detection, 0.675 for artifact detection during cross-validation and 0.965 for CT detection, 0.698 for artifact detection on the external validation set, while the ablated model showed 0.946 for CT detection, 0.735 for artifact detection during cross-validation and 0.937 for CT detection, 0.708 for artifact detection on the external validation set. While our approach is successful for CT detection, artifact detection performance is potentially depressed due to the heterogeneity of present streaking artifacts and a suboptimal number of artifact scans in our training data.

Dislocations in minerals: Fast-diffusion pathways or trace-element traps?

Element mobility is a critical component in all geological processes and understanding the mechanisms responsible for element mobility in minerals is a fundamental requirement for many geochemical and geochronological applications. Volume diffusion of elements is a commonly assumed process. However, linear defects (dislocations) are an essential component of the high-temperature creep of minerals. These defects are commonly inferred to form fast-diffusion pathways along which trace elements can more rapidly migrate. In contrast, dislocations in minerals are also energetically favourable sites of trace element segregation, which counters the notion that they enhance bulk diffusion rates by a pipe diffusion mechanism. In this paper we characterize the trace-element composition of dislocations on twin boundaries in rutile by combining atom probe tomography with transmission electron microscopy. First, morphology and correlative microstructural data are used to demonstrate that the linear compositional features in the atom probe tomography dataset represent dislocations. Assessment of dislocation composition indicates that segregation is trace element specific. The data show that dislocations in rutile act as both, fast-diffusion pathway and trace-element traps which potentially leads to erroneous estimations of the composition.

Virtual Surgeries of Nasal Cavities Using a Coupled Lattice-Boltzmann–Level-Set Approach

Abstract Fluid mechanical properties of respiratory flow such as pressure loss, temperature distribution, or wall-shear stress characterize the physics of a nasal cavity. Simulations based on computational fluid dynamics (CFD) methods are able to deliver in-depth details on respiration. Integrating such tools into virtual surgery environments may support physicians in their decision-making process. In this study, a lattice-Boltzmann (LB) flow solver is coupled to a level-set (LS) method to modify the shape of a nasal cavity at simulation run time in a virtual surgery. The geometry of a presurgical nasal cavity obtained from computer tomography (CT) datasets is smoothly adapted toward a postsurgical geometry given by the surgeon using an interpolation approach based on a LS method. The influence of the modification on the respiratory flow is analyzed in silico. The methods are evaluated by simulating a virtual surgery of a stenotic pipe and juxtaposing the results to cases using static geometries and by comparing them to literature findings. The results for both the stenotic pipe and the nasal cavity are in perfect agreement with the expected outcomes. For the nasal cavity, a shape is found that reduces the nasal resistance by 25.3% for inspiration at a volumetric flow rate of V˙=250 ml/s. The heating capability is retained despite the geometry modification. The simulation results support the surgeon in evaluating a planned surgery and in finding an improved surgery for the patient.

WMR-DepthwiseNet: A Wavelet Multi-Resolution Depthwise Separable Convolutional Neural Network for COVID-19 Diagnosis.

Timely discovery of COVID-19 could aid in formulating a suitable treatment plan for disease mitigation and containment decisions. The widely used COVID-19 test necessitates a regular method and has a low sensitivity value. Computed tomography and chest X-ray are also other methods utilized by numerous studies for detecting COVID-19. In this article, we propose a CNN called depthwise separable convolution network with wavelet multiresolution analysis module (WMR-DepthwiseNet) that is robust to automatically learn details from both spatialwise and channelwise for COVID-19 identification with a limited radiograph dataset, which is critical due to the rapid growth of COVID-19. This model utilizes an effective strategy to prevent loss of spatial details, which is a prevalent issue in traditional convolutional neural network, and second, the depthwise separable connectivity framework ensures reusability of feature maps by directly connecting previous layer to all subsequent layers for extracting feature representations from few datasets. We evaluate the proposed model by utilizing a public domain dataset of COVID-19 confirmed case and other pneumonia illness. The proposed method achieves 98.63% accuracy, 98.46% sensitivity, 97.99% specificity, and 98.69% precision on chest X-ray dataset, whereas using the computed tomography dataset, the model achieves 96.83% accuracy, 97.78% sensitivity, 96.22% specificity, and 97.02% precision. According to the results of our experiments, our model achieves up-to-date accuracy with only a few training cases available, which is useful for COVID-19 screening. This latest paradigm is expected to contribute significantly in the battle against COVID-19 and other life-threatening diseases.

Stereogram of the Living Heart, Lung, and Adjacent Structures.

Innovations in invasive cardiovascular diagnostics and therapeutics, not only limited to transcatheter approaches but also involving surgical approaches, are based on a precise appreciation of the three-dimensional living heart anatomy. Rapid advancements in three-dimensional cardiovascular imaging technologies in the 21st century have supported such innovations through the periprocedural assessment of the clinical anatomy of the living heart. However, even if high-resolution volume-rendered images are reconstructed, they cannot provide appropriate depth perception when displayed and shared on a two-dimensional display, which is widely used in clinical settings. Currently, images reconstructed from clinical datasets can visualize fine details of the cardiovascular anatomy. Therefore, this is an optimal time for cardiologists and cardiac surgeons to revisit the classic technology, stereopsis, and obtain bonus information from carefully reconstructed clinical images. Using anaglyphs or cross/uncross-fusion of paired images, striking depth perception can be readily obtained without the need for expensive equipment. This conventional technique, when applied to high-resolution volume-rendered images, may help in obtaining appropriate diagnostics, choosing optimal therapeutics, securing procedural success, and preventing complications. Furthermore, it can be used for anatomical education. In this review, we demonstrate multiple stereoscopic images reconstructed from cardiac computed tomographic datasets and discuss their clinical and educational implications.

Generalized ComBat harmonization methods for radiomic features with multi-modal distributions and multiple batch effects

Radiomic features have a wide range of clinical applications, but variability due to image acquisition factors can affect their performance. The harmonization tool ComBat is a promising solution but is limited by inability to harmonize multimodal distributions, unknown imaging parameters, and multiple imaging parameters. In this study, we propose two methods for addressing these limitations. We propose a sequential method that allows for harmonization of radiomic features by multiple imaging parameters (Nested ComBat). We also employ a Gaussian Mixture Model (GMM)-based method (GMM ComBat) where scans are split into groupings based on the shape of the distribution used for harmonization as a batch effect and subsequent harmonization by a known imaging parameter. These two methods were evaluated on features extracted with CapTK and PyRadiomics from two public lung computed tomography datasets. We found that Nested ComBat exhibited similar performance to standard ComBat in reducing the percentage of features with statistically significant differences in distribution attributable to imaging parameters. GMM ComBat improved harmonization performance over standard ComBat (− 11%, − 10% for Lung3/CAPTK, Lung3/PyRadiomics harmonizing by kernel resolution). Features harmonized with a variant of the Nested method and the GMM split method demonstrated similar c-statistics and Kaplan–Meier curves when used in survival analyses.

Deep learning for full-feature X-ray microcomputed tomography segmentation of proton electron membrane fuel cells

This study demonstrates the benefit of convolutional neural networks to accurately classify the different materials of proton exchange membrane fuel cells using X-ray micro-computed tomography. Nineteen 2D orthoslices from a 3D tomography dataset were segmented with high quality and used to train a novel U-ResNet convolutional neural network (CNN) to segment the complete volume. The results were compared with a 3D manual segmentation performed under time constraints. The CNN segmented all phases with equal or greater accuracy in comparison to the manual segmentation. In particular, the CNN excelled in separating the carbon fibres and binder phase in the gas diffusion layer, which is usually completely avoided due to difficulty. Further, permeability calculations were performed on the binder void space for both segmentations, with the CNN displaying realistic results. Therefore, CNNs have been shown to be a viable and valuable method in segmenting such fuel cells with increased efficiency and accuracy.

Automated pancreas segmentation and volumetry using deep neural network on computed tomography

Pancreas segmentation is necessary for observing lesions, analyzing anatomical structures, and predicting patient prognosis. Therefore, various studies have designed segmentation models based on convolutional neural networks for pancreas segmentation. However, the deep learning approach is limited by a lack of data, and studies conducted on a large computed tomography dataset are scarce. Therefore, this study aims to perform deep-learning-based semantic segmentation on 1006 participants and evaluate the automatic segmentation performance of the pancreas via four individual three-dimensional segmentation networks. In this study, we performed internal validation with 1,006 patients and external validation using the cancer imaging archive pancreas dataset. We obtained mean precision, recall, and dice similarity coefficients of 0.869, 0.842, and 0.842, respectively, for internal validation via a relevant approach among the four deep learning networks. Using the external dataset, the deep learning network achieved mean precision, recall, and dice similarity coefficients of 0.779, 0.749, and 0.735, respectively. We expect that generalized deep-learning-based systems can assist clinical decisions by providing accurate pancreatic segmentation and quantitative information of the pancreas for abdominal computed tomography.

Predicting personalised and progressive adaptive dose escalation to gross tumour volume using knowledge-based planning models for inoperable advanced-stage non-small cell lung cancer patients treated with volumetric modulated arc therapy

Objectives. Increased radiation doses could improve local control and overall survival of lung cancer patients, however, this could be challenging without exceeding organs at risk (OAR) dose constraints, especially for patients with advanced-stage disease. Increasing OAR doses could reduce the therapeutic ratio and quality of life. It is therefore important to investigate methods to increase the dose to target volume without exceeding OAR dose constraints. Methods. Gross tumour volume (GTV) was contoured on synthetic computerised tomography (sCT) datasets produced using the Velocity adaptive radiotherapy software for eleven patients. The fractions where GTV volume decreased compared to that prior to radiotherapy (reference plan) were considered for personalised progressive dose escalation. The dose to the adapted GTV (GTVAdaptive) was increased until OAR doses were affected (as compared to the original clinical plan). Planning target volume (PTV) coverage was maintained for all plans. Doses were also escalated to the reference plan (GTVClinical) using the same method. Adapted, dose-escalated, plans were combined to estimate accumulated dose, D99 (dose to 99%) of GTVAdapted, PTV D99 and OAR doses and compared with those in the original clinical plans. Knowledge-based planning (KBP) model was developed to predict D99 of the adapted GTV with OAR doses and PTV coverage kept similar to the original clinical plans; prediction accuracy and model verification were performed using further data sets. Results. Compared to the original clinical plan, the dose to GTV was significantly increased without exceeding OAR doses. Adaptive dose-escalation increased the average D99 to GTVAdaptive by 15.1Gy and 8.7Gy compared to the clinical plans. The KBP models were verified and demonstrated prediction accuracy of 0.4% and 0.7% respectively. Conclusion. Progressive adaptive dose escalation can significantly increase the dose to GTV without increasing OAR doses or compromising the dose to microscopic disease. This may increase overall survival without increasing toxicities.

Deep learning-based simultaneous registration and unsupervised non-correspondence segmentation of medical images with pathologies

PurposeThe registration of medical images often suffers from missing correspondences due to inter-patient variations, pathologies and their progression leading to implausible deformations that cause misregistrations and might eliminate valuable information. Detecting non-corresponding regions simultaneously with the registration process helps generating better deformations and has been investigated thoroughly with classical iterative frameworks but rarely with deep learning-based methods.MethodsWe present the joint non-correspondence segmentation and image registration network (NCR-Net), a convolutional neural network (CNN) trained on a Mumford–Shah-like functional, transferring the classical approach to the field of deep learning. NCR-Net consists of one encoding and two decoding parts allowing the network to simultaneously generate diffeomorphic deformations and segment non-correspondences. The loss function is composed of a masked image distance measure and regularization of deformation field and segmentation output. Additionally, anatomical labels are used for weak supervision of the registration task. No manual segmentations of non-correspondences are required.ResultsThe proposed network is evaluated on the publicly available LPBA40 dataset with artificially added stroke lesions and a longitudinal optical coherence tomography (OCT) dataset of patients with age-related macular degeneration. The LPBA40 data are used to quantitatively assess the segmentation performance of the network, and it is shown qualitatively that NCR-Net can be used for the unsupervised segmentation of pathologies in OCT images. Furthermore, NCR-Net is compared to a registration-only network and state-of-the-art registration algorithms showing that NCR-Net achieves competitive performance and superior robustness to non-correspondences.ConclusionNCR-Net, a CNN for simultaneous image registration and unsupervised non-correspondence segmentation, is presented. Experimental results show the network’s ability to segment non-correspondence regions in an unsupervised manner and its robust registration performance even in the presence of large pathologies.

A Novel and Practical Protocol for Three-Dimensional Assessment of Alveolar Cleft Grafting Procedures.

To evaluate the reproducibility and accuracy of a new, easy-to-use volumetric assessment of the alveolar cleft. Twelve cone-beam computed tomography (CBCT) datasets of patients with a unilateral cleft lip, alveolus, and palate were evaluated by two investigators. Residual alveolar cleft calcified volume one year after surgery was analyzed by using standardized landmarks to determine the borders of the cleft defect and semi-automatically segment the alveolar cleft defect. The Dice-coefficient between observers for the segmented preoperative alveolar cleft defect was 0.81. Average percentage of residual alveolar cleft calcified material was 66.7% one year postoperatively. This study demonstrates a reliable and practical semi-automatic three-dimensional volumetric assessment method for unilateral clefts using CBCT.

Autosegmentation of lung computed tomography datasets using deep learning U-Net architecture.

Current radiotherapy treatment techniques require a large amount of imaging data for treatment planning which demand significant clinician's time to segment target volume and organs at risk (OARs). In this study, we propose to use U-net-based architecture to segment OARs commonly encountered in lung cancer radiotherapy. Four U-Net OAR models were generated and trained on 20 lung cancer patients' computed tomography (CT) datasets, with each trained for 100 epochs. The model was tested for each OAR, including the right lung, left lung, heart, and spinal cord. Dice similarity coefficient (DSC) and Hausdorff distance (HD) were used to assess the agreement between the predicted contour and ground truth. The highest of the average DSC among the test patients for the left lung and the right lung was 0.96 ± 0.03 and 0.94 ± 0.06, respectively, and 0.88 ± 0.04 for heart, and 0.76 ± 0.07 for the spinal cord. The HD for these corresponding DSCs was 3.51 ± 0.85, 4.06 ± 1.12, 4.09 ± 0.85, and 2.76 ± 0.52 mm for left lung, right lung, heart, and spinal cord, respectively. The autosegmented regions predicted by right and left lung models matched well with the manual contours. However, in a few cases, the heart model struggled to outline the boundary precisely. The spinal cord model had the lowest DSC, which may be due to its small size. This is an ongoing study aimed to assist radiation oncologists in segmenting the OARs with minimal effort.