3D CT Images Research Articles

Deformable lung 4DCT image registration via landmark-driven cycle network.

An automated, accurate, and efficient lung four-dimensional computed tomography (4DCT) image registration method is clinically important to quantify respiratory motion for optimal motion management. The purpose of this work is to develop a weakly supervised deep learning method for 4DCT lung deformable image registration (DIR). The landmark-driven cycle network is proposed as a deep learning platform that performs DIR of individual phase datasets in a simulation 4DCT. This proposed network comprises a generator and a discriminator. The generator accepts moving and target CTs as input and outputs the deformation vector fields (DVFs) to match the two CTs. It is optimized during both forward and backward paths to enhance the bi-directionality of DVF generation. Further, the landmarks are used to weakly supervise the generator network. Landmark-driven loss is used to guide the generator's training. The discriminator then judges the realism of the deformed CT to provide extra DVF regularization. We performed four-fold cross-validation on 10 4DCT datasets from the public DIR-Lab dataset and a hold-out test on our clinic dataset, which included 50 4DCT datasets. The DIR-Lab dataset was used to evaluate the performance of the proposed method against other methods in the literature by calculating the DIR-Lab Target Registration Error (TRE). The proposed method outperformed other deep learning-based methods on the DIR-Lab datasets in terms of TRE. Bi-directional and landmark-driven loss were shown to be effective for obtaining high registration accuracy. The mean and standard deviation of TRE for the DIR-Lab datasets was 1.20±0.72mm and the mean absolute error (MAE) and structural similarity index (SSIM) for our datasets were 32.1±11.6 HU and 0.979±0.011, respectively. The landmark-driven cycle network has been validated and tested for automatic deformable image registration of patients' lung 4DCTs with results comparable to or better than competing methods.

Estimation of missing mandible from the skull using postmortem CT images

Skeletal remains often have missing mandibles owing to the environment in which the corpse was placed or damage caused by an animal. Loss of the mandible reduces the accuracy of skull identification. Although several studies have validated methods for estimating mandibular morphology from conventional anthropological measurements using skull specimens, there are no reports using three-dimensional computed tomography (3DCT) images. Here, we examined methods for establishing the mandibular morphology from the remaining skulls using postmortem computed tomography (PMCT) images. We used PMCT images from 200 Japanese subjects as samples. After verifying the morphological correlation between the cranium and mandible, we created and validated estimation models using multiple regression analysis (stepwise method) for seven sites that were necessary for understanding the morphology of the mandible. Among the regression models, the estimated model for bicondylar breadth had the highest coefficient of determination (adjR2 = 0.53). We verified the accuracy of the model on a sample independent from the specimen used to create the estimation model and found that the formulated model of bicondylar breadth had good estimation accuracy, with a high correlation coefficient between the measured and predicted values of 0.82 and a mean absolute error of 3.582 mm, indicating that the model had good estimation accuracy. Here, we established a novel method for estimating the missing mandibular morphology based on PMCT data from the Japanese population. Our estimation model can help determine the missing mandibular morphology in a cranium remnant.

Missing common Hepatic Artery and variant Right Hepatic Artery - An incidental finding on CT imaging

During abdominal surgery, the blood supply to the liver and other abdominal organs plays a vital role. The knowledge regarding hepatic artery variations is crucial for various surgical and radiological procedures concerning the liver and dutiful to be reported. We aim to report a case of variant origin of the right hepatic artery to contribute to the existing knowledge pool to improve surgical safety. The present case describes a variant hepatic vascularization encountered on 3D volume-rendered CT imaging of a 60-year-old male admitted to the hospital emergency for recurrent abdominal pain. The common hepatic artery was absent. The right hepatic artery was observed to arise from the superior mesenteric artery, while the left hepatic artery arose directly from the coeliac artery. Awareness of such abnormalities is critical to minimize morbidity and prevent operational complications in hepatic surgery or liver transplantation. KEY WORDS: hepatic artery, CT scan, variations, anatomy, superior mesenteric artery, liver.

Registration between 2D and 3D Ultrasound Images to Track Liver Blood Vessel Movement.

For the accurate positioning of surgical tools, conventional intraoperative navigation systems have been developed to recognize the relationship between target positions and the tools. However, since an internal organ is deformed during the operation, registration between realtime two-dimensional (2D) ultrasound images and three-dimensional (3D) CT or MRI images is not always effective. Therefore, this study developed image registration between 2D and 3D ultrasound images considering deformation for tracking target vessel movement in the liver. 3D ultrasound image was obtained in advance with 3D coordinates, including the target vessel. Then real-time 2D images and ultrasound probe position were simultaneously acquired using a 3D position sensor. We applied multiple image resolution registration, where rapid and fine optimizations can be expected at higher and lower levels, respectively. Meanwhile, the gradient descent method was adopted for the optimization, which determines the relative arrangements to obtain maximum similarity between 2D and 3D images. We experimentally established resolution level parameters using a phantom before applying it to track liver blood vessel movements in a normal healthy subject. Comparing the 2D images and the registered images, although the approach has some limitations in tracking large displacement, we confirmed that the cross-section of the target blood vessel was clearly visualized. This method has the potential for an ultrasound therapy targeting blood vessels under natural respiration conditions.

The hemodynamics of adrenal veins with four-dimensional computed tomography using quantitative time-density curve: a study based on aldosteronism patients

Present study quantitatively analyzed adrenal venous flow using four-dimensional computed tomography (4D CT). We reviewed 4D CT images of 55 patients [mean age, 52 years ± 11 (standard deviation); 23 females] who underwent adrenal venous sampling between August 2017 and February 2021. Time–density curves were referred for the adrenal venous enhancement. The clinical factors affecting hemodynamics were assessed using uni- and multivariate linear regression analyses. The right and left adrenal veins (RAV and LAV, respectively) were visualized in all cases. Mean peak enhancement values in RAV and LAV were 247 ± 67 and 292 ± 70 Hounsfield units (P < 0.01), and were reached at 44.43 ± 6.86 and 45.39 ± 7.53 s (P < 0.01), respectively. The body mass index (BMI), plasma renin activity and potassium were significant factors influencing the peak enhancement of RAV blood flow [standardized regression coefficients, − 0.327 (P = 0.017), − 0.346 (P = 0.013), 0.426 (P = 0.016), respectively]. A linear relationship between sex and the time-to-peak was observed for RAV [standardized regression coefficient, 0.348 (P = 0.046)]. RAV had a lower contrast effect than LAV and reached its peak faster. BMI, plasma renin activity, and potassium were associated with flow density in RAV. Sex independently influenced the time-to-peak.

Assessment of Scapholunate Instability on 4D CT Scans in Patients with Inconclusive Conventional Images.

Background Initial imaging work-up using radiography and CT arthrography sometimes can be insufficient to identify a scapholunate (SL) instability (SLI) in patients suspected of having SL ligament tears. Purpose To determine the diagnostic performance of four-dimensional (4D) CT in the identification of SLI and apply the findings to patients suspected of having SLI and with inconclusive findings on radiographs and CT arthrograms. Materials and Methods This prospective single-center study enrolled participants suspected of having SLI (recent trauma, dorsal pain, positive Watson test results, decreased grip strength) between March 2015 and March 2020. Participants with wrist fractures, substantial joint stiffness, or history of wrist surgery were excluded. Each participant underwent radiography, CT arthrography, and 4D CT on the same day. Participants were divided into three groups: those with no SLI, those with SLI, and those with inconclusive results. SL gap and radioscaphoid and lunocapitate angle were measured using semiautomatic quantitative analysis of 4D CT images by two independent readers. Receiver operating characteristic curves were used to evaluate the diagnostic performance of 4D CT. Thresholds were determined with the Youden index and were applied to the inconclusive group. Results Of the 150 included participants (mean age, 41 years ± 14 [SD]; 102 male, 48 female), there were 63 with no SLI, 48 with SLI, and 39 with inconclusive results. The maximum value and range of SL gap measurements on 4D CT scans showed high sensitivity (83% [40 of 48] and 90% [43 of 48], respectively) and high specificity (95% [59 of 62] and 81% [50 of 62], respectively) in the identification of SLI. At least one of these parameters was abnormal on 4D CT scans in 17 of 39 (44%) participants in the inconclusive group, and 10 of 17 (59%) participants had confirmed SLI. In the 22 participants in the inconclusive group with no indication of SLI at 4D CT, follow-up showed no evidence of SLI in 10 (45%) and enabled confirmation of SLI via arthroscopy in three (14%). Conclusion Scapholunate gap measurements on kinematic 4D CT scans enabled correct identification of SLI in 59% of participants with inconclusive results on conventional images. ClinicalTrials.gov registration no. NCT02401568 © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Demehri and Ibad in this issue.

Prediction of dispersion and attenuation on elastic wave velocities in partially saturated rock based on the fluid distribution obtained from three-dimensional (3D) micro-CT images

Elastic wave attenuation in partially saturated porous rock is primarily due to wave-induced fluid flow, which arises from the contrast in compressibility between air and water and is influenced by the water distribution within the rock. We propose a method for constructing a numerical model that predicts mesoscopic dispersion and attenuation. Initially, we use fluid distribution data sourced from 3D X-ray Computed Tomography images to construct the numerical model, utilizing Biot’s poroelastic equations as the governing equations. Subsequently, we implement the finite element method to derive solutions for the numerical model. Our focus is centered on two key challenges: 1) reducing memory cost, and 2) efficiently handling element intersection during the meshing process. The solutions illustrate the evolution of fluid pressure distribution and the frequency-dependent advancement of the elastic moduli, coupled with their corresponding attenuation. Ultimately, we compare these numerical predictions with previously published experimental data from a study on partially saturated Indiana limestone. The considerable agreement between our numerical results and the experimental data confirms the validity of our method, which crucially incorporates the actual fluid distribution (captured from 3D CT images) as a vital input.

4D-CT deformable image registration using unsupervised recursive cascaded full-resolution residual networks.

A novel recursive cascaded full-resolution residual network (RCFRR-Net) for abdominal four-dimensional computed tomography (4D-CT) image registration was proposed. The entire network was end-to-end and trained in the unsupervised approach, which meant that the deformation vector field, which presented the ground truth, was not needed during training. The network was designed by cascading three full-resolution residual subnetworks with different architectures. The training loss consisted of the image similarity loss and the deformation vector field regularization loss, which were calculated based on the final warped image and the fixed image, allowing all cascades to be trained jointly and perform the progressive registration cooperatively. Extensive network testing was conducted using diverse datasets, including an internal 4D-CT dataset, a public DIRLAB 4D-CT dataset, and a 4D cone-beam CT (4D-CBCT) dataset. Compared with the iteration-based demon method and two deep learning-based methods (VoxelMorph and recursive cascaded network), the RCFRR-Net achieved consistent and significant gains, which demonstrated that the proposed method had superior performance and generalization capability in medical image registration. The proposed RCFRR-Net was a promising tool for various clinical applications.

Mechanical ventilation decreases tidal volume heterogeneity but increases heterogeneity in end-expiratory volumes.

How the heterogeneous distribution of lung volumes changes in response to different mechanical ventilation (MV) strategies is unclear. Using our well-developed four-dimensional computed tomography (4DCT) high-resolution imaging technique, we aimed to assess the effect of different MV strategies on the distribution and heterogeneity of regional lung volumes. Healthy adult female BALB/c mice received either 2 h of "injurious" MV [n = 6, mechanical ventilation at high PIP with zero PEEP (HPZP)] with a peak inspiratory pressure (PIP) of 20 cmH2O and zero positive end-expiratory pressure (PEEP), or 2 h of "protective" MV [n = 8, mechanical ventilation at low PIP with PEEP (LPP)] with PIP = 12 cmH2O and PEEP = 2 cmH2O. 4DCT images were obtained at baseline (0 h) and after 2 h of MV. Tidal volume (Vt) and end-expiratory lung volume (EEV) were measured throughout the whole lung on a voxel-by-voxel basis. Heterogeneity of ventilation was determined by the coefficient of variation (COV) of Vt and EEV. Our data showed that MV had minimal impact on global Vt but decreased EEV in the HPZP group (P < 0.05). Both ventilation modes decreased the COV of Vt (39.4% for HPZP and 9.7% for LPP) but increased the COV in EEV (36.4% for HPZP and 29.2% for LPP). This was consistent with the redistribution index, which was significantly higher in the HVZP group than in the LPP group (P < 0.001). We concluded that regional assessment of the change in EEV showed different patterns in progression between LPP and HPZP strategies. Both ventilation strategies decreased heterogeneity in Vt after 2 h of MV but increased heterogeneity in EEV. Further work is required to determine the link between these effects and ventilator-induced lung injury.NEW & NOTEWORTHY Tidal volume heterogeneity decreases over time in response to mechanical ventilation, in contrast to end-expiratory volume heterogeneity which increases.

Diaphragm displacement estimation based on subspace mapping from markerless abdominal surface measurement

Respiratory induced organ and tumor motion has great influence in radiation therapy. In order to estimate the inner structures’ movement by external observation, a novel two-step subspace mapping (TSSM) algorithm is proposed to estimate the diaphragm displacement from markerless abdominal surface measurement. We first incorporate 3D image segmentation to measure the displacement of the inner diaphragm and the external abdominal surface on 4D CT images. TSSM constructs eigenspaces by PCA and projects the two organs’ displacement data into their corresponding subspaces. Then, TSSM makes a mapping between the subspaces by a Ridge regression optimization to effectively characterize their correlation. Based on the trained correlation model, the diaphragm displacement can be estimated by the abdominal surface without markers. TSSM is further extended to kernel TSSM (kTSSM) to investigate the non-linear correlation. Experiments show the proposed method can accurately estimate the displacement of the internal diaphragm by the external abdominal surface.

Comparison of Orbital Reconstructive Effect between Customized Orbital Implants Using Three-Dimensional Printed Templates and Conventional Manual-Bending Implants in Blowout Fracture Surgery

The aim of the present study was to determine the orbital reconstructive effect of customized orbital implants using three-dimensional (3D) printed templates compared with conventional manual-bending implants using computed tomography (CT)-based orbital volume measurements. This retrospective study reviewed the medical records and 3D-CT images of 90 patients who underwent medial, inferior, or inferomedial orbital wall reconstruction. The selected patients were categorized into two groups: (1) the 3D group that underwent surgery using 3D-printed customized orbital implant templates and (2) the manual group that received a conventional manual technique to mold the implant. The volume discrepancy (VD) was obtained by subtracting the volume of the contralateral unaffected eye from that of the injured eye. Of the 90 patients, 33 and 57 were divided into the 3D and manual groups, respectively. The volumes on the contralateral unaffected side and on the pre- and postoperative injured sides were 22.5 ± 2.9, 23.7 ± 3.0, and 22.3 ± 2.8 cm3 (mean ± SD), respectively, in the 3D group, and 21.5 ± 2.5, 22.7 ± 2.8, and 21.2 ± 2.7 cm3 in the manual group. The postoperative VD did not differ between the 3D (–0.2 ± 0.3 cm3) and manual (–0.3 ± 0.9 cm3) groups (p = 0.794). The volume on the postoperative injured side did not differ significantly from that on the contralateral unaffected side in the 3D group, but these did differ significantly in the manual group. Postoperative VD also increased with the preoperative VD in the manual group (Pearson correlation coefficient = 0.548, p = 0.001), whereas there was no such association in the 3D group. The orbital volume restoration effect had superior surgical outcomes for large fractures using the customized orbital implant with 3D-printed templates compared with manual-bending implants.

The Utility of 4D-CT Imaging in Primary Hyperparathyroidism Management in a Low-Volume Center.

Background: Ultrasonography (US) and the 99mTc-sestamibi parathyroid scan (SPS) may have suboptimal accuracy when detecting the localization of enlarged parathyroid gland(s) (PTG). Therefore, the more accurate four-dimensional computed tomography scan (4D-CT) has been employed for PTG imaging. Currently, there is a paucity of data evaluating the utility of 4D-CT in low caseload settings. Aim and Objectives: To evaluate the impact of PTG imaging, using 4D-CT in conjunction with its intraoperatively displayed results, on the outcomes of surgical PTX. Materials and Methods: A single-center retrospective analysis of surgically treated patients with pHPT from 01/2010 to 01/2021 was conducted. An evaluation of the impact of the preoperative imaging modalities on the results of surgical treatment was carried out. Results: During the study period, 290 PTX were performed; 45 cases were excluded due to surgery for secondary, tertiary or recurrent HPT, or due to the use of alternative imaging techniques. The remaining 245 patients were included in the study. US was carried out for PTG imaging in 236 (96.3%), SPS in 93 (38.0%), and 4D-CT in 52 patients (21.2%). The use of 4D-CT was associated with a significantly higher rate of successful localization of enlarged PTG (49 cases, 94.2%) compared to US and SPS (74 cases, 31.4%, and 54 cases, 58.1%, respectively). We distinguished between three groups of patients based on preoperative imaging: (1) PTG lateralization via US or SPS in 106 (43.3%) cases; (2) precise localization of PTG via 4D-CT in 49 (20.0%) patients; and (3) in 90 cases (36.7%), PTG imaging failed to localize an enlarged gland. The group of 4D-CT localization had significantly shorter operative time, lower rate of simultaneous thyroid resections, as well as lower rate of removal of ≥2 PTG, compared to the other groups. The 4D-CT imaging was also associated with the lowest perioperative morbidity and with the lowest median PTH in the one month follow-up; however, compared to the other groups, these differences were statistically not significant. The implementation of 4D-CT (since 01/2018) was associated with a decrease in the need for redo surgery (from 11.5% to 7.3%) and significantly increased the annual case load of PTX at our institution (from 15.3 to 41.0) compared to the period before 4D-CT diagnostics. Conclusions: 4D-CT imaging enabled to precisely locate almost 95% of enlarged PTG in patients with pHPT. Accurate localization and intraoperatively displayed imaging results are useful guides for surgeons to make PTX a faster and safer procedure in a low-volume center.

Combined evaluation of blood flow and tissue perfusion in diabetic feet by intra-arterial dynamic 4DCT imaging

Critical limb ischemia is associated with high mortality and major amputations. Intra-arterial digital subtraction angiography (IADSA) has been the reference standard but has some shortcomings including the two-dimensional projection and the lack of tissue perfusion information. The aim of this exploratory study is to examine four-dimensional computed tomography (4DCT) angiography and perfusion imaging using low-volume intra-arterial contrast injections for an improved anatomic and hemodynamic assessment in patients with foot ulcers. Three patients underwent a low-volume (2 mL) intra-arterial contrast-enhanced 4DCT examination combined with a diagnostic IADSA. An automated assessment of blood flow and tissue perfusion from the 4DCT data was performed. Vascular structures and corresponding blood flows were successfully assessed and correlated well with the IADSA results. Perfusion values of the affected tissue were significantly higher compared to the unaffected tissue. The proposed 4DCT protocol combined with the minimal usage of contrast agent (2 mL) provides superior images compared to IADSA as three phases (arterial, perfusion, and venous) are captured. The obtained parameters could allow for an improved diagnosis of critical limb ischemia as both the proximal vasculature and the extent of the perfusion deficit in the microvasculature can be assessed.Relevance statementIntra-arterial 4DCT allows for assessing three phases (arterial, perfusion and venous) using minimal contrast (2 mL). This method could lead to an improved diagnosis of critical limb ischemia as both proximal vasculature and the extent of the perfusion deficit are assessed.Trial registrationISRCTN, ISRCTN95737449. Registered 14 March 2023—retrospectively registered, https://www.isrctn.com/ISRCTN95737449Key points• Three phases (arterial, perfusion, and venous) are obtained from 2 mL intra-arterial 4DCT.• The obtained hemodynamic parameters correlated well with the IADSA findings.• 4DCT surpassed IADSA in terms of assessment of venous blood flow and inflammatory hyperperfusion.• The assessment of tissue perfusion could lead to optimizing the revascularization strategy.Graphical

Universal Lesion Detection Utilising Cascading R-CNNs and a Novel Video Pretraining Method.

According to the WHO, approximately one in six individuals worldwide will develop some form of cancer in their lifetime. Therefore, accurate and early detection of lesions is crucial for improving the probability of successful treatment, reducing the need for more invasive treatments, and leading to higher rates of survival. In this work, we propose a novel R-CNN approach with pretraining and data augmentation for universal lesion detection. In particular, we incorporate an asymmetric 3D context fusion (A3D) for feature extraction from 2D CT images with Hybrid Task Cascade. By doing so, we supply the network with further spatial context, refining the mask prediction over several stages and making it easier to distinguish hard foregrounds from cluttered backgrounds. Moreover, we introduce a new video pretraining method for medical imaging by using consecutive frames from the YouTube VOS video segmentation dataset which improves our model's sensitivity by 0.8 percentage points at a false positive rate of one false positive per image. Finally, we apply data augmentation techniques and analyse their impact on the overall performance of our models at various false positive rates. Using our introduced approach, it is possible to increase the A3D baseline's sensitivity by 1.04 percentage points in mFROC.

Prediction of lymphoma response to CAR T cells by deep learning-based image analysis.

Clinical prognostic scoring systems have limited utility for predicting treatment outcomes in lymphomas. We therefore tested the feasibility of a deep-learning (DL)-based image analysis methodology on pre-treatment diagnostic computed tomography (dCT), low-dose CT (lCT), and 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) images and rule-based reasoning to predict treatment response to chimeric antigen receptor (CAR) T-cell therapy in B-cell lymphomas. Pre-treatment images of 770 lymph node lesions from 39 adult patients with B-cell lymphomas treated with CD19-directed CAR T-cells were analyzed. Transfer learning using a pre-trained neural network model, then retrained for a specific task, was used to predict lesion-level treatment responses from separate dCT, lCT, and FDG-PET images. Patient-level response analysis was performed by applying rule-based reasoning to lesion-level prediction results. Patient-level response prediction was also compared to prediction based on the international prognostic index (IPI) for diffuse large B-cell lymphoma. The average accuracy of lesion-level response prediction based on single whole dCT slice-based input was 0.82+0.05 with sensitivity 0.87+0.07, specificity 0.77+0.12, and AUC 0.91+0.03. Patient-level response prediction from dCT, using the "Majority 60%" rule, had accuracy 0.81, sensitivity 0.75, and specificity 0.88 using 12-month post-treatment patient response as the reference standard and outperformed response prediction based on IPI risk factors (accuracy 0.54, sensitivity 0.38, and specificity 0.61 (p = 0.046)). Prediction of treatment outcome in B-cell lymphomas from pre-treatment medical images using DL-based image analysis and rule-based reasoning is feasible. This approach can potentially provide clinically useful prognostic information for decision-making in advance of initiating CAR T-cell therapy.

The accuracy of an accelerometer-based portable navigation system for total hip arthroplasty using 3D CT measurement in the supine position: a prospective multicenter study

PurposeThis multicenter prospective study aimed to evaluate the accuracy of the cup alignment of an accelerometer-based portable navigation system (AP navigation) for a total hip arthroplasty (THA) in spine position. MethodsThis study prospectively enrolled 324 THAs using supine position in 9 hospitals. An AP navigation was used for cup fixation and the intraoperative navigation data (cup alignment, pelvic tilt, and pelvic rotation) were recorded. Computed tomography (CT) of the pelvis was performed 2 weeks post-operatively and cup alignment was measured on 3D-CT images. ResultThe accuracy (absolute difference in cup alignment between the intraoperative navigation record and postoperative 3D-CT measurements) was 3 ​± ​3° (mean ​± ​standard deviation) for cup inclination and 3 ​± ​3° for cup anteversion. The pelvis tilted anteriorly in 148 hips (46%) and posteriorly in 162 hips (50%), and did not tilt in 14 hips (4%). The pelvis rotated toward the operating side in 179 hips (55%), toward the contralateral side in 112 hips (35%), and did not rotate in 33 hips (10%). Multiple regression analysis showed that the patients’ characteristics (sex, body mass index, diagnosis, Crowe classification, laterality, intraoperative pelvic tilt, intraoperative pelvic rotation, and cup size) and the surgical factors (approach, preparation time, number of screws, and institution) did not affect the accuracy of the navigation system. ConclusionsThis is the first prospective multicenter study of this navigation system. The accuracy was not affected by patient characteristics, surgical factors, or substantial pelvic movement during cup fixation.

Evaluation of Virus Concentration Analysis in the Airway by CFD

Currently, Covid-19 is an epidemic all over the world. When virus directly adhere to mucous membrane of airway by breath, some humans maybe get inflammatory responses by viruses in the first stage of infection. The airway is composed of the nasal cavity, sinuses (Maxillary Sinus, Ethmoid Sinus, Frontal Sinus and Sphenoidal Sinus) and lungs. In the infection stage, the sinuses located in the nasal cavity tend to exhibit particularly high virus concentrations. Therefore, it is important to evaluate quantitatively the areas where viruses are likely to be adhered in the nasal cavity including sinuses. In this study, by CFD including concentration analysis the areas where viruses are likely to be adhered in the nasal cavity are predicted. As for the methods, the nasal cavity was made from 2D-CT image data by Itk-SNAP. For this computation in the nasal cavity continuity equation, Navier-Stokes equation and transport equation are used. And the transport of concentration was computed in the divided 4 parts of nasal cavity. As a result, it was found that the ratio of the concentration to the initial concentration in Ethmoid Sinus is approximately 0.6. It was found that Ethmoid Sinus is the areas where viruses are likely to be adhered and the areas can be predicted by computing the concentration.

TAGNet: A transformer-based axial guided network for bile duct segmentation

Automatic segmentation of intrahepatic bile ducts and common bile ducts plays an essential role in interventional surgery for cholangiocarcinoma, directly related to the success rate of the operation. However, the large shape and appearance variances make it challenging to segment bile ducts, especially for 3D CT images. In this study, we propose a transformer-based axial guided network, dubbed TAGNet, to automatically segment intrahepatic and common bile ducts by exploiting intra- and inter-slice context modeling. The pivot is to take advantage of CNN-transformer hybrid architecture to simultaneously explore local and global contextual information from multiple adjacent slices. Especially a novel slice-axial-attention transformer module is imposed at multi-scales concurrently to capture the intra- and inter-slice feature representations along each direction, boosting long-distance contextual modeling while limiting the computation cost. Moreover, a slice-guided consistency loss function is advanced to enforce anatomical prior consistency among adjacent slices in a semi-supervised manner, thus enhancing the spatial topology of bile duct segmentation. Extensive experimental results on an in-house bile ducts CT dataset demonstrate that our method is capable of achieving promising performance, which achieves at least a 4.5% improvement in Dice and a reduction of 1.5 in HD95 than other state-of-the-art methods, indicating its potential for automated intrahepatic and common bile duct segmentation. We have made our code publicly available via https://github.com/zephyrize/TAGNet.

A Novel Imaging Technique for Analyzing Condylar Movement During Mastication in Patients With Jaw Deformities: Four-Dimensional Computed Tomography.

This study aimed to reveal characteristic condylar movements in patients with jaw deformities. Thirty patients with jaw deformities before surgery were enrolled and instructed to chew a cookie during 4-dimensional computed tomography (4DCT). The distance between the most anterior and posterior positions of the bilateral condyles on 4DCT images was measured and compared among patients with different skeletal classes. Correlations between the condylar protrusion and cephalometric values were also determined. The distances of condylar protrusion during mastication were significantly greater in the skeletal class II group than in the skeletal class III group ( P =0.0002). Significant correlations were found between the distances of condylar protrusion during mastication and the sella-nasion-B point angle ( r =-0.442, P =0.015), A point-nasion-B point angle ( r =0.516, P =0.004), sella-nasion plane to ramus plane angle ( r =0.464, P =0.01), sella-nasion plane to occlusal plane angle ( r =0.367, P =0.047), and condylion-gonion length ( r =-0.366, P =0.048). Motion analysis with 4DCT images revealed that condylar movement in patients with retrognathism was larger than in patients with mandibular prognathism. Skeletal structure was therefore correlated with condylar movement during mastication.

Applying Artificial Intelligence to Predict Complications After Endovascular Aneurysm Repair.

Objective: Complications after Endovascular Aneurysm Repair (EVAR) can be fatal. Patient follow-up for surveillance imaging is becoming more challenging as fewer patients are seen, particularly after the first year. The aim of this study was to develop an artificial intelligence model to predict the complication probability of individual patients to better identify those needing more intensive post-operative surveillance. Methods: Pre-operative CTA 3D reconstruction images of AAA from 273 patients who underwent EVAR from 2011-2020 were collected. Of these, 48 patients had post-operative complications including endoleak, AAA rupture, graft limb occlusion, renal artery occlusion, and neck dilation. A deep convolutional neural network model (VascAI©) was developed which utilized pre-operative 3D CT images to predict risk of complications after EVAR. The model was built with TensorFlow software and run on the Google Colab Platform. An initial training subset of 40 randomly selected patients with complications and 189 without were used to train the AI model while the remaining 8 positive and 36 negative cases tested its performance and prediction accuracy. Data down-sampling was used to alleviate data imbalance and data augmentation methodology to further boost model performance. Results: Successful training was completed on the 229 cases in the training set and then applied to predict the complication probability of each individual in the held-out performance testing cases. The model provided a complication sensitivity of 100% and identified all the patients who later developed complications after EVAR. Of 36 patients without complications, 16 (44%) were falsely predicted to develop complications. The results therefore demonstrated excellent sensitivity for identifying patients who would benefit from more stringent surveillance and decrease the frequency of surveillance in 56% of patients unlike to develop complications. Conclusion: AI models can be developed to predict the risk of post-operative complications with high accuracy. Compared to existing methods, the model developed in this study did not require any expert-annotated data but only the AAA CTA images as inputs. This model can play an assistive role in identifying patients at high risk for post-EVAR complications and the need for greater compliance in surveillance.