3D CT Images Research Articles

A deep-learning-based scatter correction with water equivalent path length map for digital radiography.

We proposed a new deep learning (DL) model for accurate scatter correction in digital radiography. The proposed network featured a pixel-wise water equivalent path length (WEPL) map of subjects with diverse sizes and 3D inner structures. The proposed U-Net model comprises two concatenated modules: one for generating a WEPL map and the other for predicting scatter using the WEPL map as auxiliary information. First, 3D CT images were used as numerical phantoms for training and validation, generating observed and scattered images by Monte Carlo simulation, and WEPL maps using Siddon's algorithm. Then, we optimised the model without overfitting. Next, we validated the proposed model's performance by comparing it with other DL models. The proposed model obtained scatter-corrected images with a peak signal-to-noise ratio of 44.24 ± 2.89dB and a structural similarity index measure of 0.9987 ± 0.0004, which were higher than other DL models. Finally, scatter fractions (SFs) were compared with other DL models using an actual phantom to confirm practicality. Among DL models, the proposed model showed the smallest deviation from measured SF values. Furthermore, using an actual radiograph containing an acrylic object, the contrast-to-noise ratio (CNR) of the proposed model and the anti-scatter grid were compared. The CNR of the images corrected using the proposed model are 16% and 82% higher than those of the raw and grid-applied images, respectively. The advantage of the proposed method is that no actual radiography system is required for collecting training dataset, as the dataset is created from CT images using Monte Carlo simulation.

Automatic analysis of the scapholunate distance using 4DCT imaging: normal values in the healthy wrist

Early diagnosis of scapholunate ligament (SLL) injuries is crucial to prevent progression to debilitating osteoarthritis. Four-Dimensional Computed Tomography (4DCT) is a promising dynamic imaging modality for assessing such injuries. Capitalizing on the known correlation between SLL injuries and an increased scapholunate distance (SLD), this study aims to develop a fully automatic approach to evaluate the SLD continuously during wrist motion and to apply it to a dataset of healthy wrists to establish reference values. 50 healthy wrists were analysed in this study. All subjects performed radioulnar deviation (RUD), flexion-extension (FE), and clenching fist (CF) movements during 4DCT acquisition. A novel, automatic method was developed to continuously compute the SLD at five distinct locations within the scapholunate joint, encompassing a centre, volar, dorsal, proximal, and distal measurement. The developed algorithm successfully processed datasets from all subjects. Our results showed that the SLD remained below 2 mm and exhibited minimal changes (median ranges between 0.3 mm and 0.65 mm) during RUD and CF at all measured locations. During FE, the volar and dorsal SLD changed significantly, with median ranges of 0.90 and 1.27 mm, respectively. This study establishes a unique database of normal SLD values in healthy wrists during wrist motion. Our results indicate that, aside from RUD and CF, FE may also be important in assessing wrist kinematics. Given the labour-intensive and time-consuming nature of manual analysis of 4DCT images, the introduction of this automated algorithm enhances the clinical utility of 4DCT in diagnosing dynamic wrist injuries.

Microscopic analysis of mechanical anisotropy and damage evolution of 3D printed rock-like samples under uniaxial compressive tests

Three-dimensional printing (3DP) offers valuable insight into the characterization of natural rocks and the verification of theoretical models due to its high reproducibility and accurate replication of complex defects such as cracks and pores. In this study, 3DP gypsum samples with different printing directions were subjected to a series of uniaxial compression tests with in situ micro-computed tomography (micro-CT) scanning to quantitatively investigate their mechanical anisotropic properties and damage evolution characteristics. Based on the two-dimensional (2D) CT images obtained at different scanning steps, a novel void ratio variable was derived using the mean value and variance of CT intensity. Additionally, a constitutive model was formulated incorporating the proposed damage variable, utilizing the void ratio variable. The crack evolution and crack morphology of 3DP gypsum samples were obtained and analyzed using the 3D models reconstructed from the CT images. The results indicate that 3DP gypsum samples exhibit mechanical anisotropic characteristics similar to those found in naturally sedimentary rocks. The mechanical anisotropy is attributed to the bedding planes formed between adjacent layers and pillar-like structures along the printing direction formed by CaSO4·2H2O crystals of needle-like morphology. The mean gray intensity of the voids has a positive linear relationship with the threshold value, while the CT variance and void ratio have concave and convex relationships, respectively. The constitutive model can effectively match the stress–strain curves obtained from uniaxial compression experiments. This study provides comprehensive explanations of the failure modes and anisotropic mechanisms of 3DP gypsum samples, which is important for characterizing and understanding the failure mechanism and microstructural evolution of 3DP rocks when modeling natural rock behavior.

A prospective observational study on hemodynamic risk in cerebral aneurysm growth

Introduction: Hemodynamic factors are thought to play a major role in the development, growth, and rupture of cerebral aneurysms, but the hemodynamics related to enlargement remain unclear. We prospectively enrolled patients with unruptured cerebral aneurysms and performed Computational Fluid Dynamics (CFD) analysis using 3D CT images to investigate differences in hemodynamic environments between aneurysms that grew and those that did not grow over a 3-year period. Methods: This prospective observational study was approved by the National Hospital Organization Central Ethics Review Board. Hemodynamic distributions were examined using ANSYS-CFX software. Patient-specific arterial geometries and inflow velocities were obtained from 3D-CTA and carotid ultrasound Doppler examinations. The distributions of hemodynamic metrics of 17 enlarged and 70 unenlarged aneurysms was compared. In addition, the distributions of hemodynamic metrics of the neck, body, and dome of the aneurysm was also examined in 9 enlarged and 41 non-enlarged cases. A multivariate logistic analysis was performed to investigate the influence of the six hemodynamic metrics, i.e., the time-averaged wall shear stress (TAWSS), time-averaged WSS gradient (TAWSSG), oscillatory shear index (OSI), gradient oscillatory number (GON), transverse WSS (transWSS), and normalized transWSS (NtransWSS) after adjustment forknown risk factors for aneurysm growth, i.e., age, sex, hypertension, smoking history, and aneurysm location, size, and history of subarachnoid hemorrhage. Results: Regarding the distributions of hemodynamic metrics throughout the aneurysm, NtransWSS was significantly higher in the cases with growth than in the cases without growth (odds ratio 2.09, 95% CI 1.02-4.29). the distributions of hemodynamic metrics at each cerebral aneurysm growth site showed that NtransWSS at the dome was significantly higher in the growth cases than in the non-enlargement cases (odds ratio 4.14, 95% CI 1.02-16.80). Conclusion: The data suggest that multidirectional disturbance of wall shear stress is involved in cerebral aneurysm growth, and the disturbance at the dome of the aneurysm may be particularly strongly related. This study was supported by a Grant-in-Aid from the Japanese National Hospital Organization Multi-Center Clinical Research, AMED under grant no. JP15gm0810006h0301 and a JSPS KAKENHI Grant, no. 15K10323. This is the full abstract presented at the American Physiology Summit 2024 meeting and is only available in HTML format. There are no additional versions or additional content available for this abstract. Physiology was not involved in the peer review process.

1371: Impact of breathing signal-guided dose modulation on 4DCT image reconstruction

1371: Impact of breathing signal-guided dose modulation on 4DCT image reconstruction

Deep learning-aided 3D proxy-bridged region-growing framework for multi-organ segmentation

Accurate multi-organ segmentation in 3D CT images is imperative for enhancing computer-aided diagnosis and radiotherapy planning. However, current deep learning-based methods for 3D multi-organ segmentation face challenges such as the need for labor-intensive manual pixel-level annotations and high hardware resource demands, especially regarding GPU resources. To address these issues, we propose a 3D proxy-bridged region-growing framework specifically designed for the segmentation of the liver and spleen. Specifically, a key slice is selected from each 3D volume according to the corresponding intensity histogram. Subsequently, a deep learning model is employed to pinpoint the semantic central patch on this key slice, to calculate the growing seed. To counteract the impact of noise, segmentation of the liver and spleen is conducted on superpixel images created through proxy-bridging strategy. The segmentation process is then extended to adjacent slices by applying the same methodology iteratively, culminating in the comprehensive segmentation results. Experimental results demonstrate that the proposed framework accomplishes segmentation of the liver and spleen with an average Dice Similarity Coefficient of approximately 0.93 and a Jaccard Similarity Coefficient of around 0.88. These outcomes substantiate the framework's capability to achieve performance on par with that of deep learning methods, albeit requiring less guidance information and lower GPU resources.

Detection of scapholunate interosseous ligament injury using dynamic computed tomography-derived arthrokinematics: A prospective clinical trial

Scapholunate interosseous ligament injuries are a major cause of wrist instability and can be difficult to diagnose radiographically. To improve early diagnosis of scapholunate ligament injuries, we compared injury detection between bilateral routine clinical radiographs, static CT, and dynamic four-dimensional CT (4DCT) during wrist flexion-extension and radioulnar deviation. Participants with unilateral scapholunate ligament injuries were recruited to a prospective clinical trial investigating the diagnostic utility of 4DCT imaging for ligamentous wrist injury. Twenty-one participants underwent arthroscopic surgery to confirm scapholunate ligament injury. Arthrokinematics, defined as distributions of interosseous proximities across radioscaphoid and scapholunate articular surfaces at different positions within the motion cycle, were used as CT-derived biomarkers. Preoperative radiographs, static CT, and extrema of 4DCT were compared between uninjured and injured wrists using Wilcoxon signed rank or Kolmogorov-Smirnov tests. Median interosseous proximities at the scapholunate interval were significantly greater in the injured versus the uninjured wrists at static-neutral and maximum flexion, extension, radial deviation, and ulnar deviation. Mean cumulative distribution functions at the radioscaphoid joint were not significantly different between wrists but were significantly shifted at the scapholunate interval towards increased interosseous proximities in injured versus uninjured wrists in all positions. Median and cumulative distribution scapholunate proximities from static-neutral and 4DCT-derived extrema reflect injury status.

Hindfoot kinematics and kinetics - A combined in vivo and in silico analysis approach

BackgroundThe complex anatomical structure of the foot-ankle imposes challenges to accurately quantify detailed hindfoot kinematics and estimate musculoskeletal loading parameters. Most systems used to capture or estimate dynamic joint function oversimplify the anatomical structure by reducing its complexity. Research QuestionCan four dimensional computed tomography (4D CT) imaging in combination with an innovative foot manipulator capture in vivo hindfoot kinematics during a simulated stance phase of walking and can talocrural and subtalar articular joint mechanics be estimated based on a detailed in silico musculoskeletal foot-ankle model. MethodsA foot manipulator imposed plantar/dorsiflexion and inversion/eversion representing a healthy stance phase of gait in 12 healthy participants while simultaneously acquiring 4D CT images. Participant-specific 3D hindfoot rotations and translations were calculated based on bone-specific anatomical coordinate systems. Articular cartilage contact area and contact pressure of the talocrural and subtalar joints were estimated using an extended foot-ankle model updated with an elastic foundation contact model upon prescribing the participant-specific rotations measured in the 4D CT measurement. ResultsPlantar/dorsiflexion predominantly occurred at the talocrural joint (RoM 15.9±3.9°), while inversion/eversion (RoM 5.9±3.9°) occurred mostly at the subtalar joint, with the contact area being larger at the subtalar than at the talocrural joint. Contact pressure was evenly distributed between the talocrural and subtalar joint at the beginning of the simulated stance phase but was then redistributed from the talocrural to the subtalar joint with increasing dorsiflexion. SignificanceIn a clinical case study, the healthy participants were compared with four patients after surgically treaded intra-articular calcaneal fracture. The proposed workflow was able to detect small but meaningful differences in hindfoot kinematics and kinetics, indicative of remaining hindfoot pathomechanics that may influence the onset and progression of degenerative joint diseases.

CT respiratory motion synthesis using joint supervised and adversarial learning

Objective. Four-dimensional computed tomography (4DCT) imaging consists in reconstructing a CT acquisition into multiple phases to track internal organ and tumor motion. It is commonly used in radiotherapy treatment planning to establish planning target volumes. However, 4DCT increases protocol complexity, may not align with patient breathing during treatment, and lead to higher radiation delivery. Approach. In this study, we propose a deep synthesis method to generate pseudo respiratory CT phases from static images for motion-aware treatment planning. The model produces patient-specific deformation vector fields (DVFs) by conditioning synthesis on external patient surface-based estimation, mimicking respiratory monitoring devices. A key methodological contribution is to encourage DVF realism through supervised DVF training while using an adversarial term jointly not only on the warped image but also on the magnitude of the DVF itself. This way, we avoid excessive smoothness typically obtained through deep unsupervised learning, and encourage correlations with the respiratory amplitude. Main results. Performance is evaluated using real 4DCT acquisitions with smaller tumor volumes than previously reported. Results demonstrate for the first time that the generated pseudo-respiratory CT phases can capture organ and tumor motion with similar accuracy to repeated 4DCT scans of the same patient. Mean inter-scans tumor center-of-mass distances and Dice similarity coefficients were 1.97 mm and 0.63, respectively, for real 4DCT phases and 2.35 mm and 0.71 for synthetic phases, and compares favorably to a state-of-the-art technique (RMSim). Significance. This study presents a deep image synthesis method that addresses the limitations of conventional 4DCT by generating pseudo-respiratory CT phases from static images. Although further studies are needed to assess the dosimetric impact of the proposed method, this approach has the potential to reduce radiation exposure in radiotherapy treatment planning while maintaining accurate motion representation. Our training and testing code can be found at https://github.com/cyiheng/Dynagan.

Prospective Trial of Functional Lung Avoidance Radiation Therapy for Lung Cancer: Quality of Life Report

PurposeA novel form of lung function imaging has been developed that uses 4DCT data to generate lung ventilation images (4DCT-ventilation). Functional avoidance uses 4DCT-ventilation to reduce doses to functional lung with the aim of reducing pulmonary side-effects. A 4DCT-ventilation functional avoidance, phase II, multi-center clinical trial was completed. The purpose of this work is to quantify patient reported outcomes (PRO) changes for patients treated with functional avoidance and to determine which metrics are predictive of PRO changes. Materials and MethodsPatients with locally advanced lung cancer receiving curative intent radiotherapy were accrued. Each patient had a 4DCT-ventilation image generated using 4DCT data and image processing. PRO instruments included the Functional Assessment of Cancer Therapy-Lung (FACT-L) questionnaire, administered pre-treatment, 3, 6, and 12 months post-treatment. FACT-TOI (Trial Outcome Index) and the FACT-LCS (Lung Cancer Subscale) percentage of clinically meaningful declines (CMD) were determined. A linear mixed-effects model was used to determine which patient, clinical, dose, and dose-function metrics were predictive of PRO decline. Results59 patients completed baseline PRO surveys. 83% of patients had non-small-cell lung cancer, with 75% having stage III disease. The median dose was 60 Gy in 30 fractions. CMD FACT-TOI decline was 46.3%, 38.5%, and 26.8%, at 3, 6, and 12 months, respectively. CMD FACT-LCS decline was 33.3%, 33.3%, and 29.3%, at 3, 6, and 12 months, respectively. While an increase in most dose and dose-function parameters was associated with a modest decline in PROs, none of the results were significant (all p>0.053). ConclusionThe current work provides an innovative combination of functional avoidance and PROs and is the first report of PROs for patients treated with prospective 4DCT-ventilation functional avoidance. Approximately 30% of patients had clinically significant decline in PROs at 12 months. The study provides additional data on outcomes with 4DCT-ventilation functional avoidance.

S-Net: an S-shaped network for nodule detection in 3D CT images

Objective. Accurate and automatic detection of pulmonary nodules is critical for early lung cancer diagnosis, and promising progress has been achieved in developing effective deep models for nodule detection. However, most existing nodule detection methods merely focus on integrating elaborately designed feature extraction modules into the backbone of the detection network to extract rich nodule features while ignore disadvantages of the structure of detection network itself. This study aims to address these disadvantages and develop a deep learning-based algorithm for pulmonary nodule detection to improve the accuracy of early lung cancer diagnosis. Approach. In this paper, an S-shaped network called S-Net is developed with the U-shaped network as backbone, where an information fusion branch is used to propagate lower-level details and positional information critical for nodule detection to higher-level feature maps, head shared scale adaptive detection strategy is utilized to capture information from different scales for better detecting nodules with different shapes and sizes and the feature decoupling detection head is used to allow the classification and regression branches to focus on the information required for their respective tasks. A hybrid loss function is utilized to fully exploit the interplay between the classification and regression branches. Main results. The proposed S-Net network with ResSENet and other three U-shaped backbones from SANet, OSAF-YOLOv3 and MSANet (R+SC+ECA) models achieve average CPM scores of 0.914, 0.915, 0.917 and 0.923 on the LUNA16 dataset, which are significantly higher than those achieved with other existing state-of-the-art models. Significance. The experimental results demonstrate that our proposed method effectively improves nodule detection performance, which implies potential applications of the proposed method in clinical practice.

Clubfoot Correction with Ponseti Technique: Three-Dimensional Alignment Analysis and Residual Adult Deformity Effects on Patient-Reported Outcomes

Introduction/Purpose: Few studies have assessed the long-term outcomes of the Ponseti technique and none have utilized 3- dimensional weightbearing analysis. The goal of this study was to understand how potential residual 3D deformities and abnormalities influence patient reported outcomes (PROs). This was accomplished by assessing anatomical foot and ankle alignment in adult clubfoot patients treated with the Ponseti method using 3D weightbearing CT (WBCT) imaging and then correlating residual foot and ankle malalignment with PROs. Methods: There were 37 consecutive patients (57 feet) included and 14 volunteers healthy controls (28 feet) included in this study. Every participant was evaluated using a WBCT (HiRise©) in a bipedal standing position. From these scans Cavus, Adductus, and Varus components were evaluated using multiple 3D measurements calculated using the semi-automatic segmentation software Bonelogic®. Specific Cavus related measurements included sagittal talus-first metatarsal angle and the calcaneal inclination angle. Varus related measurements included talocalcaneal angle in both the sagittal and axial planes as well as the hindfoot moment arm and the hindfoot alignment angle. Adductus deformity was evaluated using talonavicular coverage angle. These measurements were then correlated with patient reported outcome surveys, which included Visual Acuity Scale for pain, PROMIS general health, PROMIS physical function, PROMIS pain interference, pain catastrophic scale, and European foot and ankle society score. Results: There was no significant overall residual 3D-deformity observed in clubfoot patients when compared to controls, with similar FAO measurements observed between the groups, clubfoot=2.63% (95%CI=1.41%-3.85%) and control=3.2% (CI=1.6%- 4.8%,P=0.58). The sagittal talus-first-metatarsal in the clubfoot-patients had a mean-value of -0.12° compared to the controls, −5.2°. Clubfoot patients also had a decreased calcaneal-inclination-angle relative to the controls, 13.01° and 21.5° respectively. Talocalcaneal-angle for clubfoot patients in both the sagittal-plane,44.28°, and axial-plane, 17.74°, were reduced compared to the controls, 57.51° and 25.78°. Talonavicular-coverage-angle in the clubfoot-group (18.63°) was less than the controls (29.19°). Talus- first-metatarsal-angle in the sagittal-plane was significantly correlated with VAS-scores (RSquare=0.19,P=0.0118) and the EFAS- Score (RSquare=0.27,P=0.0025). Talocalcaneal-angle in the sagittal plane was also significantly correlated with the PROMIS-Pain- Interference-score (P=0.038) and PROMIS-Physical-Function-score (RSquare=0.32,P=0.0007). Conclusion: The Ponseti technique is an effective nonsurgical treatment for the overall three-dimensional foot and ankle alignment of Clubfoot. While mild, but statistically significant residual Varus and Adductus deformities were observed in adult clubfoot patients, the overall 3D alignment (FAO) was found to be similar between clubfoot patients and controls. These findings highlight the efficacy of the Ponseti technique and potentially explain the overall good PROs. The results of this study could potentially provide insight into treatment targets that may be applied to help optimize patient outcomes when treating children with Clubfoot in the future. The Ponseti technique is an effective nonsurgical treatment for Clubfoot's overall three-dimensional foot and ankle alignment. While mild, but statistically significant residual Varus and Adductus deformities were observed in adult clubfoot patients, the overall 3D alignment (FAO) was found to be similar between clubfoot patients and controls. These findings highlight the efficacy of the Ponseti technique and potentially explain the overall good PROs. The results of this study could potentially provide insight into treatment targets that may be applied to help optimize patient outcomes when treating children with Clubfoot in the future.

A model that predicts a real-time tumour surface using intra-treatment skin surface and end-of-expiration and end-of-inhalation planning CT images.

To develop a mapping model between skin surface motion and internal tumour motion and deformation using end-of-exhalation (EOE) and end-of-inhalation (EOI) 3D CT images for tracking lung tumours during respiration. Before treatment, skin and tumour surfaces were segmented and reconstructed from the EOE and the EOI 3D CT images. A non-rigid registration algorithm was used to register the EOE skin and tumour surfaces to the EOI, resulting in a displacement vector field that was then used to construct a mapping model. During treatment, the EOE skin surface was registered to the real-time, yielding a real-time skin surface displacement vector field. Using the mapping model generated, the input of a real-time skin surface can be used to calculate the real-time tumour surface. The proposed method was validated with and without simulated noise on 4D CT images from 15 patients at Léon Bérard Cancer Center and the 4D-lung dataset. The average centre position error, dice similarity coefficient (DSC), 95%-Hausdorff distance and mean distance to agreement of the tumour surfaces were 1.29 mm, 0.924, 2.76 mm, and 1.13 mm without simulated noise, respectively. With simulated noise, these values were 1.33 mm, 0.920, 2.79 mm, and 1.15 mm, respectively. A patient-specific model was proposed and validated that was constructed using only EOE and EOI 3D CT images and real-time skin surface images to predict internal tumour motion and deformation during respiratory motion. The proposed method achieves comparable accuracy to state-of-the-art methods with fewer pre-treatment planning CT images, which holds potential for application in precise image-guided radiation therapy.

The Optimal Patient Position on the Surgical Table for Lateral Lumbar Interbody Fusion in Adult Spinal Deformity Using Three-Dimensional Computed Tomography: A Retrospective Study.

Lateral lumbar interbody fusion (LLIF) techniques have been extensively used in adult spinal deformity surgery. Preoperative knowledge of the optimal position of the patient on the surgical table is essential for a safe procedure. Therefore, this study aims to determine the optimal angle for positioning the patient on the surgical table during LLIF using three-dimensional computed tomography (3DCT). Data from 59 patients (2 males, 57 females, mean age 66.3±8.6 years) with adult spinal deformities treated by performing corrective spinal surgery were included in this observational retrospective study. Simulated fluoroscopic images were obtained using 3DCT images rotated from the reference position with the spinous process of S1 as the midline to the position with the spinous process in the center of the bilateral pedicle of T12-L5. The rotation angle of each vertebra was measured and defined as the optimal rotation angle (ORA). The angle that bisected the angle between the maximum and minimum ORA was defined as the optimal mean angle of the maximum and minimum ORA (OMA) and considered the optimal angle for the patient's position on the surgical table, as this position could minimize the rotation angle of the surgical table during surgery. A multiple regression analysis was performed to predict OMA. Multiple regression analysis revealed the following equation: OMA=1.959+(0.238×lumbar coronal Cobb angle)+(-0.208×sagittal vertical axis). When the patient is placed on the surgical table by rotating them at the OMA, the rotation of the surgical table can be reduced, ensuring a safe and efficient surgical procedure.

Uncertainty-aware image classification on 3D CT lung

Early detection is crucial for lung cancer to prolong the patient’s survival. Existing model architectures used in such systems have shown promising results. However, they lack reliability and robustness in their predictions and the models are typically evaluated on a single dataset, making them overconfident when a new class is present. With the existence of uncertainty, uncertain images can be referred to medical experts for a second opinion. Thus, we propose an uncertainty-aware framework that includes three phases: data preprocessing and model selection and evaluation, uncertainty quantification (UQ), and uncertainty measurement and data referral for the classification of benign and malignant nodules using 3D CT images. To quantify the uncertainty, we employed three approaches; Monte Carlo Dropout (MCD), Deep Ensemble (DE), and Ensemble Monte Carlo Dropout (EMCD). We evaluated eight different deep learning models consisting of ResNet, DenseNet, and the Inception network family, all of which achieved average F1 scores above 0.832, and the highest average value of 0.845 was obtained using InceptionResNetV2. Furthermore, incorporating the UQ demonstrated significant improvement in the overall model performance. Upon evaluation of the uncertainty estimate, MCD outperforms the other UQ models except for the metric, URecall, where DE and EMCD excel, implying that they are better at identifying incorrect predictions with higher uncertainty levels, which is vital in the medical field. Finally, we show that using a threshold for data referral can greatly improve the performance further, increasing the accuracy up to 0.959.

End-to-end multimodal 3D imaging and machine learning workflow for non-destructive phenotyping of grapevine trunk internal structure

Quantifying healthy and degraded inner tissues in plants is of great interest in agronomy, for example, to assess plant health and quality and monitor physiological traits or diseases. However, detecting functional and degraded plant tissues in-vivo without harming the plant is extremely challenging. New solutions are needed in ligneous and perennial species, for which the sustainability of plantations is crucial. To tackle this challenge, we developed a novel approach based on multimodal 3D imaging and artificial intelligence-based image processing that allowed a non-destructive diagnosis of inner tissues in living plants. The method was successfully applied to the grapevine (Vitis vinifera L.). Vineyard’s sustainability is threatened by trunk diseases, while the sanitary status of vines cannot be ascertained without injuring the plants. By combining MRI and X-ray CT 3D imaging with an automatic voxel classification, we could discriminate intact, degraded, and white rot tissues with a mean global accuracy of over 91%. Each imaging modality contribution to tissue detection was evaluated, and we identified quantitative structural and physiological markers characterizing wood degradation steps. The combined study of inner tissue distribution versus external foliar symptom history demonstrated that white rot and intact tissue contents are key-measurements in evaluating vines’ sanitary status. We finally proposed a model for an accurate trunk disease diagnosis in grapevine. This work opens new routes for precision agriculture and in-situ monitoring of tissue quality and plant health across plant species.

High three-dimensional accuracy of component placement and lower limb alignment using a robotic arm-assisted system and gap-balancing instrument in total knee arthroplasty.

It was hypothesized that robotic arm-assisted total knee arthroplasty (RA-TKA) using additionally a gap-balancing instrumentation will show high accuracy in executing the planning in femoral and tibial component placement throughout the range of knee motion (ROM) during TKA surgery. Prospectively collected data were analysed for patients undergoing RA-TKA. A cruciate retaining cemented design was implanted using the MAKO® robotic system. Lower limb alignment at 0°, 30°, 45°, 60° and 90° of flexion was recorded at the beginning of surgery and finally after implantation of the components. A ligament tensioner was inserted after tibial precut to measure the extension and flexion gap, and final component placement was planned based on 3D CT images. Femoral and tibial component placement was measured in all three planes. A total of 104 patients were included (mean age 69.4 ± 9 years; 44 male, 60 female). The difference in component placement after planning and final implantation showed less valgus of 0.7° ± 1.4° (p < 0.001), less external rotation of 0.6° ± 1.9° (p = 0.001) and less flexion of 0.9° ± 1.8° (p < 0.001) for the femoral component. The tibial component was placed in more varus of 0.2° ± 0.9° (p = 0.056) and more posterior slope of 0.5° ± 0.9° (p < 0.001). The lower limb alignment in extension was 4.4° ± 5.2° of varus of the native knee and changed to 1.2° ± 1.9° of varus after TKA (p < 0.01). Robotic-assisted TKA helps to achieve the target of alignment and component placement very close to the planning. It allows optimal component placement of off-the-shelf implants respecting patient's specific anatomy. Level II.

Aorta Segmentation in 3D CT Images by Combining Image Processing and Machine Learning Techniques.

Aorta segmentation is extremely useful in clinical practice, allowing the diagnosis of numerous pathologies, such as dissections, aneurysms and occlusive disease. In such cases, image segmentation is prerequisite for applying diagnostic algorithms, which in turn allow the prediction of possible complications and enable risk assessment, which is crucial in saving lives. The aim of this paper is to present a novel fully automatic 3D segmentation method, which combines basic image processing techniques and more advanced machine learning algorithms, for detecting and modelling the aorta in 3D CT imaging data. An initial intensity threshold-based segmentation procedure is followed by a classification-based segmentation approach, based on a Markov Random Field network. The result of the proposed two-stage segmentation process is modelled and visualized. The proposed methodology was applied to 16 3D CT data sets and the extracted aortic segments were reconstructed as 3D models. The performance of segmentation was evaluated both qualitatively and quantitatively against other commonly used segmentation techniques, in terms of the accuracy achieved, compared to the actual aorta, which was defined manually by experts. The proposed methodology achieved superior segmentation performance, compared to all compared segmentation techniques, in terms of the accuracy of the extracted 3D aortic model. Therefore, the proposed segmentation scheme could be used in clinical practice, such as in treatment planning and assessment, as it can speed up the evaluation of the medical imaging data, which is commonly a lengthy and tedious process.

Sacroiliitis diagnosis based on interpretable features and multi-task learning

Objective. Sacroiliitis is an early pathological manifestation of ankylosing spondylitis (AS), and a positive sacroiliitis test on imaging may help clinical practitioners diagnose AS early. Deep learning based automatic diagnosis algorithms can deliver grading findings for sacroiliitis, however, it requires a large amount of data with precise labels to train the model and lacks grading features visualization. In this paper, we aimed to propose a radiomics and deep learning based deep feature visualization positive diagnosis algorithm for sacroiliitis on CT scans. Visualization of grading features can enhance clinical interpretability with visual grading features, which assist doctors in diagnosis and treatment more effectively. Approach. The region of interest (ROI) is identified by segmenting the sacroiliac joint (SIJ) 3D CT images using a combination of the U-net model and certain statistical approaches. Then, in addition to extracting spatial and frequency domain features from ROI according to the radiographic manifestations of sacroiliitis, the radiomics features have also been integrated into the proposed encoder module to obtain a powerful encoder and extract features effectively. Finally, a multi-task learning technique and five-class labels are utilized to help with performing positive tests to reduce discrepancies in the evaluation of several radiologists. Main results. On our private dataset, proposed methods have obtained an accuracy rate of 87.3%, which is 9.8% higher than the baseline and consistent with assessments made by qualified medical professionals. Significance. The results of the ablation experiment and interpreting analysis demonstrated that the proposed methods are applied in automatic CT scan sacroiliitis diagnosis due to their excellently interpretable and portable advantages.

Six-Part Classification of Femoral Intertrochanteric Fractures: A Classification Method to Improve the Diagnosis Rate of Unstable Fractures.

Three-dimensional (3D)-CT data is currently insufficient for classifying femoral trochanter fractures. Fracture classification based on fracture stability analysis is helpful to evaluate the prognosis of patients after internal fixation. Currently, there is a lack of fracture classification methods based on 3D-CT images and fracture stability analysis. The aim of this study was to propose a new six-part classification method for intertrochanteric fractures of femur based on 3D-CT images and fracture stability analysis to improve the diagnosis rate of unstable fractures. From January 2009 to December 2019, 320 patients receiving intramedullary nail surgery for femoral intertrochanteric fractures at Chengdu University's Affiliated Hospital were studied retrospectively. AO and six-part classifications were undertaken according to the 3D-CT image data of the patients, and the stability rates of two classifications were compared. According to the six-part classification stability criteria, the patients were divided into a stable and an unstable fracture group. The perioperative and follow-up indicators of the two groups were statistically analyzed, and the six-part classification's inter-observer and internal reliability was examined. There were 107 men and 213 females women the 320 patients, with an average age of 79.32 ± 11.26 years and an osteoporosis rate of 55.63% (178/320). The fracture stability rate of 39.69% (127/320) was studied using a six-part classification method. The AO classification fracture stability rate was 42.50% (136/320), with no significant difference (χ2 = 0.523, p = 0.470 > 0.05). There is no statistically significant difference between the two classification techniques in the examination of fracture stability (McNemer difference test p = 0.306 > 0.05; Kappa consistency test p < 0.001). According to the six-part classification, fracture stability and instability group were divided into two groups. The following indicators were compared between the two groups: The surgery time (p = 0.280), fracture reduction quality (p = 0.062); function independent measurement (p = 0.075); timed up and go test (TUG) (p = 0.191), and Parker-Palmer score (p = 0.146). Were as compared according to the six-part classification of stable and unstable fracture groups. Perioperative blood loss (p < 0.001), the Harris score excellent and good rate (p = 0.043), fracture healing time (p < 0.001), and the entire weight-bearing duration (p = 0.002) were statistically significant. The difference in femoral head height (FHH) (p = 0.046), the change in femoral neck shaft angle (p = 0.003), the change in medial cephalic nail length (p = 0.033), and the change in tip-apex distance (TAD) (p = 0.002) were statistically significant compared to the relevant markers of imaging stability. Fracture stability had a substantial influence on Harris ratings at 3, 6, and 12 months following surgery, according to repeated measures analysis of variance (F(1,126) = 32.604, p < 0.001). The effect of time on the Harris score was similarly significant (F(1.893,238.508) = 202.771, p < 0.001). The observer intra-observer inter-group correlation coefficient (ICC) value was 0.941 > 0.75, the inter-observer ICC value was 0.921 > 0.75, and the intra-observer and inter-observer reliability were both good. The six-part classification of femoral intertrochanteric fractures based on 3D-CT images has broader guiding relevance for femoral intertrochanteric fracture stability analysis. Clinicians will find this classification simpler and more consistent than the AO classification.