Bi-Fusion of Structure and Deformation at Multi-scale for Joint Segmentation and Registration.

Medical image segmentation and registration are two fundamental and highly related tasks. However, current works focus on the mutual promotion between the two at the loss function level, ignoring the feature information generated by the encoder-decoder network during the task-specific feature mapping process and the potential inter-task feature relationship. This paper proposes a unified multi-task joint learning framework based on bi-fusion of structure and deformation at multi-scale, called BFM-Net, which simultaneously achieves the segmentation results and deformation field in a single-step estimation. BFM-Net consists of a segmentation subnetwork (SegNet), a registration subnetwork (RegNet), and the multi-task connection module (MTC). The MTC module is used to transfer the latent feature representation between segmentation and registration at multi-scale and link different tasks at the network architecture level, including the spatial attention fusion module (SAF), the multi-scale spatial attention fusion module (MSAF) and the velocity field fusion module (VFF). Extensive experiments on MR, CT and ultrasound images demonstrate the effectiveness of our approach. The MTC module can increase the Dice scores of segmentation and registration by 3.2%, 1.6%, 2.2%, and 6.2%, 4.5%, 3.0%, respectively. Compared with six state-of-the-art algorithms for segmentation and registration, BFM-Net can achieve superior performance in various modal images, fully demonstrating its effectiveness and generalization.

DSC-Recon: Dual-Stage Complementary 4D Organ Reconstruction from X-ray Image Sequence for Intraoperative Fusion.

Accurately reconstructing 4D critical organs contributes to the visual guidance in X-ray image-guided interventional operation. Current methods estimate intraoperative dynamic meshes by refining a static initial organ mesh from the semantic information in the single-frame X-ray images. However, these methods fall short of reconstructing an accurate and smooth organ sequence due to the distinct respiratory patterns between the initial mesh and X-ray image. To overcome this limitation, we propose a novel dual-stage complementary 4D organ reconstruction (DSC-Recon) model for recovering dynamic organ meshes by utilizing the preoperative and intraoperative data with different respiratory patterns. DSC-Recon is structured as a dual-stage framework: 1) The first stage focuses on addressing a flexible interpolation network applicable to multiple respiratory patterns, which could generate dynamic shape sequences between any pair of preoperative 3D meshes segmented from CT scans. 2) In the second stage, we present a deformation network to take the generated dynamic shape sequence as the initial prior and explore the discriminate feature (i.e., target organ areas and meaningful motion information) in the intraoperative X-ray images, predicting the deformed mesh by introducing a designed feature mapping pipeline integrated into the initialized shape refinement process. Experiments on simulated and clinical datasets demonstrate the superiority of our method over state-of-the-art methods in both quantitative and qualitative aspects.

Improving Image Segmentation with Contextual and Structural Similarity.

Deep learning models for medical image segmentation are usually trained with voxel-wise losses, e.g., cross-entropy loss, focusing on unary supervision without considering inter-voxel relationships. This oversight potentially leads to semantically inconsistent predictions. Here, we propose a contextual similarity loss (CSL) and a structural similarity loss (SSL) to explicitly and efficiently incorporate inter-voxel relationships for improved performance. The CSL promotes consistency in predicted object categories for each image sub-region compared to ground truth. The SSL enforces compatibility between the predictions of voxel pairs by computing pair-wise distances between them, ensuring that voxels of the same class are close together whereas those from different classes are separated by a wide margin in the distribution space. The effectiveness of the CSL and SSL is evaluated using a clinical cone-beam computed tomography (CBCT) dataset of patients with various craniomaxillofacial (CMF) deformities and a public pancreas dataset. Experimental results show that the CSL and SSL outperform state-of-the-art regional loss functions in preserving segmentation semantics.

Facial augmented reality based on hierarchical optimization of similarity aspect graph.

BACKGROUND: The existing face matching method requires a point cloud to be drawn on the real face for registration, which results in low registration accuracy due to the irregular deformation of the patient's skin that makes the point cloud have many outlier points. METHODS: This work proposes a non-contact pose estimation method based on similarity aspect graph hierarchical optimization. The proposed method constructs a distance-weighted and triangular-constrained similarity measure to describe the similarity between views by automatically identifying the 2D and 3D feature points of the face. A mutual similarity clustering method is proposed to construct a hierarchical aspect graph with 3D pose as nodes. A Monte Carlo tree search strategy is used to search the hierarchical aspect graph for determining the optimal pose of the facial 3D model, so as to realize the accurate registration of the facial 3D model and the real face. RESULTS: The proposed method was used to conduct accuracy verification experiments on the phantoms and volunteers, which were compared with four advanced pose calibration methods. The proposed method obtained average fusion errors of 1.13 ± 0.20 mm and 0.92 ± 0.08 mm in head phantom and volunteer experiments, respectively, which exhibits the best fusion performance among all comparison methods. CONCLUSIONS: Our experiments proved the effectiveness of the proposed pose estimation method in facial augmented reality.

Embedding-Alignment Fusion-Based Graph Convolution Network With Mixed Learning Strategy for 4D Medical Image Reconstruction.

In recent years, 4D medical image involving structural and motion information of tissue has attracted increasing attention. The key to the 4D image reconstruction is to stack the 2D slices based on matching the aligned motion states. In this study, the distribution of the 2D slices with the different motion states is modeled as a manifold graph, and the reconstruction is turned to be the graph alignment. An embedding-alignment fusion-based graph convolution network (GCN) with a mixed-learning strategy is proposed to align the graphs. Herein, the embedding and alignment processes of graphs interact with each other to realize a precise alignment with retaining the manifold distribution. The mixed strategy of self- and semi-supervised learning makes the alignment sparse to avoid the mismatching caused by outliers in the graph. In the experiment, the proposed 4D reconstruction approach is validated on the different modalities including Computed Tomography (CT), Magnetic Resonance Imaging (MRI), and Ultrasound (US). We evaluate the reconstruction accuracy and compare it with those of state-of-the-art methods. The experiment results demonstrate that our approach can reconstruct a more accurate 4D image.

Augmented Reality Navigation System for Biliary Interventional Procedures With Dynamic Respiratory Motion Correction.

OBJECTIVE: Biliary interventional procedures require physicians to track the interventional instrument tip (Tip) precisely with X-ray image. However, Tip positioning relies heavily on the physicians' experience due to the limitations of X-ray imaging and the respiratory interference, which leads to biliary damage, prolonged operation time, and increased X-ray radiation. METHODS: We construct an augmented reality (AR) navigation system for biliary interventional procedures. It includes system calibration, respiratory motion correction and fusion navigation. Firstly, the magnetic and 3D computed tomography (CT) coordinates are aligned through system calibration. Secondly, a respiratory motion correction method based on manifold regularization is proposed to correct the misalignment of the two coordinates caused by respiratory motion. Thirdly, the virtual biliary, liver and Tip from CT are overlapped to the corresponding position of the patient for dynamic virtual-real fusion. RESULTS: Our system is respectively evaluated and achieved an average alignment error of 0.75 ± 0.17 mm and 2.79 ± 0.46 mm on phantoms and patients. The navigation experiments conducted on phantoms achieve an average Tip positioning error of 0.98 ± 0.15 mm and an average fusion error of 1.67 ± 0.34 mm after correction. CONCLUSION: Our system can automatically register the Tip to the corresponding location in CT, and dynamically overlap the 3D virtual model onto patients to provide accurate and intuitive AR navigation. SIGNIFICANCE: This study demonstrates the clinical potential of our system by assisting physicians during biliary interventional procedures. Our system enables dynamic visualization of virtual model on patients, reducing the reliance on contrast agents and X-ray usage.

Robot-assisted augmented reality surgical navigation based on optical tracking for mandibular reconstruction surgery.

PURPOSE: This work proposes a robot-assisted augmented reality (AR) surgical navigation system for mandibular reconstruction. The system accurately superimposes the preoperative osteotomy plan of the mandible and fibula into a real scene. It assists the doctor in osteotomy quickly and safely under the guidance of the robotic arm. METHODS: The proposed system mainly consists of two modules: the AR guidance module of the mandible and fibula and the robot navigation module. In the AR guidance module, we propose an AR calibration method based on the spatial registration of the image tracking marker to superimpose the virtual models of the mandible and fibula into the real scene. In the robot navigation module, the posture of the robotic arm is first calibrated under the tracking of the optical tracking system. The robotic arm can then be positioned at the planned osteotomy after the registration of the computed tomography image and the patient position. The combined guidance of AR and robotic arm can enhance the safety and precision of the surgery. RESULTS: The effectiveness of the proposed system was quantitatively assessed on cadavers. In the AR guidance module, osteotomies of the mandible and fibula achieved mean errors of 1.61 ± 0.62 and 1.08 ± 0.28 mm, respectively. The mean reconstruction error of the mandible was 1.36 ± 0.22 mm. In the AR-robot guidance module, the mean osteotomy errors of the mandible and fibula were 1.47 ± 0.46 and 0.98 ± 0.24 mm, respectively. The mean reconstruction error of the mandible was 1.20 ± 0.36 mm. CONCLUSIONS: The cadaveric experiments of 12 fibulas and six mandibles demonstrate the proposed system's effectiveness and potential clinical value in reconstructing the mandibular defect with a free fibular flap.

A multi-dimensional CFD framework for fast patient-specific fractional flow reserve prediction.

Fractional flow reserve (FFR) is considered as the gold standard for diagnosing coronary myocardial ischemia. Existing 3D computational fluid dynamics (CFD) methods attempt to predict FFR noninvasively using coronary computed tomography angiography (CTA). However, the accuracy and efficiency of the 3D CFD methods in coronary arteries are considerably limited. In this work, we introduce a multi-dimensional CFD framework that improves the accuracy of FFR prediction by estimating 0D patient-specific boundary conditions, and increases the efficiency by generating 3D initial conditions. The multi-dimensional CFD models contain the 3D vascular model for coronary simulation, the 1D vascular model for iterative optimization, and the 0D vascular model for boundary conditions expression. To improve the accuracy, we utilize clinical parameters to derive 0D patient-specific boundary conditions with an optimization algorithm. To improve the efficiency, we evaluate the convergence state using the 1D vascular model and obtain the convergence parameters to generate appropriate 3D initial conditions. The 0D patient-specific boundary conditions and the 3D initial conditions are used to predict FFR (FFRC). We conducted a retrospective study involving 40 patients (61 diseased vessels) with invasive FFR and their corresponding CTA images. The results demonstrate that the FFRC and the invasive FFR have a strong linear correlation (r = 0.80, p < 0.001) and high consistency (mean difference: 0.014 ±0.071). After applying the cut-off value of FFR (0.8), the accuracy, sensitivity, specificity, positive predictive value, and negative predictive value of FFRC were 88.5%, 93.3%, 83.9%, 84.8%, and 92.9%, respectively. Compared with the conventional zero initial conditions method, our method improves prediction efficiency by 71.3% per case. Therefore, our multi-dimensional CFD framework is capable of improving the accuracy and efficiency of FFR prediction significantly.

Monocular endoscopy images depth estimation with multi-scale residual fusion.

BACKGROUND: Monocular depth estimation plays a fundamental role in clinical endoscopy surgery. However, the coherent illumination, smooth surfaces, and texture-less nature of endoscopy images present significant challenges to traditional depth estimation methods. Existing approaches struggle to accurately perceive depth in such settings. METHOD: To overcome these challenges, this paper proposes a novel multi-scale residual fusion method for estimating the depth of monocular endoscopy images. Specifically, we address the issue of coherent illumination by leveraging image frequency domain component space transformation, thereby enhancing the stability of the scene's light source. Moreover, we employ an image radiation intensity attenuation model to estimate the initial depth map. Finally, to refine the accuracy of depth estimation, we utilize a multi-scale residual fusion optimization technique. RESULTS: To evaluate the performance of our proposed method, extensive experiments were conducted on public datasets. The structural similarity measures for continuous frames in three distinct clinical data scenes reached impressive values of 0.94, 0.82, and 0.84, respectively. These results demonstrate the effectiveness of our approach in capturing the intricate details of endoscopy images. Furthermore, the depth estimation accuracy achieved remarkable levels of 89.3 % and 91.2 % for the two models' data, respectively, underscoring the robustness of our method. CONCLUSIONS: Overall, the promising results obtained on public datasets highlight the significant potential of our method for clinical applications, facilitating reliable depth estimation and enhancing the quality of endoscopy surgical procedures.

PEA-Net: A progressive edge information aggregation network for vessel segmentation.

Automatic vessel segmentation is a critical area of research in medical image analysis, as it can greatly assist doctors in accurately and efficiently diagnosing vascular diseases. However, accurately extracting the complete vessel structure from images remains a challenge due to issues such as uneven contrast and background noise. Existing methods primarily focus on segmenting individual pixels and often fail to consider vessel features and morphology. As a result, these methods often produce fragmented results and misidentify vessel-like background noise, leading to missing and outlier points in the overall segmentation. To address these issues, this paper proposes a novel approach called the progressive edge information aggregation network for vessel segmentation (PEA-Net). The proposed method consists of several key components. First, a dual-stream receptive field encoder (DRE) is introduced to preserve fine structural features and mitigate false positive predictions caused by background noise. This is achieved by combining vessel morphological features obtained from different receptive field sizes. Second, a progressive complementary fusion (PCF) module is designed to enhance fine vessel detection and improve connectivity. This module complements the decoding path by combining features from previous iterations and the DRE, incorporating nonsalient information. Additionally, segmentation-edge decoupling enhancement (SDE) modules are employed as decoders to integrate upsampling features with nonsalient information provided by the PCF. This integration enhances both edge and segmentation information. The features in the skip connection and decoding path are iteratively updated to progressively aggregate fine structure information, thereby optimizing segmentation results and reducing topological disconnections. Experimental results on multiple datasets demonstrate that the proposed PEA-Net model and strategy achieve optimal performance in both pixel-level and topology-level metrics.

Automatic literature screening using the PAJO deep-learning model for clinical practice guidelines.

BACKGROUND: Clinical practice guidelines (CPGs) are designed to assist doctors in clinical decision making. High-quality research articles are important for the development of good CPGs. Commonly used manual screening processes are time-consuming and labor-intensive. Artificial intelligence (AI)-based techniques have been widely used to analyze unstructured data, including texts and images. Currently, there are no effective/efficient AI-based systems for screening literature. Therefore, developing an effective method for automatic literature screening can provide significant advantages. METHODS: Using advanced AI techniques, we propose the Paper title, Abstract, and Journal (PAJO) model, which treats article screening as a classification problem. For training, articles appearing in the current CPGs are treated as positive samples. The others are treated as negative samples. Then, the features of the texts (e.g., titles and abstracts) and journal characteristics are fully utilized by the PAJO model using the pretrained bidirectional-encoder-representations-from-transformers (BERT) model. The resulting text and journal encoders, along with the attention mechanism, are integrated in the PAJO model to complete the task. RESULTS: We collected 89,940 articles from PubMed to construct a dataset related to neck pain. Extensive experiments show that the PAJO model surpasses the state-of-the-art baseline by 1.91% (F1 score) and 2.25% (area under the receiver operating characteristic curve). Its prediction performance was also evaluated with respect to subject-matter experts, proving that PAJO can successfully screen high-quality articles. CONCLUSIONS: The PAJO model provides an effective solution for automatic literature screening. It can screen high-quality articles on neck pain and significantly improve the efficiency of CPG development. The methodology of PAJO can also be easily extended to other diseases for literature screening.

Local Contractive Registration with Biomechanical Model: Assessing Microwave Ablation after Compensation for Tissue Shrinkage.

Microwave ablation (MWA) is a minimally invasive procedure for the treatment of liver tumor. Accumulating clinical evidence has considered the minimal ablative margin (MAM) as a significant predictor of local tumor progression (LTP). In clinical practice, MAM assessment is typically carried out through image registration of pre- and post-MWA images. However, this process faces two main challenges: non-homologous match between tumor and coagulation with inconsistent image appearance, and tissue shrinkage caused by thermal dehydration. These challenges result in low precision when using traditional registration methods for MAM assessment. In this paper, we present a local contractive nonrigid registration method using a biomechanical model (LC-BM) to address these challenges and precisely assess the MAM. The LC-BM contains two consecutive parts: (1) local contractive decomposition (LC-part), which reduces the incorrect match between the tumor and coagulation and quantifies the shrinkage in the external coagulation region, and (2) biomechanical model constraint (BM-part), which compensates for the shrinkage in the internal coagulation region. After quantifying and compensating for tissue shrinkage, the warped tumor is overlaid on the coagulation, and then the MAM is assessed. We evaluated the method using prospectively collected data from 36 patients with 47 liver tumors, comparing LC-BM with 11 state-of-the-art methods. LTP was diagnosed through contrast-enhanced MR follow-up images, serving as the ground truth for tumor recurrence. LC-BM achieved the highest accuracy (97.9%) in predicting LTP, outperforming other methods. Therefore, our proposed method holds significant potential to improve MAM assessment in MWA surgeries.

CT2X-IRA: CT to x-ray image registration agent using domain-cross multi-scale-stride deep reinforcement learning.

Objective.In computer-assisted minimally invasive surgery, the intraoperative x-ray image is enhanced by overlapping it with a preoperative CT volume to improve visualization of vital anatomical structures. Therefore, accurate and robust 3D/2D registration of CT volume and x-ray image is highly desired in clinical practices. However, previous registration methods were prone to initial misalignments and struggled with local minima, leading to issues of low accuracy and vulnerability.Approach.To improve registration performance, we propose a novel CT/x-ray image registration agent (CT2X-IRA) within a task-driven deep reinforcement learning framework, which contains three key strategies: (1) a multi-scale-stride learning mechanism provides multi-scale feature representation and flexible action step size, establishing fast and globally optimal convergence of the registration task. (2) A domain adaptation module reduces the domain gap between the x-ray image and digitally reconstructed radiograph projected from the CT volume, decreasing the sensitivity and uncertainty of the similarity measurement. (3) A weighted reward function facilitates CT2X-IRA in searching for the optimal transformation parameters, improving the estimation accuracy of out-of-plane transformation parameters under large initial misalignments.Main results.We evaluate the proposed CT2X-IRA on both the public and private clinical datasets, achieving target registration errors of 2.13 mm and 2.33 mm with the computation time of 1.5 s and 1.1 s, respectively, showing an accurate and fast workflow for CT/x-ray image rigid registration.Significance.The proposed CT2X-IRA obtains the accurate and robust 3D/2D registration of CT and x-ray images, suggesting its potential significance in clinical applications.

An alternately optimized generative adversarial network with texture and content constraints for deformable registration of 3D ultrasound images.

Objective.3D ultrasound non-rigid registration is significant for intraoperative motion compensation. Nevertheless, distorted textures in the registered image due to the poor image quality and low signal-to-noise ratio of ultrasound images reduce the accuracy and efficiency of the existing methods.Approach.A novel 3D ultrasound non-rigid registration objective function with texture and content constraints in both image space and multiscale feature space based on an unsupervised generative adversarial network based registration framework is proposed to eliminate distorted textures. A similarity metric in the image space is formulated based on combining self-structural constraint with intensity to strengthen the robustness to abnormal intensity change compared with common intensity-based metrics. The proposed framework takes two discriminators as feature extractors to formulate the texture and content similarity between the registered image and the fixed image in the multiscale feature space respectively. A distinctive alternating training strategy is established to jointly optimize the combination of various similarity loss functions to overcome the difficulty and instability of training convergence and balance the training of generator and discriminators.Main results.Compared with five registration methods, the proposed method is evaluated both with small and large deformations, and achieves the best registration accuracy with average target registration error of 1.089 mm and 2.139 mm in cases of small and large deformations, respectively. The performance on peak signal to noise ratio (PSNR) and structural similarity (SSIM) also proves the effective constraints on distorted textures of the proposed method (PSNR is 31.693 dB and SSIM is 0.9 in the case of small deformation; PSNR is 28.177 dB and SSIM is 0.853 in the case of large deformation).Significance.The proposed 3D ultrasound non-rigid registration method based on texture and content constraints with the distinctive alternating training strategy can eliminate the distorted textures with improving the registration accuracy.

Adaptive tetrahedral interpolation for reconstruction of uneven freehand 3D ultrasound.

Objective.Freehand 3D ultrasound volume reconstruction has received considerable attention in medical research because it can freely perform spatial imaging at a low cost. However, the uneven distribution of the original ultrasound images in space reduces the reconstruction effect of the traditional method.Approach.An adaptive tetrahedral interpolation algorithm is proposed to reconstruct 3D ultrasound volume data. The algorithm adaptively divides the unevenly distributed images into numerous tetrahedrons and interpolates the voxel value in each tetrahedron to reconstruct 3D ultrasound volume data.Main results.Extensive experiments on simulated and clinical data confirm that the proposed method can achieve more accurate reconstruction than six benchmark methods. Specifically, the averaged interpolation error at the gray level can be reduced by 0.22-0.82, and the peak signal-to-noise ratio and the mean structure similarity can be improved by 0.32-1.83 dB and 0.01-0.05, respectively.Significance.With the parallel implementation of the algorithm, one 3D ultrasound volume data with size 279 × 279 × 276 can be reconstructed from 100 slices 2D ultrasound images with size 200 × 200 at 1.04 s. Such a quick and accurate approach has practical value in medical research.

Simulation of Postoperative Facial Appearances via Geometric Deep Learning for Efficient Orthognathic Surgical Planning.

Orthognathic surgery corrects jaw deformities to improve aesthetics and functions. Due to the complexity of the craniomaxillofacial (CMF) anatomy, orthognathic surgery requires precise surgical planning, which involves predicting postoperative changes in facial appearance. To this end, most conventional methods involve simulation with biomechanical modeling methods, which are labor intensive and computationally expensive. Here we introduce a learning-based framework to speed up the simulation of postoperative facial appearances. Specifically, we introduce a facial shape change prediction network (FSC-Net) to learn the nonlinear mapping from bony shape changes to facial shape changes. FSC-Net is a point transform network weakly-supervised by paired preoperative and postoperative data without point-wise correspondence. In FSC-Net, a distance-guided shape loss places more emphasis on the jaw region. A local point constraint loss restricts point displacements to preserve the topology and smoothness of the surface mesh after point transformation. Evaluation results indicate that FSC-Net achieves 15× speedup with accuracy comparable to a state-of-the-art (SOTA) finite-element modeling (FEM) method.

Bidirectional prediction of facial and bony shapes for orthognathic surgical planning.

This paper proposes a deep learning framework to encode subject-specific transformations between facial and bony shapes for orthognathic surgical planning. Our framework involves a bidirectional point-to-point convolutional network (P2P-Conv) to predict the transformations between facial and bony shapes. P2P-Conv is an extension of the state-of-the-art P2P-Net and leverages dynamic point-wise convolution (i.e., PointConv) to capture local-to-global spatial information. Data augmentation is carried out in the training of P2P-Conv with multiple point subsets from the facial and bony shapes. During inference, network outputs generated for multiple point subsets are combined into a dense transformation. Finally, non-rigid registration using the coherent point drift (CPD) algorithm is applied to generate surface meshes based on the predicted point sets. Experimental results on real-subject data demonstrate that our method substantially improves the prediction of facial and bony shapes over state-of-the-art methods.

Multi3:multi-templates siamese network withmulti-peaks detection andmulti-features refinement for target tracking in ultrasound image sequences.

Objective. Radiation therapy requires a precise target location. However, respiratory motion increases the uncertainties of the target location. Accurate and robust tracking is significant for improving operation accuracy.Approach. In this work, we propose a tracking framework Multi3, including a multi-templates Siamese network, multi-peaks detection and multi-features refinement, for target tracking in ultrasound sequences. Specifically, we use two templates to provide the location and deformation of the target for robust tracking. Multi-peaks detection is applied to extend the set of potential target locations, and multi-features refinement is designed to select an appropriate location as the tracking result by quality assessment.Main results. The proposed Multi3 is evaluated on a public dataset, i.e. MICCAI 2015 challenge on liver ultrasound tracking (CLUST), and our clinical dataset provided by the Chinese People's Liberation Army General Hospital. Experimental results show that Multi3 achieves accurate and robust tracking in ultrasound sequences (0.75 ± 0.62 mm and 0.51 ± 0.32 mm tracking errors in two datasets).Significance. The proposed Multi3 is the most robust method on the CLUST 2D benchmark set, exhibiting potential in clinical practice.

Joint estimation of depth and motion from a monocular endoscopy image sequence using a multi-loss rebalancing network.

Building an in vivo three-dimensional (3D) surface model from a monocular endoscopy is an effective technology to improve the intuitiveness and precision of clinical laparoscopic surgery. This paper proposes a multi-loss rebalancing-based method for joint estimation of depth and motion from a monocular endoscopy image sequence. The feature descriptors are used to provide monitoring signals for the depth estimation network and motion estimation network. The epipolar constraints of the sequence frame is considered in the neighborhood spatial information by depth estimation network to enhance the accuracy of depth estimation. The reprojection information of depth estimation is used to reconstruct the camera motion by motion estimation network with a multi-view relative pose fusion mechanism. The relative response loss, feature consistency loss, and epipolar consistency loss function are defined to improve the robustness and accuracy of the proposed unsupervised learning-based method. Evaluations are implemented on public datasets. The error of motion estimation in three scenes decreased by 42.1%,53.6%, and 50.2%, respectively. And the average error of 3D reconstruction is 6.456 ± 1.798mm. This demonstrates its capability to generate reliable depth estimation and trajectory reconstruction results for endoscopy images and meaningful applications in clinical.

Localization of Craniomaxillofacial Landmarks on CBCT Images Using 3D Mask R-CNN and Local Dependency Learning.

Cephalometric analysis relies on accurate detection of craniomaxillofacial (CMF) landmarks from cone-beam computed tomography (CBCT) images. However, due to the complexity of CMF bony structures, it is difficult to localize landmarks efficiently and accurately. In this paper, we propose a deep learning framework to tackle this challenge by jointly digitalizing 105 CMF landmarks on CBCT images. By explicitly learning the local geometrical relationships between the landmarks, our approach extends Mask R-CNN for end-to-end prediction of landmark locations. Specifically, we first apply a detection network on a down-sampled 3D image to leverage global contextual information to predict the approximate locations of the landmarks. We subsequently leverage local information provided by higher-resolution image patches to refine the landmark locations. On patients with varying non-syndromic jaw deformities, our method achieves an average detection accuracy of 1.38± 0.95mm, outperforming a related state-of-the-art method.