An optical see-through (OST) head mounted display (HMD) is an appropriate option for image-guided surgical navigation system (IGS) based on augmented reality (AR). However, the calibration of OST-HMDs has been challenging, since the augmented results can only be observed by the users wearing it. In this paper, we proposed an online calibration method based on the single point active alignment method (SPAAM) for a HoloLens using a virtual tracking system. The transformation matrices between the world coordinates to the virtual holographic coordinates are calculated using a linear model and decomposed by the Singular Value Decomposition (SVD) algorithm. The surgical workspace is analyzed, and there are nine sampling points for the offline calibration, including eight at the vertices and one at the center of a cube of the workspace. 20 groups of data at each sampling point are collected. 5 alignments should be implanted in the online calibration. The transformation matrices are compensated for the offset of every use of a HoloLens. To assess the accuracy of the calibration method, in accordance with the RANSAC algorithm, 8 groups of additional data at random non-sampling locations are collected. Three different users that are familiar with the calibration procedures performed the calibration. The average distance error of our calibration was below 6 mm, and the rotation error was up to 5°. It is a user-friendly, simple method for the online calibration of OST-HMDs.

Cardiovascular diseases (CVDs) remain the foremost cause of mortality globally, emphasizing the imperative for early detection to improve patient outcomes and mitigate healthcare burdens. Carotid intima-media thickness (CIMT) serves as a well-established predictive marker for atherosclerosis and cardiovascular risk assessment. Fundus imaging offers a non-invasive modality to investigate microvascular pathology and systemic vascular health. However, the paucity of high-quality, publicly available datasets linking fundus images with CIMT measurements has hindered the progression of AI-driven predictive models for CVDs. Addressing this gap, we introduce the China-Fundus-CIMT dataset, comprising bilateral high-resolution fundus images, CIMT measurements, and clinical data—including age and gender—from 2,903 patients. Our experiments with multimodal models reveal that integrating clinical information substantially enhances predictive performance, yielding AUC-ROC increases of 3.22% and 7.83% on the validation and test sets, respectively, compared to unimodal models. This dataset constitutes a vital resource for developing and validating AI-based early screening models for CVDs using fundus images and is now accessible to the research community.

Objective: To develop and validate an artificial intelligence diagnostic model based on fundus images for predicting Carotid Intima-Media Thickness (CIMT) in individuals with Type 2 Diabetes Mellitus (T2DM).Methods: In total, 1236 patients with T2DM who had both retinal fundus images and CIMT ultrasound records within a single hospital stay were enrolled. Data were divided into normal and thickened groups and sent to eight deep learning models: convolutional neural networks of the eight models were all based on ResNet or ResNeXt. Their encoder and decoder modes are different, including the standard mode, the Parallel learning mode, and the Siamese mode. Except for the six unimodal networks, two multimodal networks based on ResNeXt under the Parallel learning mode or the Siamese mode were embedded with ages. Performance of eight models were compared via the confusion matrix, precision, recall, specificity, F1 value, and ROC curve, and recall was regarded as the main indicator. Besides, Grad-CAM was used to visualize the decisions made by Siamese ResNeXt network, which is the best performance.Results: Performance of various models demonstrated the following points: 1) the RexNeXt showed a notable improvement over the ResNet; 2) the structural Siamese networks, which extracted features parallelly and independently, exhibited slight performance enhancements compared to the traditional networks. Notably, the Siamese networks resulted in significant improvements; 3) the performance of classification declined if the age factor was embedded in the network. Taken together, the Siamese ResNeXt unimodal model performed best for its superior efficacy and robustness. This model achieved a recall rate of 88.0% and an AUC value of 90.88% in the validation subset. Additionally, heatmaps calculated by the Grad-CAM algorithm presented concentrated and orderly mappings around the optic disc vascular area in normal CIMT groups and dispersed, irregular patterns in thickened CIMT groups.Conclusion: We provided a Siamese ResNeXt neural network for predicting the carotid intimal thickness of patients with T2DM from fundus images and confirmed the correlation between fundus microvascular lesions and CIMT.

To improve the positioning accuracy of tunnels for anterior cruciate ligament (ACL) reconstruction, we proposed an ACL reconstruction robotic positioning system based on anatomical characteristics. The system includes a preoperative path planning system, an intraoperative path planning system, and a navigation and positioning system. Brahmet line, anterior, and posterior cortical lines are used for registration of preoperative computed tomography (CT) images and intraoperative X-ray images. A new calibrator of C-arm is applied to establish the mapping between medical images and surgical space. Tunnels for ACL reconstruction can be built anatomically by the robot. The accuracy of the path planning system is 1.73 mm in the four dry bones experiments and 2.17 mm in the two cadaver experiments. The accuracy meets the accuracy requirement of ACL construction surgery.

To address the challenges of confined workspaces and high-precision requirements in thyroid surgery, this paper proposes a modular cable-driven robotic system with a hybrid rigid–continuum structure. By integrating rigid mechanisms and continuum joints within a closed-loop cable-driven framework, the system achieves a balance between flexibility in narrow spaces and operational stiffness. To tackle kinematic model inaccuracies caused by manufacturing errors, an innovative joint decoupling strategy combined with the Particle Swarm Optimization (PSO) algorithm is developed to dynamically identify and correct 19 critical parameters. Experimental results demonstrate a 37.74% average improvement in repetitive positioning accuracy and a 52% reduction in maximum absolute error. However, residual positioning errors (up to 4.53 mm) at motion boundaries highlight the need for integrating nonlinear friction compensation. The feasibility of a safety-zone-based force feedback master–slave control strategy is validated through Gazebo simulations, and a ring-grasping experiment on a surgical training platform confirms its clinical applicability.