An HMM-based recognition framework for endovascular manipulations.

Robotic surgical systems are becoming increasingly popular for the treatment of cardiovascular diseases. However, most of them have been designed without considering techniques and skills of natural surgical manipulations, which are key factors to clinical success of percutaneous coronary intervention. This paper proposes an HMM-based framework to recognize six typical endovascular manipulations for surgical skill analysis. A simulative surgical platform is built for endovascular manipulations assessed by five subjects (1 expert and 4 novices). The performances of the proposed framework are evaluated by three experimental schemes with the optimal model parameters. The results show that endovascular manipulations are recognized with high accuracy and reliable performance. Furthermore, the acceptable results can also be applied to the design of next generation vascular interventional robots.

Guide-wire detecting using a modified cascade classifier in interventional radiology.

Endovascular surgery is becoming a widespread procedure to treat cardiovascular diseases (CVDs) such as abdominal aortic aneurysm and peripheral artery disease. The guide-wire is a crucial surgical instrument inserted into vessels to offer guidance to physicians during the surgery. There are some approaches for tracking the guide-wire, most algorithms consist of two phases, namely, the initialization phase and the tracking phase. In the initialization phase, most algorithms use B-splines for modeling the guide-wire which requires manually annotated data. In the tracking phase, the guide-wire motion is non-linearity because it is deforming and changing its shape and size as a result of patients' respiration, some algorithms decompose the non-linearity motion into rigid motion and non-rigid motion, while the computational complexity is high especially for the non-rigid motion. This paper mainly presents an approach to detect the guide-wire. The algorithm has two main advantages. First, without modeling the guide-wire, this approach uses a cascade classifier which can detect the guide-wire under arbitrary motion automatically. Second, by taking the guide-wire motion direction into consideration, the detection accuracy improves significantly. The presented work has been validated on a test set of 349 frames, and the mean tracking accuracy achieves more than 95% which proves the effectiveness of the proposed method.

A dynamic EMG-torque model of elbow based on neural networks.

In this paper, a dynamic EMG-torque model of the elbow joint is developed based on ANN, and two novel test methods are proposed to validate its generalization performance. A time-delay neural network (TDNN) model is built and proved to have less risk of overfitting than the most-used multilayer feedfoward neural network (MFNN) model for dynamic EMG-torque modeling. Both EMG and kinematic features are included in the input of ANN, but the zero-EMG test shows that the trained ANN is part of the inverse joint dynamics rather than the EMG-torque model, and some random samples for ANN training are added to overcome this problem. The single-muscle test shows that an inappropriate choice of the motion type may cause the model to estimate wrong torque directions. After tuning and testing, the root mean square error (RMSE) across all subjects is 0.60±0.20 N.m.

Fast and stable guidewire simulator for minimally invasive vascular surgery.

In recent years, minimally invasive vascular surgery is widely applied in treatment of cardiovascular diseases, and the manipulation of the guidewire is the essential skill for this surgery. Lots of time and money have to be taken to achieve the skill. In this paper, we present a multithreading guidewire simulator which can help the apprentice to gain the skill and modeling the guidewire is the core technique of the simulator. The guidewire is modeled by a fast and stable method based on the Cosserat theory of elastic rods. The method describes the behavior of the guidewire with the Lagrange equations of motion and it uses the penalty method to maintain constraints. We further propose a simplified solving procedure for the guidewire model. Finally, some experiments are conducted to evaluate the effectiveness of this model.