The current convolution neural network (CNN)-based methods suffer from mass computation and low frame rate without special acceleration, they cannot be applied in some real applications. The authors propose a coarse to fine method for surgical tool tracking based on CNNs. Two CNNs are designed for the coarse and fine locations of the surgical tool. The coarse CNN is a classification network of 10 classes, and the fine CNN is a regression network for the tool tip area. The spatial and temporal context updating makes two CNNs cooperating together for the whole tracking process of surgical tool. The authors validate their method in the experiments with eight datasets, where there are two ex vivo datasets and six in vivo datasets, as well as comparing their method with other four methods in terms of accuracy and speed of tracking. The ex vivo and in vivo experiments demonstrate that the method gives consideration to both accuracy and speed, and provides a good accuracy and a high frame rate of tracking. For the future research of multi-tool tracking, the design of new regression CNN with the output of tensor will be focused, which contains multiple tools' labels and tracking information.

In order to predict how much the human ligamentum flavum will be deformed and tented during insertion of an epidural needle, experiments with the porcine spines were done to check the effect of Young's modulus of the porcine ligamentum flavum. Porcine ligamentum flavums were harvested with vertebral bones and trimmed in rectangle shape. The epidural needle was inserted at 0.8 mm/s with a linear actuator from perpendicular to stretch direction of the specimens. Reaction force as insertion resistance with a load cell and side view of the specimen and the needle as deformation of the ligamentum flavum with a CCD camera were recorded synchronously. As results, width of vertebral bones related the maximum deformation, while no relation was observed between width of vertebral bones and the maximum insertion force. Young's modulus affected the maximum insertion force. Human ligamentum flavum has longer width of vertebral bones and larger Young's modulus. Thus, larger maximum deformation and higher maximum insertion force for the human ligamentum flavum were predicted.

The number of stroke patients is rapidly increasing in the elderly society, which leads to growing demand for lower limb rehabilitation training. Currently, one patient needs two or more therapists for assistance during gait training. It results in the shortage of therapists' population, furthermore, heavy works load on the therapist. The emerging robotic technologies provide a solution to assist the therapist, and a number of corresponding researches have been reported. However, most of the existing rehabilitation robots adopt single-arm or double-arm structure, which pays less attention on motor coordination training for the stroke patients. Here, a four-arm rehabilitation robot (FARR) is proposed to assist the hemiplegic patient for motor coordination training. First, the rehabilitation demand is analysed and the corresponding robot mechanism is designed. Then, the kinematics of the robot based on the D-H expression is constructed, and the workspace is obtained. Thirdly, the speed control strategy and the cooperative control for gait training are constructed. The experiment of speed response verifies the superior tracking performance of the robotic joints, and the experiment of using the robot for gait training by a simulated subject is performed. These results prove the feasibility of the designed robot.

In practice, 3D modelling software, such as UG and Pro/E, are commonly adopted to design porous scaffolds within given contours. However, this manual method is quite time-consuming and complicated with poor adaptability. In this study, a novel facial algorithm is proposed for smart scaffold design. By forming a dendritic fractal network, the second-order uniformity and complete connectivity of pores are guaranteed while considering the axial symmetry of certain orthopaedic implants, and the algorithm is adapted for concave and hollow contour shapes. Through experiments, the stability of the algorithm and the characteristics of the two vital performance indicators, i.e. porosity and surface area ratio R, are studied. The results show that P and of the generated model are insensitive to the shape of the input model, which proves the stability of authors’ method, and the maximum available porosity reaches 78%, higher than the maximum effective porosity found in the literature.

This study aims to develop a robotic system human–machine interaction (HMI) to facilitate surgical training through visual and kinaesthetic feedbacks. This is motivated by the pressing need for effective surgical training and the unaddressed gaps in existing surgical training simulator for minimally invasive procedures. This study establishes the design concept and scope for development to facilitate the required HMI training model. Subsequently, implementation and demonstration of the model is carried out with analytical experiments to assess the feasibility of the proposed concept. The design concept of the robotic system for training is demonstrated through an user experiment. Results suggest viability and observable benefits in the authors’ proposed kinaesthetic HMI guidance for the trainee. Potential impact of this study includes the development of a novel training paradigm that engages trainees through collaborative training facilitated by human trainers and active kinaesthetic simulation. Although motivated by surgical training applications, the concept developed in this study can potentially be extended for general motor skill learning.

Robotic medical ultrasound can support diagnostics and alleviate fatigues. However, state-of-the-art ultrasound devices are too large and complex for home healthcare. Moreover, organs move in accordance with respiration. This movement changes the ultrasound image pattern and it is difficult for the operator to diagnose easily and accurately. To cope with these problems, the authors newly developed a compact portable ultrasound diagnostic robot for home healthcare, which compensate the organ motion. It can support those who find it difficult to visit a hospital for temporal, spatial, or physical reasons. A robust template matching method to servo the target was applied. Specifically, this robot moves to the target, whose position is detected by finding and identifying the image position of the target in the real-time input images. The authors also applied a multi-threading algorithm with two threads to enhance the real-time performance. One is for image processing with template matching. Another is for robot control to servo the target. Experimental results show that their proposed robot and algorithms can be suppressed to 25.2% motion in pk-to-pk for the periodic phantom organ motion (period of 3 s and of pk-to-pk 40 mm).

Drilling into the vertebra is a critical task in spinal surgeries. Traditionally, it is relied upon surgeon's intuition and experience. Robotic-assisted spinal surgery with fore feedback allows the drilling process to be performed automatically with much higher precision and safety. In this study, a new teleoperated robotic spinal surgery system with a frontend surgical instrument is designed, which is equipped with a force detection and feedback system. The proposed robotic system is manufactured and evaluated via laboratory and in vivo tests. Results prove the effectiveness of the system in precise teleoperation drilling, real-time force feedback, and achieving less surgical radiation exposure.

The brain deforms during neurosurgery, resulting in a decrease in the accuracy of the image guided neurosurgery system (IGNS). In this study, a wireless transmission-based brain shift compensation system (WBSCS) is implemented. The system consists of a laser range scanner for acquiring a cortical surface, and a mobile workstation equipped with a brain deformation correction software platform. The brain tissue deformation correction software platform includes a three-dimensional visualisation module, a calibration module, a brain tissue extraction module, a meshing module, a boundary condition acquisition module, a finite element calculation module, a preoperative image update module, and a communication module. The system exchanges data with the IGNS through wireless communication. The authors used five pigs to test the system. Results of experiments show that the system can compensate for brain deformation. It worked with IGNS to provide brain-shift-compensation-based guidance, improving the accuracy of the IGNS.

Multi-robots are used in hospital for the delivery service. To increase the delivery efficiency, an appropriate multi-robots path planning algorithm is required. This study explains a mixed path planning algorithm for multi-robots in structured hospital environment which combines the ‘corridor path planning’ and ‘room path planning’, ‘corridor path planning’ uses graph search algorithm to plan the path in corridor for multiple hospital robots, and it uses the strongly connected digraph to describe corridor environment, so it could simplify the multi-robots path planning problem. While the ‘room path planning’ uses the artificial potential field method to plan the path in room for multiple hospital robots, it makes the hospital robot move flexibly in the room, so the robot could avoid the obstacle and other robots more flexibly. The algorithm is tested by simulation, and the result shows that mixed path planning algorithm makes the robot move fast in the corridor environment and move flexibly in the room environment.

The number of operations using surgical robots are continuously increasing. To perform accurate surgeries, it is necessary to know the behaviour of intervened organs, especially their mechanical properties, which must be accurately determined. However, the physical properties of organs vary depending on age, gender, and environment, and thus, each organ exhibits particular mechanical properties. The authors propose a real-time assimilation system that identifies organ properties. Specifically, a 2D model using the finite element method and data assimilation, which is mostly used in Earth science, allows the identification of the physical parameters of organs. Data assimilation relies on a particle filter for efficiently solving the non-linear identification of parameters from a statics viewpoint. In addition, the semi-implicit Euler method discretises the proposed model and improves efficiency. The proposed approach can serve to the future implementation of a real-time and accurate framework for identifying mechanical properties of organs.

A large number of hemiplegic patients have motor dysfunction. Ankle joint plays a vital role in walking behaviours and needs rehabilitative training for the stroke survivors. However, existed rehabilitative interventions have limited effects to restore individual normal motor ability. This paper presents a novel ankle rehabilitation robot (NARR) to mimic the practical skills of the professional physiotherapist. The proposed robot consists of two symmetric mechanisms, which allows stroke survivor execute ankle rehabilitation exercise based on the individual difference. The kinematics of the NARR is calculated, and theoretically verifies the reachability of the angle range of ankle joint from the workspace. The speed response of the robotic joint in the NARR is tuned, and then, three ankle training modes – passive training, active training, and fusion training – with the NARR are developed. The experiments demonstrate that the patient can move his/her affected ankle to the set angle range in the passive training mode, and exercise the muscle power in the active training mode with the assistance from the NARR. In the fusion training mode, the hemiplegic patient can mirror exercise his/her affected ankle by collecting the ankle motion information from the corresponding sound ankle. The experimental results indicate that the NARR is capable of ankle rehabilitation.

We propose a three-dimensional-local directional pattern (3D-LDP) computation method in this study for early diagnosis of Alzheimer's disease. The proposed 3D-LDP is defined as a 1-bit decimal pattern which is calculated by computing and comparing edge response values in 18 directions. By concatenating three individual histograms extracted from three segmented sub-regions of grey matter, the accuracy, sensitivity, and specificity of the proposed 3D-LDP method were obtained as 81, 84, and 79%, respectively, comparable or even better than those of the existing 3D-LTP method (80, 85, and 78%).

High precision and servo-puncture navigation system is of great significance for the clinical application of central venous catheter (CVC) puncture-assist robot guided by ultrasound. Therefore, the authors proposed a new servo-predictive navigation system, which can track the posture of robot and predict the puncturing needle path in real time, effectively solving the limitation of conventional path plan line, such as path plan line deviation. For ensuring the accuracy of this predictive navigation system, a multi-parameter calibration method using particle swarm optimisation (PSO) based on self-adaptive system was adopted in this research, and applied to authors’ self-developed servo-predictive navigation system for puncture-assist robot MRAPS II. Experimental results indicated that by calibrating mechanical and control parameters, the authors can effectively improve the accuracy of servo-navigation system. It greatly promoted the use of CVC puncture-assist robot in clinical application.