28 January 2022, Volume 27 Issue 1 Previous Issue   
Robotics & AI in Interdisciplinary Medicine and Engineering
Eye Robotic System for Vitreoretinal Surgery
DAI Qianlin (代倩琳), XU Mengqiao (徐梦乔), SUN Xiaodong (孙晓东), XIE Le∗ (谢叻)
2022, 27 (1):  1-6.  doi: 10.1007/s12204-021-2369-2
Abstract ( 47 )   PDF (1040KB) ( 19 )  
Micro incision vitrectomy system (MIVS) is considered to be one of the most difficult tasks of eye surgery, due to its requirements of high accuracy and delicate operation under blurred vision environment. Therefore, robot-assisted ophthalmic surgery is a potential and efficient solution. Based on that consideration, a novel master-slave system for vitreoretinal surgery is realized. A 4-DOF remote center of motion (RCM) mechanism with a novel linear stage and end-effector is designed and the master-slave control system is implemented. The forward and inverse kinematics are analyzed for the controller implementation. Then, algorithms with motion scaling are also integrated into the control architecture for the purpose to enhance the surgeon’s operation accuracy. Finally, experiments on an eye model are conducted. The results show that the eye robotic system can fulfill surgeon’s motion following and simulate operation of vitrectomy, demonstrating the feasibility of this system.
References | Related Articles | Metrics
Development of a Robotic Cochlear Implantation System
CHEN Ziyun (陈子云), XIE Le (谢叻), DAI Peidong (戴培东), ZHANG Tianyu (张天宇)
2022, 27 (1):  7-14.  doi: 10.1007/s12204-021-2381-6
Abstract ( 42 )   PDF (1384KB) ( 9 )  
Traditional cochlear implantation surgery has problems such as high surgical accuracy requirement and large trauma, which cause the difficulty of the operation and the high requirements for doctors, so that only a few doctors can complete the operation independently. However, there is no research on robotic cochlear implantation in China. In response to this problem, a robotic cochlear implantation system is proposed. The robot is controlled by robot operating system (ROS). A simulation environment for the overall surgery is established on the ROS based on the real surgery environment. Through the analysis of the kinematics and the motion planning algorithm of the manipulator, an appropriate motion mode is designed to control the motion of the manipulator, and perform the surgery under the simulation environment. A simple and feasible method of navigation is proposed, and through the model experiment, the feasibility of robotic cochlear implantation surgery is verified.
References | Related Articles | Metrics
Teleoperated Puncture Robot System: Preliminary Design and Workspace Analysis
HU Bo (胡博), LIN Yanping∗ (林艳萍), CHEN Shihang (陈士行), WANG Fang (汪方), MA Xiaojun (马小军), CAO Qixin (曹其新)
2022, 27 (1):  15-23.  doi: 10.1007/s12204-021-2368-3
Abstract ( 36 )   PDF (1455KB) ( 8 )  
Radiofrequency ablation (RFA) guided by X-ray images aims to relieve herniated disc pain with minimal invasiveness and fast recovery. It requires an accurate and fast positioning of the puncture needle. We propose a teleoperated robotic system for percutaneous puncture to support RFA. We report the kinematics modelling and workspace analysis of the proposed system, which comprises preliminary and accurate positioning mechanisms. Preliminary positioning mechanism automatically drives the needle to the puncture area, and accurate positioning is then achieved by teleoperation under the guidance of X-ray images. We calculate the teleoperation workspace of the robot system using a spatial search algorithm and quantitatively analyze the optimal structural parameters aiming to maximize the workspace. The workspace of the proposed robot system complies with clinical requirements to support RFA.
References | Related Articles | Metrics
Dynamic Obstacle Avoidance for Application of Human-Robot Cooperative Dispensing Medicines
WANG Zheng (王正), XU Hui (许辉), L v Na (吕娜), TAO Wei∗ (陶卫), CHEN Guodong (陈国栋), CHI Wenzheng (迟文正), SUN Lining (孙立宁)
2022, 27 (1):  24-35.  doi: 10.1007/s12204-021-2366-5
Abstract ( 20 )   PDF (2378KB) ( 5 )  
For safety reasons, in the automated dispensing medicines process, robots and humans cooperate to accomplish the task of drug sorting and distribution. In this dynamic unstructured environment, such as a humanrobot collaboration scenario, the safety of human, robot, and equipment in the environment is paramount. In this work, a practical and effective robot motion planning method is proposed for dynamic unstructured environments. To figure out the problems of blind zones of single depth sensor and dynamic obstacle avoidance, we first propose a method for establishing offline mapping and online fusion of multi-sensor depth images and 3D grids of the robot workspace, which is used to determine the occupation states of the 3D grids occluded by robots and obstacles and to conduct real-time estimation of the minimum distance between the robot and obstacles. Then, based on the reactive control method, the attractive and repulsive forces are calculated and transformed into robot joint velocities to avoid obstacles in real time. Finally, the robot’s dynamic obstacle avoidance ability is evaluated on an experimental platform with a UR5 robot and two KinectV2 RGB-D sensors, and the effectiveness of the proposed method is verified.
References | Related Articles | Metrics
Depth Camera-Based Robot-Assisted Ultrasonic Lipolysis System
YAN Minpeng (严旻芃), CHAI Gang ∗ (柴岗), XIE Le ∗ (谢叻)
2022, 27 (1):  36-44.  doi: 10.1007/s12204-021-2343-z
Abstract ( 20 )   PDF (1756KB) ( 4 )  
With many advantages such as non-invasive, safe and quick effect, focused ultrasound lipolysis stands out among many fat-removing methods. However, during the whole process, the doctor needs to hold the ultrasound transducer and press it on the patient’s skin with a large pressure for a long time; thus the probability of muscle and bone damage for doctors is greatly increased. To reduce the occurrence of doctors’ occupational diseases, a depth camera-based ultrasonic lipolysis robot system is proposed to realize robot-assisted automatic ultrasonic lipolysis operation. The system is composed of RealSense depth camera, KUKA LBR Med seven-axis robotic arm, PC host, and ultrasonic lipolysis instrument. The whole operation includes two parts: preoperative planning and intraoperative operation. In preoperative planning, the treatment area is selected in the camera image by the doctor; then the system automatically plans uniformly distributed treatment points in the treatment area. At the same time, the skin normal vector is calculated to determine the end posture of the robot, so that the ultrasound transducer can be pressed down in the normal direction of skin. During the intraoperative operation, the robot is controlled to arrive at the treatment point in turn. Meanwhile, the patient’s movement can be detected by the depth camera, and the path of robot is adjusted in real time so that the robot can track the movement of patient, thereby ensuring the accuracy of the ultrasonic lipolysis operation. Finally, the human body model experiment is conducted. The results show that the maximum error of the robot operation is within 5mm, average error is 3.1mm, and the treatment points of the robot operation are more uniform than those of manual operation. Therefore, the system can replace the doctor and achieve autonomous ultrasonic lipolysis to reduce the doctor’s labor intensity.
References | Related Articles | Metrics
Safety Protection Method of Rehabilitation Robot Based on fNIRS and RGB-D Information Fusion
LI Dong (李栋), FAN Yulin (樊钰琳), L v Na (吕娜), CHEN Guodong∗ (陈国栋), WANG Zheng (王正), CHI Wenzheng (迟文政)
2022, 27 (1):  45-54.  doi: 10.1007/s12204-021-2365-6
Abstract ( 20 )   PDF (2503KB) ( 2 )  
In order to improve the safety protection performance of the rehabilitation robot, an active safety protection method is proposed in the rehabilitation scene. The oxyhemoglobin concentration information and RGB-D information are combined in this method, which aims to realize the comprehensive monitoring of the invasion target, the patient’s brain function movement state, and the joint angle in the rehabilitation scene. The main focus is to study the fusion method of the oxyhemoglobin concentration information and RGB-D information in the rehabilitation scene. Frequency analysis of brain functional connectivity coefficient was used to distinguish the basic motion states. The human skeleton recognition algorithm was used to realize the angle monitoring of the upper limb joint combined with the depth information. Compared with speed and separation monitoring, the protection method of multi-information fusion is safer and more comprehensive for stroke patients. By building the active safety protection platform of the upper limb rehabilitation robot, the performance of the system in different safety states is tested, and the safety protection performance of the method in the upper limb rehabilitation scene is verified.
References | Related Articles | Metrics
Multi-Lead ECG Classification via an Information-Based Attention Convolutional Neural Network
TUNG Hao (董昊), ZHENG Chao (郑超), MAO Xinsheng(毛新生), QIAN Dahong (钱大宏)
2022, 27 (1):  55-69.  doi: 10.1007/s12204-021-2371-8
Abstract ( 18 )   PDF (943KB) ( 3 )  
A novel structure based on channel-wise attention mechanism is presented in this paper. With the proposed structure embedded, an efficient classification model that accepts multi-lead electrocardiogram (ECG) as input is constructed. One-dimensional convolutional neural networks (CNNs) have proven to be effective in pervasive classification tasks, enabling the automatic extraction of features while classifying targets. We implement the residual connection and design a structure which can learn the weights from the information contained in different channels in the input feature map during the training process. An indicator named mean square deviation is introduced to monitor the performance of a particular model segment in the classification task on the two out of five ECG classes. The data in the MIT-BIH arrhythmia database is used and a series of control experiments is conducted. Utilizing both leads of the ECG signals as input to the neural network classifier can achieve better classification results than those from using single channel inputs in different application scenarios. Models embedded with the channel-wise attention structure always achieve better scores on sensitivity and precision than the plain Resnet models. The proposed model exceeds most of the state-of-the-art models in ventricular ectopic beats (VEB) classification performance and achieves competitive scores for supraventricular ectopic beats (SVEB). Adopting more lead ECG signals as input can increase the dimensions of the input feature maps, helping to improve both the performance and generalization of the network model. Due to its end-to-end characteristics, and the extensible intrinsic for multi-lead heart diseases diagnosing, the proposed model can be used for the realtime ECG tracking of ECG waveforms for Holter or wearable devices.
References | Related Articles | Metrics
Multi-Model Ensemble Deep Learning Method to Diagnose COVID-19 Using Chest Computed Tomography Images
WANG Zhiming(王志明), DONG Jingjing (董静静), ZHANG Junpeng∗ (张军鹏)
2022, 27 (1):  70-80.  doi: 10.1007/s12204-021-2392-3
Abstract ( 22 )   PDF (1418KB) ( 8 )  
Deep learning based analyses of computed tomography (CT) images contribute to automated diagnosis of COVID-19, and ensemble learning may commonly provide a better solution. Here, we proposed an ensemble learning method that integrates several component neural networks to jointly diagnose COVID-19. Two ensemble strategies are considered: the output scores of all component models that are combined with the weights adjusted adaptively by cost function back propagation; voting strategy. A database containing 8 347 CT slices of COVID- 19, common pneumonia and normal subjects was used as training and testing sets. Results show that the novel method can reach a high accuracy of 99.37% (recall: 0.998 1; precision: 0.989 3), with an increase of about 7% in comparison to single-component models. And the average test accuracy is 95.62% (recall: 0.958 7; precision: 0.955 9), with a corresponding increase of 5.2%. Compared with several latest deep learning models on the identical test set, our method made an accuracy improvement up to 10.88%. The proposed method may be a promising solution for the diagnosis of COVID-19.
References | Related Articles | Metrics
COVID-19 Interpretable Diagnosis Algorithm Based on a Small Number of Chest X-Ray Samples
BU Ran (卜冉), XIANG Wei∗ (向伟), CAO Shitong (曹世同)
2022, 27 (1):  81-89.  doi: 10.1007/s12204-021-2393-2
Abstract ( 21 )   PDF (1470KB) ( 6 )  
The COVID-19 medical diagnosis method based on individual’s chest X-ray (CXR) is achieved difficultly in the initial research, owing to difficulties in identifying CXR data of COVID-19 individuals. At the beginning of the study, infected individuals’ CXRs were scarce. The combination of artificial intelligence (AI) and medical diagnosis has been advanced and popular. To solve the difficulties, the interpretability analysis of AI model was used to explore the pathological characteristics of CXR samples infected with COVID-19 and assist in medical diagnosis. The dataset was expanded by data augmentation to avoid overfitting. Transfer learning was used to test different pre-trained models and the unique output layers were designed to complete the model training with few samples. In this study, the output results of four pre-trained models in three different output layers were compared, and the results after data augmentation were compared with the results of the original dataset. The control variable method was used to conduct independent tests of 24 groups. Finally, 99.23% accuracy and 98% recall rate were obtained, and the visual results of CXR interpretability analysis were displayed. The network of COVID-19 interpretable diagnosis algorithm has the characteristics of high generalization and lightweight. It can be quickly applied to other urgent tasks with insufficient experimental data. At the same time, interpretability analysis brings new possibilities for medical diagnosis.
References | Related Articles | Metrics
Enhancing Speech Recognition for Parkinson’s Disease Patient Using Transfer Learning Technique
YU Qing (余青), MA Yi (马祎), LI Yongfu∗ (李永福)
2022, 27 (1):  90-98.  doi: 10.1007/s12204-021-2376-3
Abstract ( 18 )   PDF (1087KB) ( 3 )  
Parkinson’s disease patients suffer from disorders of speech. The most frequently reported speech problems are weak, hoarse, nasal or monotonous voice, imprecise articulation, slow or fast speech, difficulty starting speech, impaired stress or rhythm, stuttering, and tremor. To improve the speech quality and assist the patient with speech rehabilitation therapy, we have proposed the speech recognition model for Parkinson’s disease patients using transfer learning technique (PSTL), where we have pre-trained the long short-term memory (LSTM) neural network model with our developed publicly available dataset that has been obtained from healthy people through the social media platform. Then, we applied the transfer learning technique to improve the performance of the PSTL framework. The frequency spectrogram masking data augmentation method has been used to alleviate the over-fitting problem so that the word error rate (WER) is further reduced. Even with a limited dataset, our proposed model has effectively reduced the WER from 58% to 44.5% on the original speech dataset and 53.1% to 43% on the denoised speech dataset, which demonstrated the feasibility of our framework.
References | Related Articles | Metrics
Application of Deep Learning Method on Ischemic Stroke Lesion Segmentation
ZHANG Yue (张月), LIU Shijie (刘世界), LI Chunlai (李春来), WANG Jianyu (王建宇)
2022, 27 (1):  99-111.  doi: 10.1007/s12204-021-2273-9
Abstract ( 19 )   PDF (944KB) ( 3 )  
Although deep learning methods have been widely applied in medical image lesion segmentation, it is still challenging to apply them for segmenting ischemic stroke lesions, which are different from brain tumors in lesion characteristics, segmentation difficulty, algorithm maturity, and segmentation accuracy. Three main stages are used to describe the manifestations of stroke. For acute ischemic stroke, the size of the lesions is similar to that of brain tumors, and the current deep learning methods have been able to achieve a high segmentation accuracy. For sub-acute and chronic ischemic stroke, the segmentation results of mainstream deep learning algorithms are still unsatisfactory as lesions in these stages are small and diffuse. By using three scientific search engines including CNKI, Web of Science and Google Scholar, this paper aims to comprehensively understand the state-of-the-art deep learning algorithms applied to segmenting ischemic stroke lesions. For the first time, this paper discusses the current situation, challenges, and development directions of deep learning algorithms applied to ischemic stroke lesion segmentation in different stages. In the future, a system that can directly identify different stroke stages and automatically select the suitable network architecture for the stroke lesion segmentation needs to be proposed.
References | Related Articles | Metrics
Progress and Perspective of Artificial Intelligence and Machine Learning of Prediction in Anesthesiology
XIA Ming (夏明), XU Tianyi (徐天意), JIANG Hong∗ (姜虹)
2022, 27 (1):  112-120.  doi: 10.1007/s12204-021-2331-3
Abstract ( 20 )   PDF (409KB) ( 4 )  
Artificial intelligence (AI) has long been an attractive topic in medicine, especially in light of the rapid developments in digital and information technologies. AI has already provided some breakthroughs in medicine. With the assistance of AI, more precise models have been used for clinical predictions, diagnoses, and decision-making. This review defines the basic concepts of AI and machine learning (ML), and provides a simple introduction to certain frequently used algorithms in AI and ML. In addition, the review discusses the current common applications of AI and ML in the prediction of anesthesia conditions, including those for preoperative predictions of difficult airways, intraoperative predictions of adverse events and anesthetic effects, and postoperative predictions of vomiting and pain. The use of AI in anesthesiology remains in development, even without extensive promotion and clinical application; moreover, it has immense potential to maintain further development in the future. Finally, the limitations and challenges of AI development for anesthesia are also discussed, along with considerations regarding ethics and safety.
References | Related Articles | Metrics
Panoramic and Personalised Intelligent Healthcare Mode
LIU Quanchen (刘权宸), ZHANG Pengzhu∗ (张鹏翥)
2022, 27 (1):  121-136.  doi: 10.1007/s12204-021-2274-8
Abstract ( 14 )   PDF (1453KB) ( 3 )  
Although the development of national conditions and the increase in health risk factors undoubtedly pose a huge challenge to China’s medical health and labour security system, these simultaneously promote the elevation and transformation of national healthcare consciousness. Given that the current disease diagnosis and treatment models hardly satisfy the growing demand for medical and health care in China, based on the theory of healthcare and basic laws of human physiological activities, and combined with the characteristics of the information society, this paper presents a panoramic and personalised intelligent healthcare mode that is aimed at improving and promoting individual health. The basic definition and conceptual model are provided, and its basic characteristics and specific connotations are elaborated in detail. Subsequently, an intelligent coordination model of daily time allocation and a dynamic optimisation model for healthcare programmes are proposed. The implementation of this mode is explicitly illustrated with a practical application case. It is expected that this study will provide new ideas for further healthcare research and development.
References | Related Articles | Metrics