IROS2019国际学术会议论文集A Flexible Sensor for Suture Training

Figure 1. Concept of suture simulator with the sensor for quatitative feedback in real time. Abstract Hospitals are required to train their medical staff to help improve their working conditions and provide quality health care for patients. Therefore, many types of research on medical training simulators, especially suturing skills, have been developed; however, because of drawbacks such as inconvenience and complicated configuration, most of the trainees do not use the developed simulator. We propose a new concept of a flexible sensor for suture training, which can be embedded into an artificial skin pad to measure the piercing location on a 2D plane. Using the characteristics of conductive films, and the linear relationship between distance and output resistance, we developed a flexible suture sensor to estimate the piercing position. Then, the experiments for characterization were conducted to assess the measurements, and the standard deviations obtained in the x and y coordinates were 0.57-1.95mm, 0.4-1.94mm respectively. Finally, we demonstrate the feasibility of the proposed sensor and discuss its performance and limitations. I. INTRODUCTION There has been an increased demand for a simulator-type training system for the monitoring of trainees surgical motions, as improving proficiency in a clinical training process has a significant impact on the application of medical skills in the operating room 1-8. In the past, experts observed the trainees, evaluated their skills, and then provided feedback. However this method not only makes a subjective diagnosis, but also evaluates each individual, and consequently, it takes a lot of time for apprentices to get an accurate diagnosis 9. Therefore, in recent years, surgical training tools have been developed to objectively assess the competence of apprentices and provide instant feedback on their performance 4. Development of medical skills, especially suturing, requires proper training 2. Recently, many research groups have developed suture training simulators that provide motion 10-14, force 15-18, image 19-21 and VR based feedback 22, 23. However, most of these simulators are not commercialized for several reasons. For example, motion-tracking devices and forced feedback systems evaluate the intermediate steps, but not the end-product quality. Therefore, trainees are scored only in their motion-tracking assessment, so they are likely to be negligent about the quality of their product. Researchers have been conducted to focus on the results and evaluate sutures based on images. There are several differences in the Woonjae Choi and Bummo Ahn are with the Robotics and Virtual Engineering, Korea University of Science and Technology, Daejeon, 34113, Republic of Korea and with the Robotics R fax: +82-31-8040-6870; e-mail: bmahnkitech.re.kr). stitch geometry on 2D which can be distinguished between the experts performance and that of the novices (e.g. mean number of stitches, length of stitches, stitch orientation, and so on). These geometric information are the basis for evaluating the end-product quality and trainees can improve their skills through visual and quantitative feedback by comparing their differences intuitively 20. However, it is inconvenient to take pictures with a camera to obtain proper angle images. Furthermore, the camera needs to be placed at an appropriate position for measuring image data in real time 21, which makes the developed systems even more complicated. In the case of a VR simulator, trainees can receive objective feedback in real time through the virtual image, but it is difficult to use it as a suture training tool because important factors such as the tactile and kinesthetic sensation are excluded. For these reasons, most of the trainees mainly use conventional simulators such as artificial skin, mannequins, and animals that improve only sensory skills. To address these limitations, we developed a flexible sensor that can be embedded into the artificial skin simulator. The sensor can quantitatively provide needle-piercing positions in real time, allowing trainees to receive objective and immediate feedback as shown in Fig.1. With these position data, the simulator can obtain geometric information of stitch configurations (e.g. stitch length and orientation). In addition, it can also save the most important sensory training of surgical skill 24, 25 as it is attached to the artificial skin. Therefore, it is possible to provide a comfortable and an effective training environment to the A Flexible Sensor for Suture Training Woonjae Choi, Student Member, IEEE and Bummo Ahn, Member, IEEE IEEE Robotics and Automation Letters (RAL) paper presented at the 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) Macau, China, November 4-8, 2019 Copyright 2019 IEEE trainee unlike any other suture training simulators. The sensor consists of three layers (two driving layers and one sensing layer) and is kept isolated until penetration, so that the electric potential can be properly measured. Each conductive layer was made by spreading the conductive ink on the cloth tape. To determine the voltage gradients applied to each conductive layer, we measured the electric potential at both three endpoints and averaged them. With these measurements, the sensor modules were tested by piercing the 25 points to demonstrate the ability to estimate the piercing positions Section II of this letter describes the characteristics, design, and fabrication of the sensor. Section III covers the sensors characterization and experiments results, respectively. Section IV presents an application of the proposed method, and we finally discuss the performance and limitation of the sensor in Section V along with the conclusion in section VI. II. SUTURING SENSOR A. Characteristics of conductive materials A conductive layer (driving layer) is energized (the edges operate at high and low voltages) to localize the piercing points of the suture sensor. The sheet, which consists of homogeneous conductive material, has the following resistance properties by Pouillets law: R=L/A; where , L, and A are the resistivity, length and cross-sectional area of the resistive material respectively 26. Thus, voltage gradient is linearly formed across the layer as shown in Fig.2. When the sheet is pierced by needle (conductive tool), the sensor sheet can be considered as two resistors that are serially connected at the piercing point (like the way a potential divider takes advantage of the resistors connected in series). As the electric potential varies across the layer, the insertion position can be estimated by relating it to the value of electric potential on the sheet 26. To convert the values of electric potential (V, sensor outputs) into the position values, the electric potential values of both edges of the conductive material (V and V) are required to derive an equation for reference voltage gradients. In reality, there are small voltage drops on the electrodes owing to contact resistances (in the experiment, voltage drops because of the transistor). Theoretically, because the remaining voltage drops on the conductive sheet linearly across the sensor layer (Fig. 2), the voltage values at both ends of the sensor can be used to obtain a correlation between voltage and position. 1+ 1 Where is the electric potential at the piercing point in the sensor and 1 is the distance from the electrode Figure 2. Characteristics of conductive materials. The blue line indicates the voltage gradient on the conductive materials in the ideal case. At both edges, there are additional voltage drops caused by other components in the circuit. Figure 4. Electrical schematic of the sensor module transferring electric potential from the driving layer to the sensing layer for measuring 1D coordinates. ( are the resistances of conductive layer segment, , is contact resistance, is needle resistance, is the resistance on the sensing layer, and is the resistance of the input impedance of the DAQ ) R1 R2 Rn Ri Rc2 Rs Rc1 Sensor output -+ Figure 3. Design of the Suturing Sensor composed of three conductive layers and two dielectric layers. Figure 7. Experimental setup for the sensor characterization. Figure 6. Fabrication process: (a) Spreading the conductive ink to cloth tape evenly, (b) Attaching the copper tape to both side edges, (c) Complete prototype of flexible suturing sensor module. calculated by substituting the measured value into equation (2). B. Design The 2D suture sensor has three conductive and two insulation layers. The conductive layers were used for driving the x (one layer) and y (one layer) axis voltage and sensing (one layer) both voltages. The insulation layers were used for insulating the conductive layers of the sensor. Fig. 3 shows the sensor design concept. The conductive layer for sensing of output voltages is placed under the driving layers along the x- and y- axis. In addition, all of the layers are insulated by the dielectric layer. This sensing layer is responsible for transferring the electric potential to an analog input terminal with a high impedance. Fig. 4 shows an electrical schematic of the sensor module for measuring the electric potential at a 1D piercing position with the sensing layer. If the needle is inserted through the layers completely, it creates a path to the sensing layer, and establishes the same voltage as the remaining electric potential on the driving layer. This is applied to the sensing layer when each electrode is connected to voltage and ground sources respectively. As micro current flows through the sensing layer owing to the high impedance (R10G ohm), there will be negligible voltage drops on the needle and the sensing layer (RR,R). As this voltage is proportional to the one in the driving layer, the piercing position can be calculated by measuring the voltage in the sensing layer. For measuring the two-dimensional (2D) coordinate of the piercing-point, another driving layer is added, with each electrode perpendicular to those of the existing driving layer. The three sheets of the conductive layer (two driving layers and one sensing layer) are separated by two thin insulating layers (placed between each conductive layer). All the conductive layers should be disconnected before the needle is inserted. To measure the electric potential on the two driving layers while using one sensing layer, we employed an alternative electric switching mechanism, as shown in Fig. 5. Once, the one driving layer is energized to establish a voltage gradient, the second driving layer is disabled and set to high-impedance mode simultaneously. Subsequently, they are alternated at a sufficient rate of 200 Hz, repetitively. An electronic oscillator is used to create square waveforms with different phases. Owing to the different phases of waveforms, the current is periodically cut off at both the voltage and ground sources of the two driving layers. Therefore, in the active layer, the piercing points can be measured on the same principle as the 1D measurement without any disturbance caused by the other driving layer, which is inactive. The position points for each axis can be obtained at 100 Hz (which is a sufficient rate for measuring piercing holes in suture training). C. Fabrication The conductive layer is made by spreading the conductive material (SKU-0209, Bare Conductive, UK) evenly across the cloth tape (2764, 3M, USA), and two copper tapes (TaeHwa, Korea) are attached on the edges on both sides (all Figure 5. Circuit diagram of sensing 2D coordinates applying alternative electrical switching mechanism with transistors. Vs y y x x y direction oscillator x direction oscillator Sensor output Figure 8. Electric potential at both end points (Sensor module 1): (a) x-coordinate reference values, (b) y-coordinate reference values. edges in the sensing layer). As the adhesive part of the copper tape used as the electrodes is low in conductivity, the opposite part should be attached to the edges of the conductive material. The suturing sensor module is fabricated base on the presented design as shown in Fig. 6. The electrodes (copper tape) are attached to left and right edges of one driving layer as well as the top and bottom edges of the other driving layer. These layers can be arranged in any order; however, the sensing layer should be placed under the depositions with electrodes around the edge. The thin films (OPSITE, Smith and Nephew, UK) as the dielectric layers are laid between these three conductive layers to isolate each other. Fig. 6(c) shows the fabricated prototype. III. SENSOR CHARACTERIZATION A. Experimental Setup An experimental setup was prepared as shown in Fig. 7. During the experiments, the sensor is connected to the circuit (the voltage source was set to 5V), including the transistor, and alternatively activated at rate of 200 Hz through DAQ-NI (data acquisition board USB-6343, National Instrument, U.S.A.) and a DAQ software package (Labview, National Instruments, U.S.A.). The DAQ-NI not only measures the sensor output, but also forms the electric switching mechanism by creating square waves with different phases using Labview. The suturing sensor prototypes were placed on a soft rubber to make it easier for the needle to pass through the sensor. Piercing guide is also prepared to precisely guide the needle to reference position. B. Reference Voltages at Both Ends of Sensor The reference voltages (V and V) were measured to apply the linearity of voltage gradients to the six different sensor modules. We measured the voltage values at both endpoints (with six different sensor modules), and calculated the averages of the values at each 3 points to estimate the piercing positions. Fig. 8 shows the measurements and average values of the V and V at both endpoints (3 points respectively) in only one sensor module (X and Y sheets), and the average values and the standard deviations of 12 layers (X: 6 sheets, Y: 6 sheets) are shown in Table 1. The electric potential values at both ends were shown to vary from sensor to sensor. The error was mainly caused by the fabrication process, and it varied from 0.01 to 0.04 in one sensor. Thus, the average of the values at three points is regarded as the endpoint of each sensor seat, and the linear reference voltage gradients are applied to those seats. Table 1. The average values and the standard deviations of 12 layers (X: 6 sheets, Y: 6 sheets) Sensor Average Values at , , (V) SD at , , (V) X Y X Y Vr Vg Vr Vg Vr Vg Vr Vg Module 1 3.77 0.23 3.85 0.15 0.04 0.01 0.01 0.012 Module 2 3.83 0.35 3.9 0.51 0.04 0.03 0.02 0.03 Module 3 3.65 0.46 3.61 0.42 0.01 0.01 0.01 0.01 Module 4 3.75 0.25 4 0.23 0.01 0.01 0.01 0.00 Module 5 3.85 0.22 3.92 0.43 0.03 0.02 0.03 0.04 Module 6 4.02 0.23 4.01 0.2 0.02 0.02 0.01 0.03 C. Characterization results We evaluated our sensor with twelve different gradient references from the six sensor module prototypes to confirm the characteristics and estimate the position pierced by the needle in the X and Y coordinate system. The piercing guide is used to predefine 25 piercing points with the same interval (10mm) as shown in Fig. 9 (a). While placing the guide on the sensor, we pierced 25 points (five times per each sensor module) to test its repeatability and validation. The electric potential values at the piercing point were then converted into the position values. Fig. 9(b) shows the output voltage distribution of the sensor and the real position points (red dots) corresponding to the predefined points and the calculated position values (the averages of the five measurements at the same points). The deviation of the five measurements in each sensor is less than 0.05 Volt, which is less than approximately 1 mm, calculated as distance. Table 2 shows the average values calculated from all the 25 positions. Compared with the actual positions at 25 points, the errors of estimated values for the X and Y coordinates are 0.004.08 mm, 0.0143.13 mm respectively (Sensor module 3 and 4 were shifted more than other modules.). Table 3 shows the standard deviations of each of the 25 positions in which X and Y coordinates are 0.571.95mm, 0.41.94mm respectively. Table 2. Average of experimental piercing position values of the six sensor modules calculated by the reference voltage gradients Average(mm) X(mm) Y(mm) 10 x y 20 x y 30 x y 40 x y 50 x y 50 9.71 50.37 19.66 50.00 30.17 49.82 40.80 49.98 50.60 49.06 40 9.86 40.17 20.26 39.78 30.29 39.95 40.71 39.71 49.93 39.53 30 10.02 30.23 20.06 29.73 30.09 30.51 40.35 29.90 50.00 29.41 20 10.14 19.58 20.22 19.72 29.92 19.92 40.46 19.37 50.23 19.51 10 9.38 9.44 20.38 9.88 29.44 9.56 40.61 10.14 50.10 9.87 Figure 9. Characterization experimental results to evaluate the : (a) Experimental procedure of characterization using piercing guide, (b) Result for 2D suturing sensor with piercing points corresponding 25 predefined positions. Table 3. The standard deviations of the position measurements of the six sensor modules. Standard Deviations(mm) X(mm) Y(mm) 10 x y 20 x y 30 x y 40 x y 50 x y 50 0.57 1.41 1.55 1.94 1.95 1.35 1.29 1.00 1.25 0.67 40 0.86 1.26 1.20 1.09 1.88 1.38 1.04 1.12 1.73 0.52 30 0.99 0.45 1.18 0.84 1.65 0.52 1.19 0.61 1.48 0.61 20 1.10 0.88 0.72 1.40 1.65 1.05 0.99 0.86 1.18 0.40 10 0.84 0.77 1.02 1.62 1.12 1.01 0.92 1.26 1.10 1.22 (a) (b) (c) Figure 10. (a)Suture tools for knot-tying practice, (b) Suture bites using a needle composed of an insulated and uncoated segment, (c) The actual final-product