Numerous medical interventional procedures, such as brachytherapy, biopsy, radio frequency based thermal ablation, and anesthetic drug delivery, require precise placement of needles. Most of these interventions are performed percutaneously. However, precise placement of the needle to a specific target inside human body is a difficult task. Furthermore, the involved needles are usually rigid and straight, and real-time sensory feedback of the needle is rarely available. Medical imaging and interventional delivery systems have undergone significant growth in the recent years, but the development of needles for percutaneous interventions for clinical amiability is undergoing a slow progress. Also, most of the clinically used needles are several decades old in design. Controlling the needle to reach the target precisely through a desired trajectory by the guidance of quantitative sensory feedbacks could benefit various medical procedures. Accurate and safe navigation of the needle to the desired target is the major challenge in needle based interventional techniques. Needle geometry and lack of proper actuation of the needle contribute to this major challenge.The overall system integration and control of the needling device involve three loops as shown in Fig. 1. The inner loop comprising the flexible active needle actuated by the shape memory alloy (SMA); the next outer loop of the 5 degrees of freedom robotic system guiding the needle to reach the target with high accuracy and precision; and finally, the outer most human-in-the-loop involving the physician for detecting the target and optimally planning the trajectory to reach the target.To avoid the physiological obstacle, a self-actuating flexible needle has been developed [1,2]. The developed self-actuating flexible needle deployed in our experimental work of trajectory tracking task is shown in Fig. 2. It is a 17 gage needle of length 200 mm. The SMA wire is of dimension 0.24 mm in diameter and 70 mm in length.During the movement of the needle in the soft phantom, the needle tip is detected in real-time using the gray scale value pyramid pattern matching algorithm in labview (NI, Dublin, Ireland) using the noisy ultrasound (U.S.) images (Fig. 3).The U.S. transrectal transducer (BK Medical, Peabody, MA) is utilized to detect the needle tip using a DVI2USB 2.0 frame grabber (Epiphan, Palo Alto, CA). The sampling time in obtaining the U.S. imaging feedback is 5 ms.The flexible needle and the guiding robot are controlled by discrete-time proportional integral derivative (PID)-P3 and PID control algorithms with the U.S. imaging feedback.The equation of the discrete-time PID-P3 controller is given by(1)u(k)=u(k-1)+KP[e(k)-e(k-1)] ×KPhTIe(k)+KPTD[ef(k)-2ef(k-1) +ef(k-2)]+KT[e(k)-e(k-1)]3where h is the time step, e(k) = xd(k) – x(k) is the error at kth instant; ef(k) is the low pass filtered error given by ef(k)=[τf/(τf+TS)]ef(k-1)+[TS/(τf+TS)]e(k); u(k) is the controller output at kth instant; KP, KPh/TI, and KPTD are the proportional, integral, and derivative gains, respectively.Similarly, the discrete-time PID controller is given by(2)u(k)=u(k-1)+KP[e(k)-e(k-1)] ×KPhTIe(k)+KPTD[ef(k)-2ef(k-1)+ef(k-2)]The discrete controller output u(k) is hard limited to the range 0–10 V.The curvilinear trajectory tracked by the coordinated movement of the flexible needle and the robot with U.S. imaging feedback is shown in Fig. 4(a). The corresponding error norm plot is shown in Fig. 4(b). Similarly, Fig. 5 shows the tracking results with electromagnetic (EM) sensor feedback.The coordinated control performance in trajectory tracking with U.S. imaging feedback is comparable with the EM sensory feedback based performance. The root mean square error (RMSE) values in the curvilinear trajectory tracking task with these different feedback modalities are given in Table 1. The difference in accuracy between these feedback modalities is clinically acceptable (about 0.12 mm versus 0.1 mm). The results reveal that U.S. imaging feedback based coordinated robot-needle control can be performed with in vivo tissues in real-time. The near future work involves the inclusion of the human-in-the-loop to supervise the coordinated control task.This study is fully supported by the U.S. Department of Defense (Grant Nos. W81XWH-11-1-0397/98/99).
Read full abstract
Aug 22, 2015
Світлана Ігорівна Шаповалова + 1
Open Access
