Purpose: To develop and evaluate a controller system capable of accurately guiding a shape memory alloy (SMA) actuated flexible needle to deliver therapeutic agents using a curvilinear approach for prostate brachytherapy for improved geometrical and dosimetric conformity. Methods: An experimental needling system was developed using flexible cannula of a 17‐gauge HDR needle actuated by Nitinol SMA. Experiments were performed in air at room temperature (22‐degC) and in controlled warm water‐bath (38‐degC, which is close to human body temperature). A silicone‐based coating was applied to the SMA wire‐actuator for electrical insulation in water. Two closed‐loop control algorithms, proportional‐derivative‐integral (PID) and PID‐P3 (with cubic term of positional errors), were implemented and tested using LabVIEW (National Instruments, version‐4.3.3). The needle and SMA actuator were integrated with a 5‐DOF Aurora (NDI, Ontario, Canada) electromagnetic (EM) sensor embedded at the tip of the needle for trajectory tracking. Performance was evaluated in terms of EM‐sensor feedback sampling intervals of 100ms, 500ms, and 1000ms, and needle‐tip displacements of 0.5mm, 2mm, and 2.5mm. Controller gains were experimentally determined. Results: PID‐P3 controller outperformed PID controller in all measured domains. Sampling time of 100ms yielded unstable performance for all displacements in water. Sampling time of 500ms with needle‐tip displacement of 2mm yielded the optimal performance with smaller overshoot and higher stability. In‐water results were comparable to in‐air measurements for the optimum performance at 500ms sampling time and 2mm displacement. Conclusions: PID‐P3 based SMA‐actuated needle appears to be adequate for conforming to the radius of curvature of a prostate gland, as well as operating effectively in human body. For small displacements (<0.5mm), fast rise‐time was observed, however, stability suffered. For larger displacements (>=2mm), stability increased but was hampered by slower rise‐time. Further investigation of PID‐P3 and other control algorithms (neuro‐fuzzy and adaptive) may offset slow rise‐time and increase stability. This study was funded by the U.S. Army Medical Research and Materiel Command(Garnt #W81XWH‐11‐1‐0399).
Read full abstract