위치 기반 비주얼 서보잉을 이용한 비침습형 뇌 자극 로봇 제어

Abstract

최근 비침습적으로 뇌의 특정 부위를 자극하여 뇌의 기능을 연구하거나 뇌 질환을 치료하는 비침습적 뇌자극 방법이 뇌연구자와 임상의를 중심으로 주목받고 있다. 기존의 침습적 방법에 비해, 뇌자극의 정확도 및 강도는 떨어지나, 수술에 따른 위험부담이나 부작용이 거의 없는 뛰어난 장점을 지니고 있다. 우울증 치료나 뇌졸중 치료에 사용될 수 있다고 알려져 있다. 대표적인 비침습형 뇌자극으로는, 전자기유도를 통해 뇌자극을 하는 경두개자기자극(Transcranial Magnetic Stimulation)과, 최근 등장한 집중 초음파를 통해 뇌자극을 하는 경두개초음파자극(Transcranial Ultrasound Stimulation)이 있다. 이러한 비침습형 뇌자극에서 눈에 보이지 않는 뇌의 특정부위를 정확히 자극하는 것은 매우 어려운 문제이다. 이러한 문제점을 해결하기 위해, 의료 영상을 이용하여 현재의 자극 위치를 그래픽하게 보여주는 뉴로네비게이션(Neuronavigation)이 제안되기도 하였다. 하지만 이러한 네비게이션만으로는 포지셔닝 문제가 완전히 해결될 수 없었다. 왜냐하면, 수 kg에 달하는 뇌자극기를 특정 위치와 방향에 정확히 위치시켜 고정하는 것은 현실적으로 매우 어렵기도 할뿐더러, 더욱이 자극 중에 환자의 움직임이 발생하면 어렵게 찾은 위치와 방향도 틀리게 되기 때문이다. 본 연구에서는, 이러한 문제를 해결하기 위해 뇌자극기를 정확한 위치와 방향에 위치시키고 환자의 움직임을 보상할 수 있는 6-자유도 로봇 기반 뇌자극 시스템을 제안한다. 제안한 시스템은 환자와 로봇 말단부에 부착된 마커를 통해 환자와 자극기의 위치를 추정하고, 이 상대 오차를 다시 피드백하여 최종오차를 줄여나가는 위치기반 영상 제어(Postion based Visual Servoing) 기법을 사용한다. 고정된 자세(위치 및 방향)에 대해, 시스템의 최종 오차는 평균 0.046mm(위치), 0.013rad(방향)으로 나타났다. 또한, 움직이는 환자를 가정한 실험을 통해, 환자의 움직임 보상하는 것을 확인하였다.| Recently, brain researchers and clinicians have focused on non-invasive brain stimulation methods which stimulate specific areas of the brain to study brain fuctions and treat brain diseases. Compared to conventional invasive methods, the accuracy and intensity of brain stimulation are low, but they have the advantage of having little risk of surgery or side effects. It can be used to treat depression or stroke. Typical non-invasive brain stimulation includes transcranial magnetic stimulation, which stimulates the brain through electromagnetic induction, and transcranial ultrasound stimulation, which stimulates the brain through recently introduced intensive ultrasound. It is very difficult to precisely stimulate certain areas of the brain that are invisible to these non-invasive brain stimulation. In order to solve such a problem, neuronavigation has been proposed. It is used to graphically show the current stimulation position in medical image. However, it could not completely solve the positioning problem. Because it is very difficult to fix a heavy stimulus exactly in a certain position and orientation. And if the patient's movement occurs during stimulation, the pose must be changed. In this study, we propose a brain stimulation system based on a 6-dof robot that can position the stimulus in the correct pose and compensate the patient's movement. The proposed system uses position-based visual servoing which estimates the pose of the patient and stimulus through the marker attached to the patient and the end-effector and feedbacks the relative error to reduce the final error. For a fixed pose, the final error of the system was 0.046mm and 0.013rad, respectively. In addition, it is confirmed motion compensation through the experiment in which the moving patient is assumed.; Recently, brain researchers and clinicians have focused on non-invasive brain stimulation methods which stimulate specific areas of the brain to study brain fuctions and treat brain diseases. Compared to conventional invasive methods, the accuracy and intensity of brain stimulation are low, but they have the advantage of having little risk of surgery or side effects. It can be used to treat depression or stroke. Typical non-invasive brain stimulation includes transcranial magnetic stimulation, which stimulates the brain through electromagnetic induction, and transcranial ultrasound stimulation, which stimulates the brain through recently introduced intensive ultrasound. It is very difficult to precisely stimulate certain areas of the brain that are invisible to these non-invasive brain stimulation. In order to solve such a problem, neuronavigation has been proposed. It is used to graphically show the current stimulation position in medical image. However, it could not completely solve the positioning problem. Because it is very difficult to fix a heavy stimulus exactly in a certain position and orientation. And if the patient's movement occurs during stimulation, the pose must be changed. In this study, we propose a brain stimulation system based on a 6-dof robot that can position the stimulus in the correct pose and compensate the patient's movement. The proposed system uses position-based visual servoing which estimates the pose of the patient and stimulus through the marker attached to the patient and the end-effector and feedbacks the relative error to reduce the final error. For a fixed pose, the final error of the system was 0.046mm and 0.013rad, respectively. In addition, it is confirmed motion compensation through the experiment in which the moving patient is assumed.