Khasnobish Anwesha, Konar Amit, Tibarewala D N, Nagar Atulya K
IEEE Trans Neural Syst Rehabil Eng. 2017 Jan;25(1):88-102. doi: 10.1109/TNSRE.2016.2580580. Epub 2016 Jun 14.
In visual-motor coordination, the human brain processes visual stimuli representative of complex motion-related tasks at the occipital lobe to generate the necessary neuronal signals for the parietal and pre-frontal lobes, which in turn generates movement related plans to excite the motor cortex to execute the actual tasks. The paper introduces a novel approach to provide rehabilitative support to patients suffering from neurological damage in their pre-frontal, parietal and/or motor cortex regions. An attempt to bypass the natural visual-motor pathway is undertaken using interval type-2 fuzzy sets to generate the approximate EEG response of the damaged pre-frontal/parietal/motor cortex from the occipital EEG signals. The approximate EEG response is used to trigger a pre-trained joint coordinate generator to obtain the desired joint coordinates of the link end-points of a robot imitating the human subject. The robot arm is here employed as a rehabilitative aid in order to move each link end-points to the desired locations in the reference coordinate system by appropriately activating its links using the well-known inverse kinematics approach. The mean-square positional errors obtained for each link end-points is found within acceptable limits for all experimental subjects including subjects with partial parietal damage, indicating a possible impact of the proposed approach in rehabilitative robotics. Subjective variation in EEG features over different sessions of experimental trials is modeled here using interval type-2 fuzzy sets for its inherent power to handle uncertainty. Experiments undertaken confirm that interval type-2 fuzzy realization outperforms its classical type-1 counterpart and back-propagation neural approaches in all experimental cases, considering link positional error as a metric. The proposed research offers a new opening for the development of possible rehabilitative aids for people with partial impairment in visual-motor coordination.
在视觉-运动协调中,人类大脑在枕叶处理代表复杂运动相关任务的视觉刺激,为顶叶和前额叶生成必要的神经元信号,进而生成与运动相关的计划,以激发运动皮层执行实际任务。本文介绍了一种新颖的方法,为前额叶、顶叶和/或运动皮层区域遭受神经损伤的患者提供康复支持。利用区间二型模糊集尝试绕过自然视觉-运动通路,从枕叶脑电信号生成受损前额叶/顶叶/运动皮层的近似脑电响应。该近似脑电响应用于触发预训练的关节坐标生成器,以获得模仿人类受试者的机器人连杆端点的期望关节坐标。在此,机器人手臂用作康复辅助工具,通过使用众所周知的逆运动学方法适当激活其连杆,将每个连杆端点移动到参考坐标系中的期望位置。对于所有实验对象,包括部分顶叶受损的对象,每个连杆端点获得的均方位置误差都在可接受范围内,这表明所提出的方法在康复机器人技术中可能具有影响。本文使用区间二型模糊集对不同实验试验阶段脑电特征的主观变化进行建模,因为其具有处理不确定性的内在能力。所进行的实验证实,考虑连杆位置误差作为度量标准,区间二型模糊实现方法在所有实验情况下均优于其经典一型对应方法和反向传播神经方法。所提出的研究为开发针对视觉-运动协调部分受损人群的可能康复辅助工具开辟了新途径。