A number of rehabilitation robots have been developed in the past two decades to aid functional recovery of impaired limbs in patients with brain injury (Volpe et al., 2000; Hesse et al., 2005; Kahn et al., 2006; Lum et al., 2006; Masiero et al., 2007; Nef et al., 2007; Coote et al., 2008; Housman et al., 2009; Chang et al., 2014). In the field of rehabilitation, high intensive, task-oriented and repetitive execution of movements is effective for functional recovery of impaired upper limbs following brain injury (Bütefisch et al., 1995; Kwakkel et al., 2004; Schaechter, 2004; Levin et al., 2008; Murphy and Corbett, 2009; Oujamaa et al., 2009). Rehabilitation robots can easily and precisely provide these labor-intensive rehabilitative treatments, and the effect of rehabilitation robots on functional recovery in patients with brain injury has been demonstrated in many studies (Volpe et al., 2000; Hesse et al., 2005; Lum et al., 2006; Masiero et al., 2007; Coote et al., 2008; Norouzi-Gheidari et al., 2012). Compared to conventional therapy (CT) provided by a therapist, the effectiveness of robot assisted therapy (RT) is questionable (Masiero et al., 2011; Norouzi-Gheidari et al., 2012). There is no difference between RT and intensive CT of the same duration/intensity condition, and extra sessions of RT in addition to CT bring better motor recovery of the shoulder and elbow (not for hand and wrist) compared with CT (Norouzi-Gheidari et al., 2012). To make the best use of robot for upper limb rehabilitation, increased efficacy of robotic rehabilitation is necessary. However, research on the optimal conditions to maximize the rehabilitative effect during treatment with a rehabilitation robot has been limited (Reinkensmeyer et al., 2007).
Brain plasticity, the ability of our brain system to reorganize its structure and function, is the basic mechanism underlying functional recovery in patients with brain injury (Schaechter, 2004; Murphy and Corbett, 2009). The underlying principle of rehabilitation in terms of brain plasticity is based on the modulation of cortical activation induced by the manipulation of external stimuli (Kaplan, 1988). Little is known about the cortical effects resulting from rehabilitation robot treatment (Li et al., 2013; Chang et al., 2014; Jang et al., 2015).
Functional neuroimaging techniques, including functional MRI (fMRI), Positron Emission Tomography (PET) and functional Near Infrared Spectroscopy (fNIRS) provide important information about the activation of the brain by external stimuli (Frahm et al., 1993; Willer et al., 1993; Miyai et al., 2001; Fujii and Nakada, 2003; Perrey, 2008; Kim et al., 2011; Leff et al., 2011; Gagnon et al., 2012). Of these, fNIRS provides a non-invasive method for measurement of the hemodynamic responses associated with activation of the cerebral cortex based on the intrinsic optical absorption of blood (Arenth et al., 2007; Irani et al., 2007; Perrey, 2008; Ye et al., 2009; Leff et al., 2011). Compared with other functional neuroimaging techniques, fNIRS has a unique advantage of less sensitivity to motion artifact and metallic material. Therefore, fNIRS appears suitable for the study of brain response during treatment with rehabilitation robots (Perrey, 2008; Mihara et al., 2010; Leff et al., 2011; Li et al., 2013; Chang et al., 2014).
In this study, we hypothesized that there exists optimal conditions for robotic rehabilitation to enhance the rehabilitative effect. The speed of movement performed by rehabilitation robot could be a unique aspect of robot rehabilitation, because varied speed can be provided consistently only with the robot. To confirm our hypothesis, using fNIRS, we examined the optimal speed of passive wrist movements performed by a rehabilitation robot that induces cortical activation through proprioceptive input by passive movements (Radovanovic et al., 2002; Francis et al., 2009; Lee et al., 2012). As a part of upper limb, the wrist enhances the usefulness of the hand by allowing it to take different orientations with respect to the elbow (van der Lee, 2001). If there exists an optimal speed that offers the greatest cortical activation, it could be applicable for robotic rehabilitation and research for other optimal conditions such as duration.
Subjects and Methods
Healthy right-handed subjects (15 males, 8 females; mean age 26.5, range 21–30) with no history of neurological, psychiatric, or physical illness were recruited for this study. Handedness was evaluated using the Edinburg Handedness Inventory (Oldfield, 1971). All subjects were fully informed about the purpose of the research and provided written, informed consent prior to participation in this study. The study protocol was approved by the Institutional Review Board of the Daegu Gyeongbuk Institute of Science and Technology (DGIST). Data from two subjects were excluded because the subjects did not follow the required instructions during the data collection.
Regarding flexion and extension only, the human wrist can be simplified as a one degree of freedom (DOF) kinematic model with one revolute joint (Zatsiorsky, 2002). As mentioned above, the wrist rehabilitation robot was designed and manufactured as a simplified kinematic model of the wrist. The robot used for wrist rehabilitation has three parts: hand, wrist joint and forearm, and provides passive movement of flexion and extension (Figure 1). It has a gear driven mechanism using a single motor. The actuation system for the wrist part is composed of DC, a brushless motor with encoder (EC-i 40, Maxon motor), harmonic drive (CSF-11-50, Sam-ik THK, gear ratio 50:1), and force-torque sensor (Mini 45, ATI). In house developed software was used to control the robot. For the real-time control, Linux Fedora 11 and the Real Time Application Interface for Linux (RTAI) Ver 3.8 systems were mounted. Real-time sensing control was achieved using an encoder and Sensoray s626 board, in which time delay control (TDC) was used for precise position control. The robot showed a position error of 0.1°–1° during the experiment.