Background: Hemiplegia is one of the main reasons why stroke survivors lose their walking and balancing ability. Many studies point out that cycling is an effective means for lower limb rehabilitation. However, during training, the unaffected limb may compensate for the affected one resulting in suboptimal rehabilitation. To address this issue, the investigators developed the virtual reality-cycling training system (VRCTS) which can acquire force and speed signals in real-time through a cycling module. The system then analyzes the acquired data and uses a 3D VR rehabilitation program to help patients to train their affected side. The aim of the study was to develop the VRCTS, verify its function and test system function on both normal subjects and stroke patients.

Methods: In this system, the investigators designed a cycling device that is embedded with load cell and encoder sensors to detect cycling force and angle in real-time. A Cycling Graph User Interface Control and Data Recode System (Cycling CR System) was applied for signal analysis and feedback control. The investigators designed a 3D interactive VR rehabilitation program that can guide and train the users through visual feedback. Each user performed a pre-test to examine determine condition, left-right balance and other parameters, which allows the system to be customized.

Inclusion Criteria:

• single stroke with unilateral hemiplegia (Brunnstrom stage III);
• hypertonia in the affected leg, modified Ashworth scale (MAS) grade ≦ 2;
• no significant perceptual, cognitive, or sensory deficits;
• no history of osteoarthritis, severe cardiopulmonary disease, or vascular disease in lower limbs
• lack of a fixed contracture in the affected lower limb