The purpose of this study was to compare the relative accuracy of a range of computer-based analysis with respect to EMG onset determined visually by an experienced examiner. Ten healthy students (6 male, 4 female) were recruited and three times randomly selected trials of isometric contraction of wrist flexion and extension were evaluated using four technique. These methods were compared which varied in terms of EMG processing, threshold value and the number of samples for which the mean must exceed the defined threshold, and beyond 7% of maximum amplitude. To identify determination of onset time, ICCs(Intraclass Correlation Coefficients) was used and inter-rater arid intra-rater reliability ranged good in visually derived onset values. The results of this study present that in wrist flexion and extension, the reliability of the inter and intra-examiner muscle contraction onset times through visual analysis showed beyond .971 with ICCs. The reliability of the muscle contraction onset time decision through visual reading, tested with computer analysis, showed a relationship of all the selected analysis methods with ICCs .859 and .871. The objective computer-based analysis comparing with visual reading at the same time is the effective and qualitative data analysis method, considering the specificity of each study method.

In this paper, the prototype of surface EMG (ElectroMyoGram) sensor is developed for the robotic rehabilitation applications, and the developed sensor is composed of the electrodes, analog signal amplifiers, analog filters, ADC (analog to digital converter), and DSP (digital signal processor) for coding the application example. Since the raw EMG signal is very low voltage, it is amplified by about one thousand times. The artifacts of amplified EMG signal are removed by using the band-pass filter. Also, the processed analog EMG signal is converted into the digital form by using ADC embedded in DSP. The developed sensor shows approximately the linear characteristics between the amplitude values of the sensor signals measured from the biceps brachii of human upper arm and the joint angles of human elbow. Finally, to show the performance of the developed EMG sensor, we suggest the application example about the real-time human elbow motion acquisition by using the developed sensor.