Cerebral palsy is a neurological disorder which caused by a brain injury or developmental malfunction of the brain. Cerebral palsy may cause many adverse results such as movement and postural stability problems. One of the medical conditions that may occur with such problems is scoliosis. There are few treatment options, one of the common methods is the use of a back brace. The patients can be more active when they wear the brace. The purpose of our research is to clarify the difference between the conditions with and without brace focusing on the difference of surface electromyography (s- EMG) activities. In many cases, however, s-EMG signal is too weak to analyse in typical way. In this paper, we discuss the filtering method of the weak s-EMG signals. We use a notch filter whose cut-off frequencies are determined by fast Fourier transform of the raw signals. We show that the filter does not make strong influence for the features of s-EMG signals to be able to discuss the differences between the conditions which are with and without wearing a back brace.