Somatosensory evoked potentials (SEP) have been shown to be an important electrophysiological measure to assess the integrity of the spinal cord. However the peaks in the SEP waveform are often undetectable due to low signal-to-noise (SNR) ratio. Sometimes they also become indistinct during injury when the SEP flattens. Hence time-domain analysis methods are often subject to errors, and need human-expert intervention. In this paper, we propose a new technique for analyzing the shape of the evoked potentials, in which slope information is obtained for the entire signal in specific time bins. Apart from solving the problems associated with present methods, this technique has an added advantage of analyzing the SEP signal as a whole rather than simply a few peaks. The efficacy of this technique was investigated on SEP signals recorded from 12 rats before and after contusion spinal cord injury at thoracic vertebra T8. The statistical analysis results revealed significant effect of injury to hindlimbs, whereas almost none to forelimbs. Thus, the results show high potential of this technique to differentiate between normal and injured spinal cord.