Life is one of the most complex nonlinear systems and heart is the core of this lifecycle system. The complexity of electrocardiogram (ECG) signals may reflect the physiologic function and health status of the heart. In this paper, we introduced two novel parameters of the multifractal mass exponent spectrum curvature and area. The evaluation of Cantor set validated that the two indicators are entirely effective in exploring the complexity of chaotic time series. Using the multiscale analysis method, we studied the mass exponent spectra of ECG signals taken from the cohorts of healthy, ischemia and myocardial infarction (MI) sufferer under different sampling frequencies and data lengths. Then we compared these new indicators with other nonlinear parameters and also expected to acquire some valuable nonlinear eigenvalues to distinguish the healthy from the heart diseased through those individual discrepancies. The classification indexes and ANOVA testing results both indicated that our method could achieve better results. These conclusions may be of much value in early diagnoses and clinical applications.