An Improved QT Interval Measurement based on Singular Value Decomposition

This paper proposes an improved method of automatic ECG QT interval measurement based on the singular value decomposition (SVD) of multiple lead ECG signals. SVD separates multiple lead ECG record into orthogonal signals. Major orthogonal signals associated with high singular values are selected first for subsequent analysis. Instantaneous norm of the major three orthogonal signals are used for estimating Q wave initiating time t_Q. Two dimensional trajectory of the major orthogonal signals are utilized for T wave end time t_TE estimation. The T wave trajectory stagnates at t_TE. For the accurate t_TE estimation, this stagnation of the trajectory is proposed to be detected by its change in tangential angle. The proposed method was applied to 17 ECG data from normal subjects, patients of long QT syndrome (LQTS) and left ventricular hypertrophy (LVH) to demonstrate its effectiveness. Good consistent agreement, mean relative error of 5.01%, between estimated QT intervals and those of manual measurement by an experienced cardiologist was achieved