In principle, the T wave results from the differences in durations of
action potentials (AP) of different ventricular regions. Based on this
concept, a mathematical model has been developed that represents the
TU wave morphology as a summation of four AP-like functions: TU = S1 -
S2 + L1 - L2. The sigmoidal shape of AP-like curves is produced by Hi
ll's equation V(t) = a . t(n)/(b(n) + t(n)). Each of the decomposition
functions is characterized by two parameters: the amplitude at the be
ginning of QRS (A(max)), and the duration at 5% of A(max) (D-95). The
set of four decomposition functions leads to eight parameters that pro
vide detailed characteristics of the TU wave morphology. The model was
validated using 170 TU wave complexes recorded digitally in leads V-2
-V-6 from 22 normal subjects and 12 patients with abnormal TU wave mor
phologies (negative, biphasic, and notched T waves). The electrocardio
graphic signals were sampled at 100 Hz and a best-fit procedure was us
ed to obtain the decomposition. In all cases the coefficients of corre
lation between original TU patterns and their mathematical models were
greater than or equal to 0.99. The mean absolute difference between t
he observed and modeled values of the TU patterns was similar in cases
with normal and abnormal TU wave morphologies (4.65 +/- 0.41 mu V vs
5.19 +/- 0.48 mu V, respectively) demonstrating that the model is capa
ble of describing and categorizing various TU patterns by a set of eig
ht numerical parameters.