Various methods for automatic electrocardiogram T-wave detection and Q-T in
terval assessment have been developed. Most of them use threshold level cro
ssing. Comparisons with observer detection were performed due to the lack o
f objective measurement methods. This study followed the same approach. Obs
erver assessments were performed on 43 various T-wave shapes recorded: (il
with 100 mm s(-1) equivalent paper speed and 0.6 mV cm(-1) sensitivity; and
(ii) with 160 mm s(-1) paper speed and vertical scaling ranging from 0.07
to 0.02 mV cm(-1) depending on the T-wave amplitude. An automatic detection
algorithm was developed by adequate selection of the T-end search interval
, improved T-wave peak detection and computation of the angle between two 1
0 ms long adjacent segments along the search interval. The algorithm avoids
the use of baseline crossing and direct signal differentiation. It perform
s well in cases of biphasic and/or complex T-wave shapes. Mean differences
with respect to observer data are 13.5 ms for the higher gain/speed records
and 14.7 ms for the lower gain/speed records. The algorithm was tested wit
h 254 various T-wave shapes. Comparisons with two other algorithms are pres
ented. The lack of a 'gold standard' for the Tend detection, especially if
small waves occur around if, impeded adequate interobserver assessment and
evaluation of automatic methods, it is speculated that a simultaneous prese
ntation of normal and high-gain records might turn more attention to this p
roblem. Automatic detection methods are in fact faced with 'high-gain' data
, as high-resolution analogue-to-digital conversion is already widely used.