Secondary abstract: |
Electrocardiography is amongst the basic and relatively accessible examinations in
cardiology. One of the key tasks of electrocardiography is accurately detecting each
heart beat or the so-called QRS complex, the most recognizable electrocardiogram
section. In thesis we have first presented automated QRS complex detection as a
principal process in the automatic electrocardiogram analysis, and afterwards implemented,
as well as improved upon, two QRS complex detectors. We have evaluated
both detectors on two standard databases of annotated electrocardiograms, namely
the MIT-BIH DB arrythmia database and the LTST DB database, as well as on the
selected, challenging electrocardiograms. With the first detector we have achieved
96.09 % sensitivity with 99.57 % positive predictive value on MIT-BIH DB arrythmia
database, and 94.90 % sensitivity with 99.30% positive predictive value on the
LTST DB database. With our implementation of the second detector, presented at
last year’s Engineering in Medicine & Biology Society conference, we have reached
to 99.81 % sensitivity with 99.90 % positive predictive value on the MIT-BIH DB
arrythmia database, and to 99.96 % sensitivity with 99.89 % positive predictive
value on the LTST DB database. |