One way of detecting interesting sequences is to analyze the RR intervals. For this purpose, a QRS detector is needed. Other pathologies may reflect only in changes in the QRS morphologies. Such situations cannot be seen from RR sequences and therefore they cannot be detected using a QRS detector only. On the other hand, manual detection of these changes - especially from long-term ECG recordings - is very time consuming.

Open the singal 212.hea, which is taken from the MIT-BIH Arrhythmia Database on PhysioNet and scroll to second 320. What you see is an ECG in normal sinus rhythm. This ECG seems to stem from a sane patient. The first step in ECG analysis would be QRS detection. Therefore, start the QRS Detector and additionally apply the correlation classificator to the signal. Make sure that the checkbox Open with Single Beat Editor is selected when starting the classificator.

Signal 212.hea is quite an interesting signal, since there are two different types of cardiac activities, which cannot be separated by analyzing the RR intervals only. E.g. you could select Result 2D on the lower right side of the Single Beat Editor and select BCI (beat coupling intervals) als the Y-Parameter. From the resulting figure no abnormalities can be detected:

Looking at the Single Beat Editor you can see the signal from two leads of one single heart beat. When the Single Beat Editor is opened after the Correlation Classification, the first beat of correlation class 1 is displayed. Select Line styles... in the lower left Channels menu, press Select all ->>, choose a color, line style and line width and press Apply to signal. Now enable the Hold checkbox in the lower Show menu of the Single Beat Editor. Thereafter choose correlation class 2 in the Markers menu. Select another linestyle in the Line Style Editor and press Apply to signal again. What you see is a figure similar to the one below.

Both beat classes show very similar morphologies at the beginning of the QRS complexes, but - due to a bundle branch block - the QRS complexes of beat class 1 are widened. Select Result2D... again and choose SC (correlation class) as the Y-Parameter. In order to focus on the two classes "Normal" and "Right Bundle Branch Block" - represented by the correlation classes 1 and 2 - press Advanced selection..., mark "1" and "2" in the SC field and press the AND button. Select Ok to close the Event Selection Editor and click on Ok to start Result2D.

In the Result2D figure you can see how the cardiac activity changes in between normal sinus rhythm (beats of correlation class 2) and heart beats with right bundle branch block (beats of correlation class 1). Several short episodes of normal depolarisation are alternating with right bundle branch block episodes.

Finally you can apply the Characteristic Points Detector and Averaged Beat Calculation to the signal and start the Advanced Classification. In the Data Editor each heart beat is annotated as "Normal" or "Complete BBB" now. In the Result2D a summery of the heart beats found is plotted which highly correlates with the one on PhysioNet. As you can see, correlation class 1 refers to the bundle branch block beats. Class 2 represents normal heart beats. Class 3, 5 and 6 stem from noisy sequences. Class 4 is similar to class 2 but provides a different beat morphology in lead V1.