• fragmented QRS complex
• myocardial scar
• Selvester Scoring System
• stationary wavelet transform

This thesis was developed as part of a research visit to the University of Southampton in collaboration with expert cardiologists from Southampton Hospitals NHS Trust.
This project describes an algorithmic procedure to detect and categorize fragmented QRS complexes (f-QRS) in 12 lead-electrocardiogram. In fact, the f-QRS is considered a biomarker of various heart diseases, such as myocardial ischemia and scar. The proposed algorithm of detecting f-QRS can then be applied in the development of an automated Selvester scoring system, which is used to localize and quantify the size of myocardial scar in the heart’s tissue. The latest refined version of the Selvester score is heavily based on the presence and categorization of f-QRS, which allows the discrimination among the various confounding condition considered, such as Left Bundle Brunch Block (LBBB), Right Bundle Branch Block (RBBB), Left Ventricular Hypertrophy (LVH) and Left Anterior Fascicular block (LAFB).
An algorithm in Matlab was developed to detect and categorize the f-QRS into the three fragmentation categories, notching, slurring, slowing, using the Stationary Wavelet Transform (SWT). Following, the rules of the Selvester Scoring System were implemented as an extension to this algorithm resulting in a fully automated Selvester scoring technique.
12-lead standard electrocardiogram from 62 patients with confirmed myocardial scar was used, in order to test the f-QRS detection and the Selvester scores and compare the algortihm’s results with manual Selvester scoring results made by cardiologists.