Type 2 Brugada Syndrome (T2-BrS) is the most prevalent Brugada ECG phenotype in the general population, yet its characteristic saddleback morphology is not independently diagnostic without pharmacological provocation and specialist interpretation. No published end-to-end automated pipeline targets the binary T2-BrS versus Non-BrS classification task directly from standard 12-lead recordings. This thesis addresses that gap. A labelled dataset of 362 patient-disjoint 12-lead ECG recordings (188 confirmed Type 2 BrS and 174 Non-BrS) was assembled through two acquisition pathways: semi-automated digitisation of approximately 1,000 paper ECGs from a Sardinian Brugada patient registry using the ECGD software tool, and direct extraction of digital signals from GE MAC2000 XML archives.