Brugada Syndrome (BrS) is an inherited cardiac channelopathy associated with sudden arrhythmic death in structurally normal hearts. Among its three recognised ECG pattern types, the Type-3 pattern is defined by right precordial ST elevation below 2mm in leads V1–V2 and is the most morphologically subtle and clinically ambiguous manifestation. No published machine learning pipeline has previously addressed the binary discrimination of Type-3 BrS from non-Brugada (Non-BrS) ECGs as a dedicated classification task under a patient-disjoint evaluation protocol. This thesis presents BrugadaECGNet, an end-to-end pipeline for Type-3 BrS classification from standard 12-lead ECGs. A dataset of 237 recordings was assembled from two heterogeneous acquisition sources: 63 Type-3 BrS recordings in GE Sapphire DCAR XML format from a clinical centre in Torino, Italy, and 174 Non-BrS recordings digitised from paper ECGs.