Abstarct: One of the main approaches for identifying subsurface geological structures is the use of geophysical methods, particularly seismic reflection surveying. Salt domes are among the most significant subsurface structures, serving as important hydrocarbon traps and key natural geological features for hydrocarbon storage. In seismic reflection studies, salt domes can be identified through effective subsurface imaging; however, there are limitations in accurately delineating their boundaries. To address this, seismic attributes are commonly employed. Seismic attributes are highly effective tools in seismic interpretation, enabling a clearer and more convenient characterization of geological features by revealing hidden information within seismic reflection data. Nevertheless, many seismic attributes—especially certain conventional texture-baxsed attributes—lack the capability to precisely represent complex structures such as salt domes. This limitation is primarily due to the non-linear nature, complex geometry, and reflector interference surrounding these structures, which hinder some attributes from clearly distinguishing their boundaries. Therefore, selecting an optimal set of combined attributes and applying machine learning algorithms, such as the Support Vector Machine (SVM), plays a critical role in enhancing interpretation accuracy and improving the detection of salt structures. One of the main challenges in this study was training the SVM model using labeled data baxsed on 22 selected seismic attributes. The aim of this process was to accurately identify the boundaries and three-dimensional geometry of salt domes within the seismic data. The results demonstrated that the SVM, when applied with these attributes, was able to fully reveal the 3D structure of the salt domes with high accuracy and clarity, providing a clear depiction of their boundaries.