This work examines several volume attributes extracted from 3D seismic data with the goal of seismic facies classification and lithology prediction in intraslope minibasins. The study area is in the Keathley Canyon protraction (KC), within the middle slope of the Northern Gulf of Mexico (GOM). It lays within the tabular salt and minibasins province downdip of the main Pliocene and Pleistocene deltaic depocenters. Interaction between sedimentation and mobile salt substrate lead to the emergence of many stratigraphic patterns in the intraslope minibasins. Interest in subsalt formations left above salt formations poorly logged. Facies classification using Artificial Neural Network (ANN) was applied in those poorly logged areas. The resultant facies classes were calibrated and used to predict the lithology of the recognized facies patterns in an intraslope minibasin, away from well control. Three types of facies classes were identified: Convergent thinning, convergent baselaping and bypassing. The convergent baselaping are found to be the most sand rich among all other facies.