The Koroshina Dolomite Formation is an important gas-bearing carbonate reservoir in the Sadad Gas Field, central Syria, where strong structural heterogeneity and limited well control make the prediction of reservoir properties challenging. This study aims to predict the spatial distribution of reservoir porosity within the Koroshina Dolomite Formation by integrating well-log data, two-dimensional (2D) post-stack seismic data, seismic inversion, and artificial neural network modeling. Well-log data from the SADAD-1 and SADAD-2 wells were integrated with a migrated 2D seismic profile crossing both wells. Model-based post-stack seismic inversion was performed using Hampson–Russell software to obtain an acoustic impedance model constrained by well-log data. Seismic attributes derived from the inversion results were then analyzed and used as input variables for a Probabilistic Neural Network (PNN), with measured well-log porosity used as the target property. Porosity prediction using individual seismic attributes produced a correlation coefficient of approximately 39.7%, which increased to 75.1% when 28 seismic attributes were combined. The final PNN model achieved a correlation coefficient of approximately 83.5% between predicted and measured porosity. The resulting porosity model revealed pronounced lateral and vertical heterogeneity within the reservoir. Higher predicted porosity near the SADAD-2 well and the associated reverse-fault zone is interpreted as fracture-related secondary porosity, whereas lower-porosity areas are associated with compaction and cementation. These results demonstrate that integrating post-stack seismic inversion, seismic attributes, and PNN modeling can provide a practical approach for extending porosity prediction into interwell areas and improving characterization of structurally complex carbonate reservoirs.