Damghan University Press Iranian Journal of Astronomy and Astrophysics 2322-4924 12 4 2026 07 28 Gravitational Lens Detection in Simulated LSST Data: A Comprehensive Feature Extraction Approach 347 357 2140 10.22128/ijaa.2026.3309.1246 EN Seyedeh Nahid Hasani Department of Physics, University of Guilan, 41335-1914, Rasht, Iran Nasibe Alipour Department of Physics, University of Guilan, 41335-1914, Rasht, Iran 0000-0003-3643-5121 Reza Saffari Department of Physics, University of Guilan, 41335-1914, Rasht, Iran 0000-0002-7186-8371 Journal Article 2026 04 30 Identifying strong gravitational lensing systems is a major challenge in modern astronomical surveys. This research presents a machine learning-based method to distinguish lensed objects from non-lensed objects using multiband data.<br />To this end, a comprehensive set of physically interpretable features is extracted from simulated data across the g, r, i, and z bands. These features capture statistical, spatial, amplitude, and frequency information. Then, using the XGBoost algorithm, we identify the features that have the greatest impact on the classification process. The model used in this study primarily relies on indicators of structural complexity (such as spread, entropy, and kurtosis) and asymmetry (such as skewness), especially in the g band. Our analysis shows that the proposed model achieves strong performance, with an accuracy of 0.88 and a TSS score of 0.75. Additionally, the Area Under the Curve (AUC) of 0.94 demonstrates the model's high ability to distinguish between gravitational lenses and non-lensed systems. Overall, the proposed framework can provide a reliable approach for identifying gravitational lens candidates. https://ijaa.du.ac.ir/article_2140_0afd552686ebc3db5d6d4954792b7877.pdf