The paper investigates the spatial properties of the Galactic center gamma-ray excess (GCE) observed by the Fermi-LAT telescope, focusing on the impact of different models for Galactic diffuse emission and the masking of the Galactic plane and point sources. The authors compare a spherically symmetric morphology, potentially indicating dark matter (DM) annihilation, with a boxy morphology, suggesting unresolved sources in the Galactic bulge. They find that DM-motivated templates, which have been favored in recent studies, perform worse than other templates for the Galactic bulge. A non-parametric model of the Galactic bulge derived from the VVV survey yields a significantly better fit for the GCE compared to DM-based templates, regardless of the background model used. This conclusion holds even when additional freedom is allowed in the background models, allowing for non-parametric modulation of the components. The optimized background models consistently favor a boxy bulge morphology for the GCE, regardless of the mask applied to the Galactic plane. The paper also discusses the reproducibility and improvements of previous works, including the use of different bulge templates and ring-based models, and addresses systematic uncertainties related to mask size and goodness of fit using Monte Carlo simulations.The paper investigates the spatial properties of the Galactic center gamma-ray excess (GCE) observed by the Fermi-LAT telescope, focusing on the impact of different models for Galactic diffuse emission and the masking of the Galactic plane and point sources. The authors compare a spherically symmetric morphology, potentially indicating dark matter (DM) annihilation, with a boxy morphology, suggesting unresolved sources in the Galactic bulge. They find that DM-motivated templates, which have been favored in recent studies, perform worse than other templates for the Galactic bulge. A non-parametric model of the Galactic bulge derived from the VVV survey yields a significantly better fit for the GCE compared to DM-based templates, regardless of the background model used. This conclusion holds even when additional freedom is allowed in the background models, allowing for non-parametric modulation of the components. The optimized background models consistently favor a boxy bulge morphology for the GCE, regardless of the mask applied to the Galactic plane. The paper also discusses the reproducibility and improvements of previous works, including the use of different bulge templates and ring-based models, and addresses systematic uncertainties related to mask size and goodness of fit using Monte Carlo simulations.