Abstract: The Gaussian beam migration method is widely used in seismic imaging due to its high efficiency, flexibility, and good adaptability to anisotropic media. Least-squares migration (LSM) based on Gaussian beams theoretically offers superior imaging accuracy, yet it still faces challenges in computational efficiency and stability. To improve the imaging quality of subsurface complex structures, this paper proposes an image-domain, non-iterative least-squares Gaussian beam migration method suitable for transversely isotropic (TI) media. First, anisotropic ray tracing is employed to compute the traveltimes and amplitudes of Gaussian beams, thereby constructing the Green's functions for TI media. Subsequently, the point spread function (PSF) of the subsurface space is analytically characterized based on the Green's function, serving as an approximation of the Hessian matrix. Finally, high-dimensional deconvolution is applied to the classical anisotropic Gaussian beam migration result using the point spread function, achieving non-iterative least-squares migration in the image domain for TI media. Validation using synthetic models demonstrates the method's good imaging performance in TI media, and application to real seismic data further confirms its potential.