In this paper we study a new kind of geometric mean of positive definite operators A and B, named the near-geometric mean, of the form for \(t\in [0,1]\) \(\begin{aligned} A \star _{t} B = A^{1/2} (B^{1/2} A^{-1} B^{1/2})^{t} A^{1/2}, \end{aligned}\) arising from the determinantal inequality in diffusion tensor imaging. We show the geodesic property of near-geometric mean for the Thompson metric, the monotonicity on parameters and the boundedness of near-geometric mean. Moreover, we compare other known geometric means and Rényi mean with near-geometric mean in terms of the log-majorization.