Diffuse low grade gliomas are slowly growing tumors that always recur after treatment. In this paper, we revisit the modeling of the tumor radius evolution before and after the radiotherapy process and propose a novel model that is simple, yet biologically motivated, and that remedies some shortcomings of previously proposed ones. We confront it with clinical data consisting in time-series of tumor radius for 43 patient records, using a stochastic optimization technique and obtain very good fits in all the cases. Since our model describes the evolution of the tumor from the very first glioma cell, it gives access to the possible age of the tumor. Using the technique of profile-likelihood to extract all the information from the data, we build confidence intervals for the tumor birth age and confirm the fact that low-grade glioma seem to appear in the late teenage years. Moreover, an approximate analytical expression of the temporal evolution of the tumor radius allows us to explain the correlations observed in the data.