Cancer stem cells (CSCs) appear to be an essential target for cancer therapies, in particular, in brain tumors such as Glioblastoma. Nevertheless, their isolation is made difficult by their low content in culture or tumors (<5% of the tumor mass) and is essentially based on the use of fluorescent or magnetic labeling techniques, increasing the risk of differentiation induction. The use of label-free separation methods such as sedimentation field-flow fractionation (SdFFF) is promising, but it becomes necessary to consider a coupling with a detection and characterization method for future identification and purification of CSCs from patient-derived tumors. In this study, we demonstrate for the first time the capability of using an ultrahigh-frequency range dielectrophoresis fluidic biosensor as a detector. This implies an important methodological adaptation of SdFFF cell sorting by the use of a new compatible carrier liquid DEP buffer (DEP-B). After SdFFF sorting, subpopulations derived from U87-MG and LN18 cell lines undergo biological characterization, demonstrating that using DEP-B as a carrier liquid, we sorted by SdFFF subpopulations with specific differentiation characteristics: F1 = differentiated cells/F2 = CSCs. These subpopulations presented high-frequency crossover (HFC) values similar to those measured for standard differentiated (around 110 MHz) and CSC (around 80 MHz) populations. This coupling appeared as a promising solution for the development of an online integration of these two complementary label-free separation/detection technologies.