Abstract This study introduces a multicontraction microfluidic channel that can differentiate glioma cells from normal glio cells. As cells pass through successive constriction channels, the incremental velocity and varying size profiles of the cells will be collected, reflecting their biophysical properties. The data of high-dimensional variables were analyzed , including the cell sizes, velocities, and velocity increments. The prediction value is used to represent the difference between two groups using the established classification model from high-dimensional variables. At the same time, we prepared three groups of primary tumor cells from patients with different grades of glioma to verify the efficacy of this classification method. The results show that in this microfluidic channel, the diagnostic model made of cell biophysical properties can well identify glioma cells, which gives a novel method for efficient identification of circulating tumor cells and rapid pathological diagnosis.