The majority of copyright protection scheme are focused on 2D medical images, with only few being proposed for the copyright protection of complex medical data, especially for Diffusion tensor-Magnetic resonance imaging (DT-MRI). We propose a robust zero watermarking scheme for DT-MRI tensor image based on orthogonal ray-casting tensor sampling (ORCTS) and the quaternion generic polar complex exponential transform (QGPCET), which can effectively provide integrity verification and copyright protection measures for DT-MRI, also can greatly improve the efficiency of watermarking. First we propose the ORCTS for DT-MRI tensor data, which allows feature engineering of the physiological tissue information represented by DT-MRI, using ORCTS to generate three feature images of DT-MRI in coronal, sagittal and axial planes. Then uses three feature images as the quaternion’s three imaginary components, calculates their QGPCET coefficients to constructing the coefficients feature image; Finally the zero watermarking image is obtained by performing the XOR operation between the coefficients feature image and watermarking information. Compared to other zero watermarking schemes for 2D medical images, the results of experiments show that proposed scheme can resist common image processing attacks and geometric attacks with better robustness, and can guarantee the medical availability of DT-MRI tensor data. In the best case, the scheme can reduce the watermarking storage consumption to 0.07%. and reduce the time consumption to 0.75%.