When gravitational waves (GWs) pass by a massive object on its way to Earth, a strong gravitational lensing effect will happen. Thus, the GW signal will be amplified, deflected, and delayed in time. Through analyzing the lensed GW waveform, physical properties of the lens can be inferred. On the other hand, neglecting lensing effects in the analysis of GW data may induce systematic errors in the estimating of source parameters. As a space-borne GW detector, TianQin will be launched in the 2030s. It is expected to detect dozens of mergers of massive black hole binaries (MBHBs) as far as z=15 and thus will have high probability to detect at least one lensed event during the mission lifetime. In this article, we discuss the capability of TianQin to detect lensed MBHB signals. Three lens models are considered in this work: the point mass model, the singular isothermal sphere (SIS) model, and the Navarro-Frenk-White (NFW) model. The sensitive frequency band for space-borne GW detectors is around millihertz, and the corresponding GW wavelength could be comparable to the lens gravitational length scale, which requires us to account for wave diffraction effects. In calculating lensed waveforms, we adopt the approximation of geometric optics at high frequencies to accelerate computation, while precisely evaluating the diffraction integral at low frequencies. Through a Fisher analysis, we analyze the accuracy to estimate the lens parameters. We find that the accuracy can reach to the level of 10-3 for the mass of point mass and SIS lens and to the level of 10-5 for the density of the NFW lens. We also assess the impact on the accuracy of estimating the source parameters and find that the improvement of the accuracy is dominated by the increasing of signal-to-noise ratio.