X-ray differential phase contrast computed tomography imaging based on grating interferometer system can reconstruct not only the linear attenuation coefficient, but also the phase shift coefficient and the linear scattering coefficient of the object. In practical application, it is very difficult to make a large area grating, so the sample is often larger than the grating. When the sample is scanned with a grating smaller than the sample, the part of the sample beyond the field of view of the grating will cause the differential phase projection information to be truncated. In this paper, a method of reconstructing the region of interest for differential phase contrast computed tomography is proposed. The method is based on the approximate linear relation between the phase shift coefficient of the object and the linear attenuation coefficient (i.e. the decrement in the real part of the refractive index and the imaginary part of the refractive index), the phase shift coefficient of the region of interest is approximately reconstructed by the polynomial of Lambda function of the phase shift coefficient and Lambda inverse function of linear attenuation coefficient. In this paper, according to the Fresnel diffraction theory and differential phase grating phase step-by-step method of imaging a simulation experiment is performed. In the experiment, conducted is the approximate reconstruction by using the first order polynomial and quadratic polynomial of Lambda function of the phase shift coefficient and Lambda inverse function of linear attenuation coefficient. The sample size is five times of grating imaging field, and the results show that this method can approximately reconstruct the region of interest for the sample image. We also carry out the actual data experiment. The actual data are obtained by the Talbot grating interferometer system of Shanghai synchrotron radiation BL13W1 station, and the standard model and biological sample are imaged. The method of reconstructing the region of interest is proposed in this paper. This method can be applied to the multi-material samples with a similar relationship between the decrement in the real part of the refractive index and the decrement in the imaginary part of the refractive index, and also to single-material samples. The comparison between the numerical simulations and the actual experimental results verifies the effectiveness of the proposed method.