Objective MRI sequences in breast are continuous with fixed positions, yet 2D images cannot conveniently provide the information of the depth and the whole shape of the tumor to be observed. In this case, the paper proposes a method of 3D reconstruction on the segmentation of breast and tumor. Methods Firstly, the contour of the breast is extracted by the use of the combination method of Otsu and region growing based on region processing. Then, it makes use of the level set based on energy and gradient to segment the tumor exactly from breast. Finally, it chooses volume rendering to deal with the segmentation results and adopts the ray casting method based on the bounding box to realize the reconstruction. Results This paper conducts the experiments on multiple groups of patients’ MRI, and evaluates objectively on several segmentation algorithms. Experiments of 3D reconstruction get the stereoscopic image of the breast and tumor to be observed for doctors. Conclusions Segmentation based on level set of region processing and 3D reconstruction based on ray casting method provide a more vivid 3D visual effect for doctors, which can not only contribute to the diagnosis of the severity of the tumor, but help surgeons to decide the surgical site accurately in breast-conserving surgery.