Abstract: The continuous development of CT imaging techniques has significantly reduced radiation dose to the patient.However,maintaining high image quality in low-dose CT(LDCT)reconstruction is still important concerns.In order to improve the image quality and reduce noise artifact disturbance,we propose a wavelet domain convolutional sparse coding algorithm for LDCT reconstruction.With the predetermined filters,the convolutional sparse coding is introduced in the high frequency sub-band of wavelet domain to construct a new LDCT reconstruction objective function.The CT image and the wavelet domain convolutional sparse coding are sequentially updated using an alternating minimization scheme.Quantitative evaluations on Sheep-Logan phantom,AAPM simulation dataset and UIH real dataset,compared with the traditional total variation,dictionary learning and convolutional sparse coding with gradient regularization reconstruction algorithm,demonstrate that the proposed method can achieve satisfactory performance in terms of noise suppression and structural preservation.Further,the parameters analysis experiments show that this method can be easily implemented with good robustness in parameter setting.