This article presents the preliminary results of the research that aims, as its main objective, the development of a method to improve an image obtained in a mammographic examination. It is known that, in order to increase the quality of mammographic images, it is necessary to increase the radiation dose used in the exam, since with low doses, the quantum noise interferes with the quality of the same, however, this solution is not feasible since it would put at risk the patient's health. Therefore, the method proposed in this article proposes an alternative way to improve these images, combining wavelet transforms and convolutional neural networks. This method was tested on a set of 10 mammographic images taken from an online database and its effectiveness was verified through objective and non-objective metrics. The objective metrics used were the peak signal-to-noise (PSNR) and the structural similarity index (SSIM), for the PSNR metric we had an average of 32.52 while for the SSIM we had an average of 0.71. And regarding the subjective analysis of the method, we see that the improved images have a greater accentuation in their lines and edges.