Image segmentation is the process of partitioning an image into a set of distinct regions. Some of its practical applications are in medical image processing, facial and fingerprint recognition, object detection, and more. During image acquisition or transmission, the images may be corrupted by noise which affects the accuracy of segmentation applications. In this paper, we present a modified version of the Chan-Vese model for segmenting images with Gaussian noise. The modification is mainly a change on the total variation (TV) term of the discretized Chan-Vese model where we considered the use of a nonconvex TV q – type norm where q∈(0,1) . To solve both global and local regularized versions of the proposed model, we used a steepest descent method. Numerical results are provided in this paper to show the efficiency of the proposed model in segmenting noisy images.