This paper investigates a 3D novel dual active contours approach to segment multiple regions in medical images. The locally based segmentation approaches can handle the heterogeneity of the image as well as the noise artefacts. In this light, a locally based dual active contours approach is proposed to separate among three regions constituting the image. The dual contours approach combines the local information along each point in the two curves conjointly with the information between them. Different parameters in this approach determine its accuracy, including the initial distance between the two curves and how much local the information is used in each curve. The approach’s efficiency is evaluated on synthetic images as well as HRpQCT and MRI data compared to state-of-the-art techniques. The computational cost of this approach is reduced using the convolution operator and the FFT transform. The experimental evaluation of the approach demonstrates its segmentation performance on synthetic images and real medical images.