Rheumatic Heart Disease (RHD) is a serious cardiovascular disease in children and young adults, and accurate diagnosis depends heavily on echocardiography. However, echocardiographic images often have low contrast and high noise, compromising diagnostic reliability. To address this challenge, we present a two-stage enhancement framework that couples Contrast Enhancement via Dynamic Histogram Equalisation (CEDHE) with Generative Adversarial Networks (GANs). Unlike conventional histogram equalisation, CEDHE emphasises low-contrast regions while preserving high-intensity details, and the subsequent GAN refinement further improves image quality by removing noise and restoring delicate structures. The framework was implemented in Python and evaluated on a real-time echocardiography dataset. It achieved up to 31.06 dB in PSNR and 74% in SSIM, outperforming baseline methods and producing high-quality images with clear detection of cardiac structures in the experiment. In this regard, these technologies facilitate characterisation of valve morphology, chamber measure and functional defects, which are essential for the diagnosis of RHD. This represents an acceptable method for improving the interpretation of echocardiographic images in clinical practice.