Abstract— This paper introduces a transformative watermarking technique tailored for multi-channel images, utilizing discrete Krawtchouk moments to capture intricate features in medical images. The method facilitates seamless watermark embedding in the transform domain, preserving the diagnostic integrity of medical images. The proposed scheme offers a robust framework for watermark insertion and extraction, resilient against twelve types of attacks. Beyond watermarking, our approach extends to reconstructing multi-channel images, a significant advancement in medical imaging. This enables clinicians to recover pristine images for precise diagnostics while ensuring data security. Experimental evaluations on an 89-image diabetic retinopathy dataset confirm the efficacy of our techniques, showcasing the ability to embed and extract watermarks without compromising diagnostic quality and successfully reconstruct original images.