Deep Learning-based Brain Tumor Segmentation and Secure Watermarking

The digitization of medical imaging has revolutionized healthcare, enabling efficient diagnosis and storage. However, it also introduces challenges in ensuring the security and integrity of sensitive medical data. This paper presents a comprehensive framework integrating deep learning-based tumour segmentation, robust watermarking using DWT-HD-SVD techniques, and AES encryption for secure transmission. The U-Net architecture is utilized for automated segmentation, achieving high accuracy. Watermarks are embedded imperceptibly and evaluated for robustness against attacks. AES encryption with SHA-256 enhances security during transmission. Extensive simulations validate the methodology’s effectiveness, making it suitable for telemedicine applications, enhancing both security and reliability of medical imaging.