Secure Two-Layer Encryption for DICOM Annotations: Ensuring Confidentiality and Real-Time Efficiency

Effective image annotation is a fundamental pillar in computer vision and artificial intelligence, crucial for the development of accurate machine learning models.The research proposes a solution to the problem of encrypting DICOM images synthesis with the outline of advanced security measures, particularly using the Advanced Encryption Standard (AES). This feature enables users to interactively mark interest points on medical images, as well as encrypt the annotation coordinates as well as the comments added by the user. The annotation and comment stored contained in the DICOM metadata are protected by two-level security systems that require two different passwords for decryption. This work is validated with image metrics measurements and Peak Signal to Noise Ratio (PSNR) as well as Structural Semeantics Index (SSIM) of the DICOM images that were annotated, to ensure proper security and quality of the annotated DICOM images. The preservation of data integrity, in addition to other critical assets confidentiality, allows for unrestricted linking of medical images between professionals in the field without breaching safety issues.