Lossless digital signature embedding for medical image integrity assurance

Medical image security considers three major issues: confidentiality, authenticity, and integrity. With integrated medical imaging systems being extensively used in clinics for healthcare delivery, image security consideration is no longer limited to images in transit but also in archive. This dissertation focuses on the image integrity issue with three aims: (1) Investigate two general two-dimensional (2D) lossless digital signature embedding (LDSE) methods for assuring medical image integrity when a clinical image is in transit and in archive; (2) Research a three-dimensional (3D) LDSE method to efficiently preserve the integrity of a large image volume; and (3) Develop a systematic auditing method for tracing the integrity of a clinical image during its life in a Picture Archiving and Communication System (PALS).; The first aim permanently embeds the digital signature (DS) of a clinical image into image pixels using lossless data embedding approaches. The permanently embedded DS can be verified to ensure image integrity anytime, at anywhere, and therefore integrity can be assured during the lifetime of the image. These lossless embedding methods do not change the visual appearance of the image and can completely restore the original image, both of which are important to satisfy the clinical requirements.; The LDSE method can be extended from 2D to 3D. This extension is significant, because various imaging modalities currently used in the daily clinical practice generate 3D volumetric image sets, each of which may contain hundreds of images. There is the need for an efficient 3D LDSE method to assure the image integrity of these large image sets. Research of this method is the second aim.; The third aim is prompted by the Health Insurance Portability and Accountability Act (HIPAA) mandate. An auditing method for tracking image access has been developed as a means to track the usage of any image throughout its lifetime in a PALS. This auditing method utilizes methods developed in Aims 1. Experiments of these three aims were tested and validated with clinical data and rigorous evaluation plans. Results indicated that LDSE and auditing methods were effective for assuring the integrity of medical images in large-scale clinical imaging systems.