Authentication and Self-Correction in DNA Identification based on Agarose-Gel Images Vassilis Fotopoulos, Argyro Sgourou, Athanassios N. Skodras School of Science and Technology Hellenic Open University GR-26222, Patras, Greece Email: vfotop1@eap.gr Abstract— Agarose gel electrophoresis is an ideal method for separating and analyzing DNA and RNA molecules. Analysis is based on digital images taken by cameras equipped with UV filters and finds its main use in DNA identification. It is thus of paramount importance that these images employ authentication mechanisms and if possible, self-restore capabilities against ma- licious tampering. The proposed approach achieves the required authentication and self-correction by hiding a copy of the regions- of-interest into the region-of-non-interest of the images. I. I NTRODUCTION Agarose gel electrophoresis is the easiest and commonest way of separating and analyzing DNA and RNA molecules (although proteins can also be separated on agarose gels). The separation is achieved by utilizing the mobilities with which different-sized DNA or RNA can transverse a viscous medium (agarose gel), in the presence of an electric field. The size of the molecule affects its mobility; smaller fragments move closer to the anode than longer ones, within a specific time frame. Thus, the DNA molecules are separated into classes based on length. Larger separation can be achieved using higher running times or higher voltages. The DNA/RNA fragments can then be visualized using ethidium bromide as a fluorescent dye, under UV light exposure [1]. Fragment size is most commonly determined by comparison to DNA fragments of known size, known as DNA ladders (arrows in Fig. 1). There exist a number of applications of the agarose gel elec- trophoresis method including research protocols, medicine, forensics etc. The method allows a rough estimation of DNA/RNA quantity, quality and size of DNA/RNA molecules using the afforementioned DNA ladder [2]. In general it suits the purpose of the primary step for other techniques that depend on the separation of nucleic acid molecules in an agarose gel matrix. Data from the method are visualized on a transilluminator and images of the gels are captured on specialized digital cameras with UV filters, in order to keep a record of the experiments. In order to avoid accidental or deliberate dis- tortion of facts in either research or medical and forensic results, securing this imaging data is of critical importance. For instance, forensics deal with evidence collected from crime scenes and criminal investigations. The agarose electrophoresis gels are used for the separation of the isolated DNA from different sources and different lanes on the gel represent DNA Fig. 1. A typical agarose gel electrophoresis image from the victim, the suspect, or DNA isolated from other elements found at the crime scene. All DNA samples are then subjected to genetic analysis in order to detect similarities between the suspect’s genetic material and DNA left on evi- dence. Possible disguise of the original data (just by changing the order of DNA samples), may eventuate in prosecution of innocent people and this raises ethical and social issues [3]. The same issues are addressed in the case of medical data, where sensitive genetic information could be obtained from the DNA samples, such as family relationships and disease susceptibility. Researchers using such DNA databases have access to people’s DNA without their consent. This can be seen as an intrusion of personal privacy and a violation of civil liberties. We are currently working on developing a protecting mech- anism for sensitive digital data that come from agarose gel electrophoresis. The protection process can later be applied in other biologically inspired assays where results are stored as digital images. In the present work, we explore image authen- ticity by attempting to identify tampered regions. Furthermore, in case tampering has occurred exclusively into the part of the image that contains the DNA bands, the original data can be fully recovered from those saved within the image. Our general aim is to impose certain restrictions regarding alter- ations performed on the original image. These authentication applications are crucial for protecting any kind of imaging data (either medical or forensic) where unauthorized manipulations could result in adverse consequences.