Secure Medical Data Abstraction Using Convolutional Neural Network

One of the most crucial components of medicine is data security and privacy which is vital while transferring it over the internet. There are some situations where the patient is not satisfied with the disclosure of his medical information to outsiders. Steganography is the technique where secret information can be hidden inside another ordinary file or document. The problem here is that by using traditional methods, the data may not be extracted properly at the destination. Consequently, this study suggests a methodology for image steganography based on convolutional neural networks (CNN) that hide the patient’s confidential data like identity, medical condition, treatment, and comments of doctors inside an MRI scan of a patient with Dicom image format. The data is encrypted before being incorporated into the cover image for an additional layer of security, and the obtained information can be deciphered at the destination to retrieve the original information. The result in terms of performance will be an increase in hiding capacity and level of abstraction of data.