Stega Care: Securing Virtual Therapy Images with AI-Driven Image Forensics

Abstract

With the rapid evolution of digital healthcare, virtual therapy platforms have become essential tools for delivering mental health services remotely. These platforms enhance accessibility, especially for individuals in remote or underserved areas. However, the transmission of therapeutic images—such as visual assessments or expressive content generated during sessions—raises significant cybersecurity concerns. Given the sensitive nature of such data, robust protection mechanisms are required to ensure privacy, integrity, and patient trust. This research proposes an artificial intelligence-driven framework designed to enhance the security of virtual therapeutic images by integrating image forensics and steganography techniques. Central to this approach is a deep learning-based steganalysis classifier capable of detecting hidden alterations and unauthorised data embedding in medical images. By leveraging convolutional neural networks (CNNs), the classifier accurately identifies covert manipulations while maintaining image fidelity and confidentiality. The system is trained and evaluated using benchmark steganographic image datasets, demonstrating high effectiveness in identifying steganographic threats and detecting tampered content in real-time. Experimental results indicate that the proposed model performs well even in complex attack scenarios involving sophisticated data-hiding techniques. The framework offers a scalable and proactive solution for safeguarding sensitive therapeutic content in telehealth environments. By embedding this AI-powered detection capability into virtual therapy platforms, healthcare providers can significantly enhance their cybersecurity posture.