Partial Encryption Scheme of Medical Images Based on DWT, Secret Image Sharing and Hyperchaotic System

There has been a significant increase in the demand for secure image storage in healthcare organizations in recent years. Encryption is used to address the challenge of encrypting sizeable digital image files, as full encryption can be computationally expensive and take a longAtime to process. In this paper, partial and selective encryption is proposed for medical images. First, deep learning based on U-Net is used to localize a tumor region called (ROI) a region of interest. A diffusion phase of the proposed system handles pixel values and positions based on linear and hyperchaotic systems. It includes converting an image's pixel values and repositioning pixels in a predetermined order. In the confusion phase, one level of Integer Discrete Wavelet Transform (IWT) is applied to divide the scrambled region into four sub-bands. Then, a Feistel network based on polynomial-based secret image sharing (SIS) Aencrypts the lowest frequency band only while the three bands LH, HL, and HHAare diffused using a mapping technique based on the Morton scan to swap coefficients positions and then confused based on the hyperchaotic system. The culmination of these techniques results in generating a test image cipher characterized by robust confusion and diffusion properties. Importantly, this methodology has yielded remarkable results, reducing the encryption time by up to 96%. AThis efficiency is achieved without compromising the security or quality of the encrypted medical images. as high entropy is attained post- encryption. Furthermore, by employing the Integer Discrete Wavelet Transform (IWT), the integrity and fidelity of the encrypted images remain uncompromised. AAdditionally, to bolster the level of confusion in the encryption process, a substantial key space of 2 1628 has been employed, further enhancing the resilience of the encryption method.