3D medical images security via light-field imaging

Area of Science:

• Medical Imaging
• Information Security
• Computational Science

Background:

• Three-dimensional (3D) medical imaging generates complex datasets requiring secure handling.
• Existing encryption methods for 3D medical data can be computationally intensive or lack selectivity.
• Light-field imaging offers a way to capture 3D information efficiently.

Purpose of the Study:

• To propose a selective encryption scheme for 3D medical images.
• To enhance the security of sensitive regions within 3D medical data.
• To improve the efficiency of 3D medical image encryption compared to slice-based methods.

Main Methods:

• Utilizing convolutional neural networks (CNNs) to identify salient regions of interest (ROIs) in elemental images (EIs) from 3D light-field data.
• Employing two-dimensional (2D) Moore cellular automata (MCA) with a balanced rule for encrypting the extracted ROIs.
• Reconstructing the encrypted elemental image array (EIA) into a full-color, full-parallax 3D image.

Main Results:

• Successful extraction and selective encryption of ROIs in 3D medical images.
• Demonstration of reconstruction into a viewable 3D image without requiring 3D image preprocessing.
• Achieved higher efficiency compared to traditional slice-by-slice encryption techniques.

Conclusions:

• The proposed selective encryption scheme effectively secures 3D medical image data.
• The method enables efficient and secure visualization for improved medical diagnosis and treatment planning.
• This approach offers a promising advancement in the field of medical image security and analysis.