Medical Image Reconstruction With Encrypted Slice Shuffling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Medical image data used in PACS systems cannot be directly utilized for training deep learning algorithms, and there are security concerns regarding personal information and body images, necessitating a system for pre-processing and post-processing, as well as encryption for secure transmission.
Innovation Solution
A medical image reconstruction and transmission method using an artificial neural network model that encrypts and decrypts medical image data, shuffles image slices, and classifies data based on feature information for secure and high-quality image reconstruction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If medical image data is transmitted between PACS and external servers for deep learning processing, then advanced medical determination capabilities are improved, but security risks of personal information leakage increase
Solution Approach 1:
The patent applies preliminary action by performing encryption on medical image data before transmission to external servers. The encryption process converts original medical images into encrypted representations that preserve structural information needed for deep learning analysis while removing personally identifiable information. This preliminary security measure ensures that even if data is intercepted during transmission, personal information cannot be accessed
Solution Approach 2:
The patent introduces an intermediary encryption mechanism that acts as a mediator between the PACS system and external deep learning servers. The encryption process serves as an intermediary layer that transforms medical image data into a form that can be processed by external algorithms while preventing direct access to personal information. The encrypted data structure maintains the necessary features for medical determination without exposing sensitive patient identifiers
2Object-affected harmful factors
If medical image data is encrypted through pseudonymization before transmission, then security against personal information leakage is improved, but the complexity of pre-processing and post-processing increases
Solution Approach 1:
The patent applies parameter changes by modifying the data representation parameters through encryption while maintaining the essential structural parameters needed for medical image analysis. The encryption process changes parameters such as pixel values and data formatting, but preserves spatial relationships and anatomical features that deep learning algorithms require. This selective parameter transformation achieves security without requiring complete reprocessing of the medical images
3Measurement precision
If deep learning algorithms are trained with massive medical image data, then medical determination accuracy is improved, but the requirement for pre-processing and post-processing systems increases
Solution Approach 1:
The patent applies preliminary action by implementing automated pre-processing routines that prepare medical image data for deep learning training. The system automatically performs encryption, data formatting, and quality control checks before data is fed into training algorithms. This preliminary preparation reduces the manual pre-processing burden and ensures data is ready for immediate processing by deep learning systems
Solution Approach 2:
The patent implements self-service mechanisms where the encryption and pre-processing system automatically manages the complexity of data preparation. The system includes automated workflows that handle data encryption, validation, and formatting without requiring extensive manual intervention. This self-service approach reduces the operational complexity despite the sophisticated processing requirements
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A medical image reconstruction and transmission method employing a medical image reconstruction and transmission system according to an embodiment of the present invention includes receiving medical image data including at least one of k-space data obtained through accelerated imaging and digital imaging and communications in medicine (DICOM) data generated on the basis of the k-space data obtained through accelerated imaging, reconstructing the received medical image data using an artificial neural network model, and transmitting the reconstructed medical image data on the basis of an address from which the medical image data is received.