Systems and methods for de-identifying medical and healthcare data
Inventors
Rosenberg, Michael • BENITZ, MALCOLM • SUTTLES, JASON • WOODLIEF, CHRIS • GOLDSTEIN, BRENT
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-11270027-B2
Publication Date
2022-03-08
Expiration Date
Abstract
The invention relates generally to systems and methods for protecting patient privacy when health care information is shared between various entities and, in particular, to systems and methods that implement a multi-stage sanitizing routine for de-identifying patient data from medical reports and diagnostic images to ensure patient privacy, while preserving the ability for sanitized medical reports and diagnostic images to be re-identified.
Core Innovation
The invention provides a method for de-identifying medical data that preserves re-identification while deleting protected health information. A server retrieves a medical image to be de-identified, determines a modality associated with the medical image, and retrieves a de-identification profile for the modality that specifies at least one area of the medical image containing patient information.
A sanitizing engine coupled to the server applies a blackout zone over the area specified in the de-identification profile and deletes the patient information within the blackout zone. After the blackout zone is applied, optical character recognition is performed in the area to determine whether any characters are detected, and if no characters are detected, a boundary for a diagnostic portion or a gradient boundary for a diagnostic portion is detected.
The sanitizing engine then detects whether non-black pixels are present outside the boundary and converts the non-black pixels outside the boundary to black pixels, or encapsulates the medical image into a DICOM format if non-black pixels are not detected outside the boundary. The workflow includes detecting and processing patient-information content in diagnostic images, quality control by optical character recognition, and re-encapsulating into DICOM format for sharing, with re-identification enabled via stored DICOM header data and parseable document data linked using unique identification strings generated with a hexadecimal salt hashing mechanism.
Claims Coverage
The independent claims cover three closely related inventive features: modality-specific de-identification profiles for medical images, previously generated de-identification profile handling with gradient boundary conversion, and a system that applies blackout-zone sanitization and transmits a DICOM file. Across the claims, the focus is on deleting patient information in a defined area, optical character recognition-based character detection, boundary detection for a diagnostic portion, pixel conversion outside the boundary, and DICOM encapsulation, with optional linkage supporting re-identification via stored header data.
Modality-specific blackout-zone sanitization with OCR-based character detection and diagnostic boundary handling
Retrieving a medical image of patient information, determining a modality, retrieving a de-identification profile for the modality that specifies an area containing patient information, applying a blackout zone over the specified area to delete patient information, performing optical character recognition in the area after the blackout zone is applied, determining if any characters are detected, detecting a boundary for a diagnostic portion if no characters are detected, detecting if non-black pixels are present outside the boundary, and converting non-black pixels outside the boundary to black pixels or encapsulating the medical image into a DICOM format when non-black pixels are not detected outside the boundary.
Previously generated de-identification profile with gradient boundary conversion of characters outside the diagnostic portion
Retrieving a medical image of patient information, applying a sanitizing process in an area determined by a previously generated de-identification profile that specifies the area containing patient information and wherein the sanitizing process deletes the patient information, detecting if any characters are present in the area after the sanitizing process, detecting a gradient boundary for a diagnostic portion if no characters are detected, and converting pixels of any characters detected outside of the gradient boundary to an inverse color.
System with blackout-zone deletion, gradient boundary pixel conversion, DICOM encapsulation, and transmission to a remote computing system
A database configured to store a medical record including a medical image and a medical report; a sanitizing engine communicatively coupled to the database and configured to import the medical record, apply a blackout zone over an area containing patient information, delete the patient information within the blackout zone, detect if characters exist in the area where the blackout zone was applied, detect a gradient boundary for a diagnostic portion, convert non-white pixels outside the gradient boundary to white pixels, and encapsulate the medical image into a DICOM file; and a server communicatively coupled to the database and the sanitizing engine configured to transmit the DICOM file to a remote computing system.
The claims collectively cover server-driven de-identification of medical images using a de-identification profile to define a patient-information area, sanitizing that area with a blackout zone and deletion, using optical character recognition to detect remaining characters, detecting a diagnostic portion boundary including a gradient boundary, and converting pixels outside the diagnostic portion or encapsulating the sanitized image as a DICOM file. The system claim additionally includes a database, sanitizing engine, DICOM encapsulation, and transmission of the DICOM file to a remote computing system, with linkage supporting re-identification via stored DICOM header data and identifiers linked through a hexadecimal salt hashing mechanism.
Stated Advantages
Preserves re-identification while deleting protected health information by enabling re-identification through stored DICOM header data and parseable document data linked using unique identification strings.
Deletes patient information within a blackout zone specified by a modality-associated de-identification profile.
Uses optical character recognition after blackout sanitization to determine whether any characters remain.
Ensures that pixels outside a detected diagnostic boundary are converted to black pixels or white pixels, or that character pixels outside a gradient boundary are converted to an inverse color.
Encapsulates the de-identified medical image into a DICOM format for transmission/sharing.
Documented Applications
De-identifying medical data for sharing by encapsulating sanitized medical images into a DICOM file and transmitting the DICOM file to a remote computing system.
Re-identification-enabled workflows for de-identifying medical data while preserving a path to re-identification through stored header/parseable document linkage.
Interested in licensing this patent?