Intelligent image segmentation prior to optical character recognition (OCR)
Inventors
Lemay, Paul Roland • Agnello, Alessandro Simone
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12322197-B2
Publication Date
2025-06-03
Expiration Date
Abstract
A medical device monitoring system and method extract information from screen images from medical device controllers, with a single OCR process invocation per screen image, despite critical information appearing in different screen locations, depending on which medical device controller's screen image is processed. For example, different software versions of the medical device controllers might display the same type of information in different screen locations. Copies of the critical screen information, one copy from each different screen location, are made in a mosaic image, and then the mosaic image is OCR processed to produce text results. Text is selectively extracted from the OCR text results, depending on contents of a selector field on the screen image, such as a software version number or a heart pump model identifier.
Core Innovation
The invention relates to a medical device monitoring system in which a server requests and receives, via a computer network, an image of contents displayed on a screen of a medical device controller. Instead of using multiple OCR runs per screen image, the server generates a mosaic image by copying different source regions that encompass different screen coordinates into predefined destination regions at different coordinates in the mosaic image. This mosaic generation creates mapped destination regions so that OCR outputs can be associated with specific coordinate locations.
After the mosaic image is generated, the system has the mosaic image optical-character-recognized to produce text results. The text results comprise a first text result corresponding to the destination region at the fourth coordinates, a second text result corresponding to the destination region at the fifth coordinates, and a selector text result corresponding to the destination region at the sixth coordinates. The selector text result determines which text result is selected, i.e., one of the first text result or the second text result is selected based on the selector text result.
The described mosaic copying approach supports handling different on-screen text locations across medical device controller outputs by arranging multiple screen regions into a single mosaic image for OCR processing. The disclosed processing also includes optional mosaic pre-processing behaviors, including magnifying a source region to fill a larger destination region and converting colored pixels to white pixels and using black pixel thresholding to improve OCR contrast. The document further includes validation and cleanup of OCR outputs and storing and using the extracted results for remote monitoring and alarms.
Claims Coverage
The independent claims cover a medical device monitoring system, a method, and a non-transitory computer-readable medium. Across these, the claims define four main inventive features: automated receipt of an on-screen image, mosaic generation by copying three defined source regions into predefined destination regions at different coordinates, OCR producing coordinate-corresponding text results including a selector text result, and selection of one of two text results based on the selector text result.
Mosaic image generated by copying source regions into predefined destination regions
Generate a mosaic image by copying a first source region encompassing first screen coordinates to a predefined destination region at fourth coordinates, a second source region encompassing second screen coordinates to a predefined destination region at fifth coordinates different from the fourth, and a third source region encompassing third screen coordinates to a predefined destination region at sixth coordinates different from the fourth and fifth.
OCR producing coordinate-corresponding text results including a selector text result
After generating the mosaic image, have the mosaic image optical-character-recognized to produce text results comprising a first text result corresponding to the fourth coordinates, a second text result corresponding to the fifth coordinates, and a selector text result corresponding to the sixth coordinates.
Selecting one of the OCR text results based on the selector text result
Select one of the first text result or the second text result based on the selector text result.
Requesting and receiving an image of contents displayed on a medical device controller screen
Request and receive, via a computer network from a medical device controller, an image of contents displayed on a screen of the medical device controller.
Overall, the independent claims require receiving an on-screen image, constructing a mosaic by copying three source regions into distinct destination coordinates, running OCR on the mosaic to obtain two coordinate-linked text results plus a selector text result, and selecting between the first and second text results according to the selector text result. Dependent claims additionally refine selector content and mosaic generation and pixel handling behaviors, including magnifying to fill a larger destination region and converting non-black pixels to white pixels.
Stated Advantages
Avoids multiple OCR runs per screen image by processing mapped destination regions in a mosaic.
Enables selecting the relevant extracted text result based on a selector text result.
Improves OCR contrast via optional conversion of colored pixels to white pixels and related pixel thresholding.
Supports monitoring with remote use of extracted text results for monitoring stations and alarms.
Documented Applications
Medical device monitoring using remote monitoring stations and alarms based on selected OCR-derived text results.
Interested in licensing this patent?