Pixel Scoring for Medical Image Anomaly Detection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current image enhancement techniques in medical imaging fail to effectively detect faint anomalies and imperceptible details, necessitating improved methods for enhancing images to aid in medical anomaly and disease detection.

Innovation Solution

A machine-implemented method that scores pixels in images based on neighborhood relationships, adjusting their values to reveal hidden data, using a system comprising a receiver module, score generation module, pixel value adjustment module, and image adjustment module to generate enhanced images.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If current image enhancement techniques are used, then general image processing is achieved, but faint anomalies and imperceptible details cannot be detected

Engineering Contradiction:
Improvedetection precisionVSAvoidloss of faint details
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent applies local quality by making different parts of the image processing have different functions. Specifically, pixels that are determined to be part of anomalies receive different enhancement treatment compared to normal pixels. The system calculates anomaly scores for different regions and applies selective enhancement, thereby improving detection precision for faint anomalies while preserving overall image quality.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent changes parameters dynamically based on local image characteristics. The enhancement amount is adjusted according to calculated anomaly scores, which are derived from comparing pixel values with their neighborhoods. This parameter adaptation allows the system to enhance faint anomalies selectively without uniformly processing the entire image, thus improving detection capability.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If pixel values are adjusted to reveal hidden data, then image enhancement is improved, but noise and artifacts may be introduced

Engineering Contradiction:
Improveanomaly detection capabilityVSAvoidnoise and artifacts
Core Design Contradiction:
Measurement precisionVSObject-affected harmful factors

Solution Approach 1:

The patent applies partial action by selectively enhancing only those pixels that are determined to be anomalies, rather than uniformly enhancing the entire image. The system calculates anomaly scores and applies enhancement proportionally to the detected anomaly regions, thereby reducing the introduction of noise and artifacts in non-anomaly areas while still improving anomaly detection capability.

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The patent uses feedback mechanisms by calculating anomaly scores based on neighborhood comparisons and using these scores to guide subsequent enhancement actions. The system iteratively processes pixels, adjusting enhancement based on the detected anomaly characteristics, which helps prevent over-enhancement and reduces artifact generation.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP2828824B1Scoring and adjusting pixels based on neighborhood relationships for revealing data in images
Publication Date: 2018.10.24 PAUL REED SMITH GUITARS LLP
  • EP2828824B1 patent drawingFigure 1A
  • EP2828824B1 patent drawingFigure 1B
  • EP2828824B1 patent drawingFigure 1C

AI summary

A machine/computer implemented system, method, and computer program product for scored pixel intensity value adjustment of a digital image is disclosed. The system is configured to obtain a digital image from data storage and perform pixel-by-pixel comparisons to generate per pixel scores. The types of comparisons include discovering minima and maxima per pixel scores by comparing to neighboring non-adjacent pixel pairs, delta pair scores by comparing to neighboring pixels, and multiple vector score types by comparing to vectors made up of individual pixels. This new information is applied to adjust each pixel's value. The system is further configured to generate a collection of such scores for a plurality of pixels in a digital linage and to generate a multi-dimensional scored pixel adjusted image. The scored pixel adjustment yields a new digital image, wherein the value of a given pixel is adjusted based on one or more of the score types.