Lung Point Detection Using M-Mode Ultrasound Classification

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for assessing lung sliding in ultrasound imagery struggle to accurately and efficiently detect the absence of lung sliding, which is indicative of life-threatening conditions such as pneumothorax, especially in emergency situations.

Innovation Solution

A computing system processes B-mode video frames to generate M-mode images, applies image classifiers and machine learning models to determine the presence or absence of lung sliding, and uses a clip prediction module to output a binary class prediction, incorporating techniques like pleural line detection, Radon Transform, and convolutional neural networks to enhance accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If automated processing of B-mode video frames is implemented to generate M-mode images and apply image classifiers, then detection speed and reliability are improved, but device complexity increases

Engineering Contradiction:
Improvedetection reliabilityVSAvoidsystem complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The system segments the lung ultrasound video into individual B-mode video frames, then processes each frame to generate M-mode images. This segmentation allows the complex detection task to be broken down into manageable steps: frame extraction, M-mode conversion, classifier application, and result aggregation, improving reliability while keeping each processing stage modular and manageable

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

M-mode images serve as an intermediary representation between the raw B-mode video frames and the final classification output. The system converts B-mode frames to M-mode images, which capture the temporal dynamics of lung sliding more effectively, then applies image classifiers to these intermediate representations. This intermediary step enhances detection reliability by transforming the data into a more suitable format for classification

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If machine learning models are used to detect pleural line and classify M-mode images, then measurement precision is improved, but computing time and processing complexity increase

Engineering Contradiction:
Improvepleural line detection precisionVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary processing by converting B-mode video frames to M-mode images before applying the image classifier. This preliminary transformation prepares the data in an optimal format that enhances the precision of subsequent pleural line detection and lung sliding classification, while the pre-processing step can be performed offline or in advance

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system creates M-mode images as simplified representations or copies of the complex B-mode video data. These M-mode copies capture the essential temporal patterns of lung sliding in a compressed format, allowing classifiers to achieve high measurement precision while processing smaller, more manageable data structures that require less computing time

Inventive Principle:
Principle #26Copying

3Measurement precision

If multiple processing steps including Radon Transform and contour finding are applied, then detection accuracy is improved, but device complexity and processing duration increase

Engineering Contradiction:
Improvelung sliding detection accuracyVSAvoidprocessing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The detection process is segmented into distinct processing steps: Radon Transform application, thresholding, contour finding, and classification. Each step performs a specific function that contributes to overall detection accuracy. This segmentation allows the system to apply multiple sophisticated processing techniques while maintaining modularity, where each component can be independently optimized and validated

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system applies a continuous sequence of processing operations where each step builds upon the previous one: B-mode frames are continuously transformed to M-mode images, which then undergo Radon Transform, thresholding, contour finding, and finally classification. This continuous pipeline ensures that useful information is preserved and enhanced through each processing stage, improving detection accuracy while maintaining efficient data flow

Inventive Principle:
Principle #20Continuity of useful action

Data Source

PatentEP4542485B1Systems and methods for detecting lung point
Publication Date: 2026.02.25 DEEP BREATHE
  • EP4542485B1 patent drawingFigure 1
  • EP4542485B1 patent drawingFigure 2
  • EP4542485B1 patent drawingFigure 3

AI summary

Systems and methods for detecting an absent lung sliding condition from B-mode or M-mode ultrasound imaging. Pleural line detection and M-mode designation are performed to enhance classifier performance. A convolutional neural network binary classifier operates to predict lung sliding or absent lung sliding for a series of clips. A further algorithm uses constituent M-mode-level prediction confidences from the series of clips to produce a final prediction.