Deep Learning BPE Assessment in Breast Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing manual workflow for assessing background parenchymal enhancement (BPE) in breast imaging is time-consuming, resource-intensive, and prone to high variability between radiologists, leading to inconsistent diagnoses.
Innovation Solution
A deep learning (DL) model is trained on various types of medical images to automatically assess BPE levels in breast images, reducing the reliance on manual radiological interpretation and enhancing consistency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual workflow is used for assessing BPE, then radiologists can perform the assessment, but the process is time-consuming and resource-intensive
Solution Approach 1:
The patent replaces the manual mechanical process of radiological interpretation with an automated deep learning system. The DL model automatically assesses BPE levels in breast images, eliminating the need for radiologists to manually analyze each image, thereby reducing time consumption while maintaining assessment reliability through consistent algorithmic evaluation
Solution Approach 2:
The system enables self-service by allowing the deep learning model to autonomously perform BPE assessment without requiring human intervention. The model independently processes images, evaluates enhancement patterns, and provides diagnoses, freeing radiologists from routine tasks and reducing overall time requirements
2Reliability
If manual workflow is used for assessing BPE, then radiologists can perform the assessment, but there is high variability between radiologists leading to inconsistent diagnoses
Solution Approach 1:
The patent applies homogeneity by using a standardized deep learning algorithm that provides consistent BPE assessment results across different cases and operators. The model maintains uniform evaluation criteria and decision-making logic, eliminating the variability inherent in manual interpretation by different radiologists and ensuring diagnostic consistency
Solution Approach 2:
The system replaces variable human interpretation with a fixed, reproducible computational algorithm. The deep learning model processes images through consistent neural network layers and activation functions, ensuring that the same input image always produces the same BPE assessment, thereby eliminating inter-radiologist variability
3Reliability
If manual workflow is used for assessing BPE, then radiologists can perform the assessment, but imaging system resources are over-utilized
Solution Approach 1:
The patent extracts the BPE assessment function from the manual radiological workflow and implements it through a dedicated deep learning system. This separation allows the imaging system resources to be optimized for image acquisition while the computational burden of assessment is handled by the DL model, reducing overall resource consumption
Solution Approach 2:
The system substitutes manual radiological assessment with automated computational processing. The deep learning model efficiently analyzes image data through parallel neural network operations, consuming significantly fewer resources compared to manual review while maintaining or improving assessment accuracy
Data Source
AI summary
Methods and systems are provided for automatically assessing a level of BPE of a patient of an imaging system based on one or more medical images of one or more breasts of the patient, using a deep learning (DL) model. The medical images may include contrast enhanced mammography (CEM) images, magnetic resonance (MR) images, or a different type of images. The one or more images may include one or more images of a same breast, where the BPE assessment outputted by the DL model may include a score, such as a percentage of BPE detected in the images. The one or more images may include images of a left breast and a right breast of the patient, where the BPE assessment outputted by the DL model may include whether an asymmetry between BPE assessments of the left breast and the right breast is detected.