Dual-Layer Image Model for Lesion Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current medical image processing algorithms struggle to accurately identify and mark the specific position of lesions on pathological tissues, as they often rely on rough markings that do not accurately reflect the true position of lesions.
Innovation Solution
An image processing model generation method that includes an initial image processing model with a classification layer and a marking layer, which iteratively updates parameters based on loss values calculated from classification and position loss values to refine the identification and marking of lesion regions, ensuring accurate detection and marking of disease types in medical images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current medical image processing algorithms are used, then the processing speed is relatively fast, but the marking precision of lesion regions is poor
Solution Approach 1:
The model is divided into two independent but coordinated output branches: a classification layer for disease type identification and a marking layer for lesion region localization. This segmentation allows each layer to specialize in one task while sharing the same feature extraction backbone, thereby improving marking precision without proportionally increasing overall complexity
Solution Approach 2:
The shared convolutional neural network backbone serves dual purposes: extracting features for both classification and marking tasks. This multi-functionality reduces redundant computation and model parameters, achieving improved marking precision while controlling the increase in device complexity
2Productivity
If manual lesion marking is performed, then the marking precision is high, but the processing time and labor cost increase
Solution Approach 1:
The model enables automated self-marking of lesion regions by learning to predict bounding box coordinates directly from input images. The marking layer automatically generates precise lesion markings without human intervention, significantly improving processing efficiency while maintaining high marking precision through trained neural network predictions
Solution Approach 2:
The model uses loss function feedback during training to continuously improve marking accuracy. The loss calculation compares predicted markings with ground truth annotations, providing gradient feedback that adjusts model parameters to reduce marking errors, thereby achieving high precision automated marking that replaces manual processes
3Measurement precision
If the model uses more parameters for better accuracy, then the detection precision improves, but the training time and computational resources increase
Solution Approach 1:
The classification and marking tasks are merged into a single unified model with shared feature extraction layers. This combination allows the model to learn complementary representations that improve detection accuracy while avoiding the redundant training time that would result from separate models, as the shared backbone is trained once to serve both purposes
Data Source
AI summary
An image processing model generation method includes: inputting at least one training sample lesion image into an initial image processing model, the initial image processing model including a classification layer and a marking layer; calling the classification layer; calling the marking layer; obtaining a loss value of the at least one training sample lesion image in the initial image processing model; determining whether the loss value is within a preset range; if not, updating parameters of the initial image processing model, an image processing model with updated parameters being used as an initial image processing model in next training; and repeating above steps until the loss value is within the preset range.