Medical Image Focus Detection via Sliding Window and Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computer-aided detection technologies for medical images face challenges in achieving both accuracy and efficiency, often resulting in high false positives and slowed detection speeds due to excessive Region of Interest (ROI) or Volume of Interest (VOI) overlap.
Innovation Solution
A focus detection apparatus and method that utilize a sliding window with a side length at least twice the target focus size and a stride not larger than the target focus size, combined with machine learning techniques, candidate aggregation, and multi-size adjustments to improve detection accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the overlap range of different ROI or VOI is increased to improve detection accuracy, then detection accuracy is improved, but the number of ROI or VOI increases excessively causing detection speed to slow down
Solution Approach 1:
The patent changes the parameters of the sliding window (size and stride) to optimize the balance between detection accuracy and speed. By setting the window size to at least twice the target focus size and the stride to not larger than the target focus size, the system achieves effective focus coverage without excessive ROI generation
Solution Approach 2:
The patent applies partial action by using a sliding window size that is at least twice the target focus size, which provides sufficient coverage to detect the focus while avoiding the excessive overlap that would generate too many ROIs and slow down detection
2Measurement precision
If the sliding window size is increased to ensure complete focus coverage, then detection accuracy is improved, but the number of ROIs generated increases causing processing time to increase
Solution Approach 1:
The patent optimizes the sliding window parameters by setting the window size to at least twice the target focus size and the stride to not larger than the target focus size. This parameter configuration ensures complete focus coverage while minimizing the number of ROIs generated, thus reducing processing time
3Measurement precision
If the stride of the sliding window is decreased to improve focus detection precision, then detection accuracy is improved, but the number of scan positions increases causing detection efficiency to decrease
Solution Approach 1:
The patent sets the stride parameter to not larger than the target focus size, which provides sufficient detection precision while avoiding excessively small stride values that would create too many scan positions and reduce detection efficiency
Data Source
AI summary
A focus detection apparatus and a method thereof are provided. In the method, a medical image is obtained. Size of a target focus and a sliding window are determined, and side length of the sliding windows is at least twice the side length of the target focus. The medical image is scanned through the sliding window, and a stride which the sliding windows moves each time is not greater than the side length of the target focus. At least one area of interest is obtained based on the scan result. Then, the area of interest is identified based on machine learning techniques, and perform candidate aggregation and multiple size aggregation to determine at least one focus position. Accordingly, the computational time and the detection accuracy can be improved.