Image omics feature selection method based on momentum adaptive harmony search

Abstract

The invention discloses a radiomics feature selection method based on momentum adaptive harmony search, and belongs to the technical field of radiomics. The method comprises the following steps of: 1, checking a region of interest (ROI) of the medical image; 2, adrenal tumor CT image data containing the ROI are read; 3, extracting original image features from the read image data; 4, defining an objective function of the method for minimizing the problem; 5, initializing a harmony memory bank; 6, initializing the maximum number of iterations, the HMCR and the PAR; 7, generating a new solution j; 8, updating the harmony memory bank; and 9, repeating the steps 7-8, and outputting an optimal feature subset. According to the method, the HMCR and the PAR are dynamically adjusted, so that sinking into a local optimal solution is avoided, and the robustness of the algorithm is enhanced; the concept of momentum gradient descent is fused, and convergence is accelerated; the fitness function is adjusted, and the essence of the feature search optimization problem is better conformed.

Description

technical field [0001] The invention belongs to the technical field of radiomics, and in particular relates to a feature selection method of radiomics based on momentum adaptive harmony search. Background technique [0002] Adrenal adenoma (AA) is the most common benign tumor of the adrenal gland. It originates from the adrenal cortex. It usually has no specific clinical symptoms. It is often found accidentally in physical examination or other examinations of the chest and abdomen. Clinically, it is divided into no and no tumors according to whether the tumor has endocrine function or not. Functional adenoma and functional adenoma, most of which are non-functioning adenoma clinically. At present, according to the different CT values ​​of adenomas, 10 HU is used as the critical point, and adenomas > 10 HU are defined as fat-poor adrenal adenoma (LP-AA). Adrenal metastases (AM) are the most common malignant tumors of the adrenal glands, second only to benign AA in incidenc...

AI Technical Summary

Problems solved by technology

The current feature selection methods are often difficult to consider the correlation between features, or the calculation cost is high, and feature selection is essentially an optimization problem, so it is feasible to use the harmony search algorithm to solve such problems

However, discrete harmony search tends to fall into local optimum when solving such problems, which reduces the efficiency and accuracy of search.

Images