Sparse Contribution Feature Selection and Classifiers Optimized by Concave-Convex Variation for HCC Image...

Area of Science:

• Medical image analysis
• Computational pathology
• Machine learning in oncology

Background:

• Accurate classification of hepatocellular carcinoma (HCC) images is crucial for effective pathology diagnosis and treatment planning.
• Existing classification methods may require optimization for improved accuracy in complex pathological images.

Purpose of the Study:

• To propose and evaluate a novel concave-convex variation (CCV) method for optimizing machine learning classifiers for hepatocellular carcinoma (HCC) image classification.
• To enhance the accuracy of HCC image classification using optimized random forest, support vector machine, and extreme learning machine models.

Main Methods:

• Preprocessing of hematoxylin-eosin (H&E) pathological images using a bilateral filter.
• Extraction of complete features from HCC image patches guided by pathologists.
• Development of a sparse contribution (SC) feature selection model to identify beneficial features.
• Application of the concave-convex variation (CCV) method to optimize classifier performance.

Main Results:

• The proposed CCV classifiers demonstrated significant performance improvements over original classifiers.
• CCV-optimized random forest (CCV-RF) achieved the best results in HCC image recognition among the tested methods.
• Experiments were conducted on 1260 HCC image patches, validating the effectiveness of the CCV approach.

Conclusions:

• The concave-convex variation (CCV) method is effective in enhancing the accuracy of machine learning classifiers for hepatocellular carcinoma (HCC) image classification.
• CCV-optimized random forest offers superior performance for HCC image recognition, aiding in more precise pathological diagnosis.