A New Feature Selection Method Based on Hybrid Approach for Colorectal Cancer Histology Classification

Colorectal cancer (CRC) is one of the most common malignant cancers worldwide. To reduce mortality, early diagnosis and treatment are essential in leading to a greater improvement survival length patients. In this paper, hybrid feature selection technique (RF-GWO) based on random forest (RF) algorithm gray wolf optimization (GWO) was proposed for handling high dimensional redundant datasets colorectal (CRC). Feature aims properly select minimal relevant subset features out vast amount complex noisy data reach classification accuracy. Gray were utilized find suitable histological images human dataset. Then, best-selected features, artificial neural networks (ANNs) classifier applied classify multiclass texture analysis cancer. A comparison between GWO another optimizer particle swarm (PSO) also conducted determine which successful enhancement RF algorithm. Furthermore, it crucial an having capability removing attaining optimal therefore achieving CRC performance terms accuracy, precision, sensitivity rates. The Heidelberg University Medical Center Pathology archive used check method found outperform benchmark approaches. results revealed that (GWO-RF) has outperformed other state art methods where achieved overall rates 98.74%, 98.88%, 98.63%, respectively.