A new hybrid fuzzy bio-inspired classifier for cancer detection using cuckoo optimization and hyper-planes

Purpose Various methods are used for cancer detection such as genetic tests, scanning, MRI, mammography, etc. These methods help collect data on patients, which can be utilized for comparing a new patient’s information with the aggregated data to detect cancer. The main step in this process is data classification. There are several cancer detection methods with their own disadvantages in flexibility, non-linear complexity and sensitive in imbalance data. In this paper, a new fuzzy bio-inspired based classification method is designed to classify the imbalance medical data. Design/methodology/approach In this paper, a new fuzzy bio-inspired-based classification method is designed to classify the imbalance of medical data. The method consists of a new fuzzy draft of the Cuckoo Optimization Algorithm (COA) and separating hyper-planes based on assigning binary codes to separated regions that are called Hyper-Planes Classifier (HPC). Based on the technical review is done in the paper, the HPC has a better structural superiority than the other classification algorithms. The Fuzzy Cuckoo Optimization Algorithm (FCOA), which fills up its challenge in proper tuning parameters, is proposed to optimize the weights of the separating hyper-planes with linear complexity time. Findings The experimental results were presented in five steps. Step1, the details of the average and the best results of the proposed methods were reported and compared. Step2, the quality of the detection methods with different numbers of hyper-planes were compared. The obtained weights of different numbers of hyper-planes were reported in Step3. Step4, the convergence process of the FCOA and the COA were shown. Step5, the best obtained results were compared with the best reported one in previous literature. The experimental results and the presented comparisons show that the proposed hybrid detection method is comparable to other methods and operates better than them in most cases. Originality/value A technical review has been done based on classifying the applied classification methods to cancer detection and analyzing advantages (+) and disadvantages (−) of the methods and their optimizer algorithms. A new fuzzy draft of COA has been designed to dynamically tuning the Egg Laying Radius based on a fuzzy inference system with four fuzzy rules. A novel hybridization of the hyper-planes classification method and the designed FCOA has been proposed to optimize the hyper-planes' weights. The effectiveness of the proposed hybridization has been examined in famous UCI cancer datasets based on one, two, three and four hyper-planes and compared with more than 30 previous researches.