- Afyon Kocatepe Üniversitesi Fen Ve Mühendislik Bilimleri Dergisi
- Vol: 23 Issue: 4
- Classification of T-ALL, B-ALL and T-LL Malignancies Using Adaptive Network-Based Fuzzy Inference Sy...
Classification of T-ALL, B-ALL and T-LL Malignancies Using Adaptive Network-Based Fuzzy Inference System Approach Combined with Nature-Inspired Optimization on Microarray Dataset
Authors : Fatma Akalin, Nejat Yumuşak
Pages : 941-954
Doi:10.35414/akufemubid.1259929
View : 55 | Download : 47
Publication Date : 2023-08-31
Article Type : Research Article
Abstract :Leukemia is the formation of cancer with different characteristic findings. According to the progress type of disease in the body is called acute or chronic. Acute leukemias are characterized by the presence of blast cells that proliferate uncontrollably in the bone marrow and then go into the blood and tissues. Determination of T/B or non T/B cell class is important in the immunophenotypic evaluation related to subtypes of blast cells. Because the diagnosis and treatment processes of B-ALL, T-ALL and T-LL subtypes, which are composed of B and T cell lines, are different. Therefore, correct diagnosis is vital. In this study, the molecular diagnosis was provided for the accurate detection of T-ALL, B-ALL and T-LL subtypes through microarray datasets. But, microarray datasets have a multidimensional structure. Because it contains information related to the disease as well as information not related to the disease. This situation also affects the training situation and computational cost of the model. For this, the whale optimization algorithm was used in the first stage of the study. Thus, related genes were selected from the data set. Secondly, the selected potential genes were given as input to the ANFIS structure. Then, in order to improve the inference power, parameter optimization related to the membership function of the ANFIS structure was provided with ABC and PSO optimization algorithms. Finally, the predictions obtained from the ANFIS, ANFIS+ABC, and ANFIS+PSO methods for each sample were classified using the logistic regression algorithm and, an accuracy rate of 86.6% was obtained.Keywords : Mikrodizi Veri Kümesi, Metasezgisel Optimizasyon Algoritmaları, Uyarlanabilir Ağ Tabanlı Bulanık Çıkarım Sistemi, Lojistik Regresyon