Neutrosophic Logic Empowered Machine Learning Algorithm with Salp Swarm Optimization for Biomedical Image Analysis

Adwan A. Alanazi¹, Abdelgalal O. I. Abaker^2,*, Sayed Abdel-Khalek³, Fahad Mohammed Alhomayani^4,5,
M. Aripov⁶

¹Department of Computer Science and Information, University of Hail, Saudi Arabia

²Applied College, Khamis Mushait, King Khalid University, Abha, Saudi Arabia

³Department of Mathematics and Statistics, College of Science, Taif University, P. O Box 11099, Taif 21944, Saudi Arabia

⁴College of Computers and Information Technology, P.O. Box 11099, Taif University, Taif 21944, Saudi Arabia

⁵Applied College, P.O. Box 11099, Taif University, Taif 21944, Saudi Arabia0

^6,1Department of Applied Mathematics and Computer Analysis, Faculty of Mathematics, NUU, Uzbekistan

Emails: a.alanazi@uoh.edu.sa, aoadrees@kku.edu.sa, sayedquantum@yahoo.co.uk, fahad@tu.edu.sa, aripovmersaid@gmail.com

^*Corresponding Author: aoadrees@kku.edu.sa

Abstract

Leukemia recognition and classification contain the identification of dissimilar kinds of leukemia, a group of blood cancers that affects the bone marrow and blood. A classical model containing microscopic analysis of blood smears to classify abnormal cells analytic of leukemia. Leukemia recognition employing a united technique of neutrosophic logic and deep learning (DL) signifies a new and complete approach to handling uncertainty and difficulty in medical data. Neutrosophic logic permits the representation of unstated or imperfect data, which is general in medical analyses. DL mainly convolutional neural networks (CNN) or recurrent neural networks (RNN), which can mechanically remove difficult patterns from medicinal imageries, improving the accuracy of leukemia recognition. The neutrosophic logic module accommodates the characteristic uncertainty in medicinal data, offering a formalism to manage imperfect or inaccurate data linked with the analysis procedure. The combination of these dual techniques generates a robust structure which capable of leveraging both the control of DL in image analysis and the flexibility of neutrosophic logic in dealing with uncertainties, contributing to more trustworthy and interpretable leukemia recognition methods.  This study develops a new Salp Swarm Algorithm with a Neutrosophic Logic SVM (SSA-NSVM) model for Leukemia Detection and Classification. The SSA-NSVM technique mainly exploits Neutrosophic Logic (NL) concepts with the DL model for the detection of leukemia. To attain this, the SSA-NSVM model uses bilateral filtering (BF) based image pre-processing. In addition, the SSA-NSVM approach applies a modified densely connected networks (DenseNet) technique for learning complex and intrinsic feature patterns. Besides, the hyperparameter range of the modified DenseNet system takes place utilizing a SSA. At last, the NSVM technique is employed for the detection and identification of leukemia. The performance validation of the SSA-NSVM algorithm is verified utilizing a benchmark medicinal image dataset. The simulation values emphasized that the SSA-NSVM model reaches better detection outcomes than other existing approaches.

Keywords: Leukemia Detection; Neutrosophic Logic; Bone Marrow; Salp Swarm Algorithm; Blood Cancer