Development of lung cancer prediction system using meta-heuristic optimized deep learning model

Mohamed Shakeel, Pethuraj (2023) Development of lung cancer prediction system using meta-heuristic optimized deep learning model. Doctoral thesis, Universiti Teknikal Malaysia Melaka.

	Text (24 Pages) Development of lung cancer prediction system using meta-heuristic optimized deep learning model.pdf - Submitted Version Download (508kB)
	Text (Full Text) Development of lung cancer prediction system using meta-heuristic optimized deep learning model.pdf - Submitted Version Restricted to Registered users only Download (5MB)

Abstract

Lung cancer is a serious disease that completely affects the human respiratory system for both men and women. The lung cancer symptoms create complexity in detecting lung cancer in the earlier stage. For this purpose, an automatic computer-aided lung cancer detection system is required to minimize the mortality rate by recognizing it in an earlier stage. The traditional system fails to predict the accurate affected cancer region with minimum computation complexity and error rate. For overcoming the difficulties, effective and optimized meta-heuristic machine learning techniques indicate lung cancer in the earlier stage. The proposed optimized automatic lung cancer prediction process minimizes the entire miss-classification and improves lung cancer recognition accuracy. The machine learning techniques are used to predict lung cancer in this study. First, the Computed Tomography (CT) images obtained from the Cancer Imaging Archive (CIA) dataset processed with the help of a weighted mean histogram equalization approach that used to eliminate the noise information from CT image. After that cancer-affected region in the lung is segmented with the help of the proposed Butterfly Optimization Algorithm-based K-Means Clustering (BOAKMC) algorithm. The algorithm detects the affected region depending on pixel similarity computation process. Then different features are derived from the segmented region using Gray Intensity Co-Occurrence Distribution Matrix (GICDM) which is processed by applying a proposed Supervised Jaya Optimized Rough Set based Feature Selection (SJORSFS) algorithm. These algorithms select the best features according to the fitness value, and its redundancy is to be reduced. Finally, the classification is implemented using an ensemble classifier, deep learning instantaneously trained a neural network and an Autoencoder-based Recurrent Neural Network (ARNN) classification algorithm. The proposed lung cancer prediction model recognizes the lung cancer up to 96.39% of accuracy, 0.981% of precision value, 0.9839% of F1-score, 6.438% of false positive rate and 448.607ms of classification time.

Item Type:	Thesis (Doctoral)
Uncontrolled Keywords:	Lung cancer, Respiratory system, Automatic detection system
Subjects:	T Technology > T Technology (General) T Technology > TA Engineering (General). Civil engineering (General)
Divisions:	Library > Tesis > FKP
Depositing User:	MUHAMAD HAFEEZ ZAINUDIN
Date Deposited:	16 Dec 2024 08:30
Last Modified:	16 Dec 2024 08:30
URI:	http://eprints.utem.edu.my/id/eprint/28305
Statistic Details:	View Download Statistic

Actions (login required)

View Item