Health of Things Algorithms for Malignancy Level Classification of Lung Nodules
Lung cancer is one of the leading causes of death worldwide. Several computer-aided diagnosis systems have been developed to help reduce lung cancer mortality rates. This paper presents a novel structural co-occurrence matrix (SCM)-based approach to classify nodules into malignant or benign nodules...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2018-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/8322412/ |
| _version_ | 1818415948237897728 |
|---|---|
| author | Murillo B. Rodrigues Raul Victor M. Da Nobrega Shara Shami A. Alves Pedro Pedrosa Reboucas Filho Joao Batista F. Duarte Arun K. Sangaiah Victor Hugo C. De Albuquerque |
| author_facet | Murillo B. Rodrigues Raul Victor M. Da Nobrega Shara Shami A. Alves Pedro Pedrosa Reboucas Filho Joao Batista F. Duarte Arun K. Sangaiah Victor Hugo C. De Albuquerque |
| author_sort | Murillo B. Rodrigues |
| collection | DOAJ |
| description | Lung cancer is one of the leading causes of death worldwide. Several computer-aided diagnosis systems have been developed to help reduce lung cancer mortality rates. This paper presents a novel structural co-occurrence matrix (SCM)-based approach to classify nodules into malignant or benign nodules and also into their malignancy levels. The SCM technique was applied to extract features from images of nodules and classifying them into malignant or benign nodules and also into their malignancy levels. The computed tomography exams from the lung image database consortium and image database resource initiative datasets provide information concerning nodule positions and their malignancy levels. The SCM was applied on both grayscale and Hounsfield unit images with four filters, to wit, mean, Laplace, Gaussian, and Sobel filters creating eight different configurations. The classification stage used three well-known classifiers: multilayer perceptron, support vector machine, and k-nearest neighbors algorithm and applied them to two tasks: (i) to classify the nodule images into malignant or benign nodules and (ii) to classify the lung nodules into malignancy levels (1 to 5). The results of this approach were compared to four other feature extraction methods: gray-level co-occurrence matrix, local binary patterns, central moments, and statistical moments. Moreover, the results here were also compared to the results reported in the literature. Our approach outperformed the other methods in both tasks; it achieved 96.7% for both accuracy and F-Score metrics in the first task, and 74.5% accuracy and 53.2% F-Score in the second. These experimental results reveal that the SCM successfully extracted features of the nodules from the images and, therefore may be considered as a promising tool to support medical specialist to make a more precise diagnosis concerning the malignancy of lung nodules. |
| first_indexed | 2024-12-14T11:43:06Z |
| format | Article |
| id | doaj.art-d665634e73704e0e8735478757efb7ff |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-12-14T11:43:06Z |
| publishDate | 2018-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-d665634e73704e0e8735478757efb7ff2022-12-21T23:02:44ZengIEEEIEEE Access2169-35362018-01-016185921860110.1109/ACCESS.2018.28176148322412Health of Things Algorithms for Malignancy Level Classification of Lung NodulesMurillo B. Rodrigues0Raul Victor M. Da Nobrega1Shara Shami A. Alves2Pedro Pedrosa Reboucas Filho3Joao Batista F. Duarte4Arun K. Sangaiah5Victor Hugo C. De Albuquerque6https://orcid.org/0000-0003-3886-4309Laboratório de Processamento de Imagens e Simulação Computacional, Instituto Federal de Educação, Ciencia e Tecnologia do Ceará, Maracanau, BrazilLaboratório de Processamento de Imagens e Simulação Computacional, Instituto Federal de Educação, Ciencia e Tecnologia do Ceará, Maracanau, BrazilLaboratório de Processamento de Imagens e Simulação Computacional, Instituto Federal de Educação, Ciencia e Tecnologia do Ceará, Maracanau, BrazilLaboratório de Processamento de Imagens e Simulação Computacional, Instituto Federal de Educação, Ciencia e Tecnologia do Ceará, Maracanau, BrazilPrograma de Pós-Graduação em Informática Aplicada, Universidade de Fortaleza, Fortaleza, BrazilSchool of Computing Science and Engineering, Vellore Institute of Technology (VIT), Vellore, IndiaPrograma de Pós-Graduação em Informática Aplicada, Universidade de Fortaleza, Fortaleza, BrazilLung cancer is one of the leading causes of death worldwide. Several computer-aided diagnosis systems have been developed to help reduce lung cancer mortality rates. This paper presents a novel structural co-occurrence matrix (SCM)-based approach to classify nodules into malignant or benign nodules and also into their malignancy levels. The SCM technique was applied to extract features from images of nodules and classifying them into malignant or benign nodules and also into their malignancy levels. The computed tomography exams from the lung image database consortium and image database resource initiative datasets provide information concerning nodule positions and their malignancy levels. The SCM was applied on both grayscale and Hounsfield unit images with four filters, to wit, mean, Laplace, Gaussian, and Sobel filters creating eight different configurations. The classification stage used three well-known classifiers: multilayer perceptron, support vector machine, and k-nearest neighbors algorithm and applied them to two tasks: (i) to classify the nodule images into malignant or benign nodules and (ii) to classify the lung nodules into malignancy levels (1 to 5). The results of this approach were compared to four other feature extraction methods: gray-level co-occurrence matrix, local binary patterns, central moments, and statistical moments. Moreover, the results here were also compared to the results reported in the literature. Our approach outperformed the other methods in both tasks; it achieved 96.7% for both accuracy and F-Score metrics in the first task, and 74.5% accuracy and 53.2% F-Score in the second. These experimental results reveal that the SCM successfully extracted features of the nodules from the images and, therefore may be considered as a promising tool to support medical specialist to make a more precise diagnosis concerning the malignancy of lung nodules.https://ieeexplore.ieee.org/document/8322412/Computer-aided diagnosispulmonary noduleslung cancertextural featuresstructural co-occurrence matrixmalignancy classification |
| spellingShingle | Murillo B. Rodrigues Raul Victor M. Da Nobrega Shara Shami A. Alves Pedro Pedrosa Reboucas Filho Joao Batista F. Duarte Arun K. Sangaiah Victor Hugo C. De Albuquerque Health of Things Algorithms for Malignancy Level Classification of Lung Nodules IEEE Access Computer-aided diagnosis pulmonary nodules lung cancer textural features structural co-occurrence matrix malignancy classification |
| title | Health of Things Algorithms for Malignancy Level Classification of Lung Nodules |
| title_full | Health of Things Algorithms for Malignancy Level Classification of Lung Nodules |
| title_fullStr | Health of Things Algorithms for Malignancy Level Classification of Lung Nodules |
| title_full_unstemmed | Health of Things Algorithms for Malignancy Level Classification of Lung Nodules |
| title_short | Health of Things Algorithms for Malignancy Level Classification of Lung Nodules |
| title_sort | health of things algorithms for malignancy level classification of lung nodules |
| topic | Computer-aided diagnosis pulmonary nodules lung cancer textural features structural co-occurrence matrix malignancy classification |
| url | https://ieeexplore.ieee.org/document/8322412/ |
| work_keys_str_mv | AT murillobrodrigues healthofthingsalgorithmsformalignancylevelclassificationoflungnodules AT raulvictormdanobrega healthofthingsalgorithmsformalignancylevelclassificationoflungnodules AT sharashamiaalves healthofthingsalgorithmsformalignancylevelclassificationoflungnodules AT pedropedrosareboucasfilho healthofthingsalgorithmsformalignancylevelclassificationoflungnodules AT joaobatistafduarte healthofthingsalgorithmsformalignancylevelclassificationoflungnodules AT arunksangaiah healthofthingsalgorithmsformalignancylevelclassificationoflungnodules AT victorhugocdealbuquerque healthofthingsalgorithmsformalignancylevelclassificationoflungnodules |