Fig 1.
The proposed methodology.
Fig 2.
The data preprocessing steps.
Fig 3.
The architecture of the proposed CNN model for lung cancer detection, where F, K, and S indicate the filters, kernels, and strides, respectively.
Fig 4.
The flowchart of the EOSA-CNN algorithm showing the optimization process for the computed solution vector.
Table 1.
CNN hyperparameter configuration.
Table 2.
Notations and description of variables and parameters for SEIR-HDVQ.
Fig 5.
An illustration of samples from the original dataset showing images with normal, benign, and malignant labels.
(a) Normal, (b) Bengin, and (c) Malignant.
Fig 6.
Illustration of the transformed binary images of normal, benign, and malignant samples into grayscale.
(a) Normal, (b) Bengin, and (c) Malignant.
Fig 7.
Outcome of application of Gaussian blur filter works on the lung cancer images of normal, benign, and malignant samples.
(a) Normal, (b) Bengin, and (c) Malignant.
Fig 8.
Illustrates the outcome of the Otsu’s method on normal, benign, and malignant samples.
(a) Normal, (b) Bengin, and (c) Malignant.
Fig 9.
Shows normalized lung cancer on normal, benign, and malignant samples.
(a) Normal, (b) Bengin, and (c) Malignant.
Fig 10.
Shows the erosion (a), (b), and (c) images and dilation (d), (e), and (f) images for normal, benign, and malignant samples, respectively. (a) Normal, (b) Bengin, and (c) Malignant, (d) Normal, (e) Bengin, and (f) Malignant.
Fig 11.
Explain the output of the CLAHE filter for normal, benign, and malignant lung cancer samples.
(a) Normal, (b) Bengin, and (c) Malignant.
Fig 12.
Explains the output of the wavelet filter for normal, benign, and malignant lung cancer samples.
(a) Normal, (b) Bengin, and (c) Malignant.
Fig 13.
Shows the LL component of output from the wavelet filter function for normal, benign, and malignant lung cancer samples.
(a) Normal, (b) Bengin, and (c) Malignant.
Table 3.
Standard benchmark functions used for the experimentation of EOSA and other similar optimization algorithms.
Table 4.
Structure of the confusion matrix.
Fig 14.
Convergent curves of EOSA on standard benchmark functions over 1, 50, 100, 200, 300, 400 and 500 epochs.
Fig 15.
Convergent curves of EOSA and related optimization algorithms benchmark functions over 1, 50, 100, 200, 300, 400 and 500 epochs.
Table 5.
Comparison of best, worst, mean, median and standard deviation values for ABC, WOA, BOA PSO, EOSA, DE, GA, HGSO, SOA, and BMO metaheuristic algorithms using the classical benchmark functions over 500 runs and 100 population size.
Table 6.
The overall and per-class performance of the GA-CNN, LCBO-CNN, MVO-CNN, SBO-CNN, WOA-CNN, and EOSA-CNN hybrid algorithms as compared with the basic CNN architecture.
Table 7.
Results comparison of the best, mean, standard deviation, median, and worst overall performance based on the GA-CNN, LCBO-CNN, MVO-CNN, SBO-CNN, WOA-CNN, and EOSA-CNN hybrid algorithms compared with the basic CNN architecture.
Table 8.
Class-based results comparison for the best, mean, standard deviation, median, and worst of class-based performance based on the GA-CNN, LCBO-CNN, MVO-CNN, SBO-CNN, WOA-CNN, and EOSA-CNN hybrid algorithms and as compared with the basic CNN architecture.
Fig 16.
Overlapped confusion matrix for all hybrid algorithms with CNN.
(a) GA-CNN, (b) LCBO-CNN, (c) MVO-CNN (d) SBO-CNN, (e) WOA-CNN, (f) EOSA-CNN, and (g) CNN.
Table 9.
Performance comparison of the proposed method and some similar methods of CNN for the classification of lung cancer.