HRDEL: High ranking deep ensemble learning-based lung cancer diagnosis model

Highlights

• We proposed a novel intelligent method for lung cancer diagnosis.

• We proposed BF-SSA for mitigating the dimension of features.

• We develop HRDEL learning approach with a high-ranking process.

• Evaluation of the performance of the proposed approach for two different data sets.

• The efficiency of the proposed approach is better than existing methods.

Abstract

Among all the diseases in human beings, lung cancer is known as the most hazardous disease that often leads to death rather than other cancer ailments. Lung cancer is asymptomatic, and so, it is unable to detect at the early stage. But, the rapid identification of lung cancer helps for sustaining the survival rate of people. Hence, many researchers develop various techniques for detecting lung cancer by undergoing different studies. Recently, computer technology has been used for solving these diagnosis problems. These developed systems involve diverse deep and machine learning approaches along with certain image-processing techniques for forecasting the severity level of lung cancer. Hence, this methodology plans to develop a novel intelligent method for diagnosing lung cancer. Initially, data is gathered by downloading two benchmark datasets, which include attribute information from various patients' health records. Furthermore, two standard techniques, “Principal Component Analysis (PCA) and t-Distributed Stochastic Neighbor Embedding (t-SNE)” have been used to extract features. Further, the deep features are retrieved from “the pooling layer of Convolutional Neural Network (CNN)”. Further to choose the significant features, the feature selection is taken place by the Best Fitness-based Squirrel Search Algorithm (BF-SSA), which is known as optimal feature selection. This hybrid optimization concept is considered to be superior in various domains to explore the search space efficiently and makes better performance in exploiting the feature selection. In the final phase called prediction, High Ranking Deep Ensemble Learning (HR-DEL) takes place concerning five forms of detection models. Finally, the high ranking of all the classifiers yields the final predicted output. The developed HR-DEL makes accurate prediction up to 8.79% better than the conventional methods and provides high robustness by reducing the dispersion or spread of the classification and model efficiency. The classification is performed, and the results are evaluated with the performance comparison of various algorithms.

Introduction

Cancer is considered a severe issue that may result in deaths with high mortality in both women and men owing to unclear clinical examinations and non-invasive treatments. The survival rate of lung cancer patients is very low while comparing with other cancers (Suresh & Mohan, 2020). The challenging task is to detect the nodule regions that are present in the soft lung tissues in the earlier stage of lung cancer (Wang and Charkborty, 2021, Sori et al., 2021). Computerized Tomography (CT) and Chest Radiograph (CXR) is employed for detecting pulmonary nodules for detecting lung cancer (Tian et al., 2021). Further, advanced technology is used to scan the whole chest region within a single breath hold and ensures low noisy images (Faruqui et al., 2021). Generally, detecting the pulmonary nodule is done through radiologists when cancer-related decisions (Xiao & Wu, 2021) are provided by the oncologists. But, the size of the nodule is small at the primary stage of cancer which makes the doctor take more time for checking the lung cancer (Naik & Edla, 2021) even with the help of experienced radiologists. The radiologists suggest different opinions when slight variations of morphology occur between the benign and malignant nodule (Li, 2021).

For achieving an accurate diagnosis and for detecting the entire nodules, it is necessary to involve expert doctors and radiologists (McWilliams et al., 2015, Yin et al., 2018). However, the symptoms of lung cancer are shown after the lung cancer reaches its severity, where it is impossible to cure. This is due to two vital issues such as high time consumption for a complete diagnosis of lung cancer and lack of recognition of the lung cancer symptoms at the early stage (Lee, 2018). The time taken for diagnosing and treating lung cancer needs to be improved for enhancing the probability of curing the lung cancer (Masood, 2020). Hence, the screening of lung cancer is the most important step by utilizing better identification approaches that must improve the patient health (Cirujeda, 2016). On the other hand, cancer detection using the lung image leads to true negative and false positives that further cause tension and additional investigation and cost to the patients and also for the doctor to suffer from the additional burden. The recent development of computer technology in lung cancer detection (Ozdemir et al., May 2020, Pang et al., 2020, He, June 2020) ensures better performance in attaining better detection accuracy that overrides human performance.

Different deep learning techniques achieve better efficiency in natural image recognition and detection, which is also extended to diverse medical imaging modalities and problems (Wu et al., 2019, Chen et al., 2020, Liu et al., 2013). Here, the deep CNN model helps to ensure huge success in different medical image processing when compared with the conventional methods (Mingxiang Feng, Xin Ye, Baishen Chen, Juncheng Zhang, Miao Lin, Haining Zhou, Meng Huang, Yanci Chen, Yunhe Zhu, 2021, Yang Jian and Zhou Yikai, 2004, Xing and Kejing, 1999). Especially in the field of medical image, the Convolutional network attains superior accuracy and sensitivity to the performance of the human expert (Shuji Sakai et al., 2006). Then, 3D-CNN is introduced for detecting lung cancer which achieves better performance in detection (Ramami, 2020). But it is restricted by certain limitations like time and memory complexity of the network. The conventional lung cancer detection model using CNN has not succeeded in performing more than 90 % levels for practical applications. Therefore, it is a prerequisite to raising a new lung cancer detection model with deep processing of lung images with an objective of enhanced deep learning approaches.

The goal of the research discussed in this paper is the proposed lung cancer diagnosis model deploys RNN to generate an ensemble learning model based on the proposed BF-SSA, in which five sets of features are assigned to each classification stage for lung cancer classification. This ensemble-based classification is used to improve the classification accuracy for correctly diagnosing lung cancer, which helps save lives by recognizing it early.

The residual part of the developed model is sequenced as follows. Part II explains the related works and their challenging issues. Part III depicts the initial process of the proposed lung cancer diagnosis model. The feature extraction process is explored in Part IV. Part V describes the lung cancer classification and the improved algorithm. Part VI discusses the simulation result. Part VIII summarizes the suggested lung cancer, diagnosis model.