Non-Small Cell Lung Cancer Classification from Histopathological Images using Feature Fusion and Deep CNN 
Nur-A-Alam1, Md. Mahbubur Rahman2, Khandaker Mohammad Mohi Uddin3, Al Bashir4, Jahanara Akhtar5
1Nur-A-Alam, is working as a lecturer in the Department of Computer Science and Engineering, Dhaka International University (DIU). Dhaka, Bangladesh.
2Md. Mahbubur Rahman* , is working as a lecturer in the Department of Computer Science and Engineering, Dhaka International University (DIU). Dhaka, Bangladesh.
3Khandaker Mohammad Mohi Uddin, is working as an academic researcher and a lecturer in the Department of Computer Science and Engineering, Dhaka International University, Dhaka.
4Al Bashir, is a researcher and faculty member of Dhaka International University in the Department of Computer Science and Engineering. Dhaka, Bangladesh.
5Mst. Jahanara Akhtar, is currently serving as an Associate Professor of Computer Science and Engineering Department, Dhaka International University, Dhaka, Bangladesh. She is a Research Fellow (PhD) in the department of Computer Science and Engineering, Jahangirnagar University, Dhaka, Bangladesh.
Manuscript received on June 01, 2020. | Revised Manuscript received on June 08, 2020. | Manuscript published on June 30, 2020. | PP: 1013-1018 | Volume-9 Issue-5, June 2020. | Retrieval Number: E9266069520/2020©BEIESP | DOI: 10.35940/ijeat.D8935.069520
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Lung cancer is the overgrowth of cells in digestive organs. Identifying different types of lung cancer (squamous cell cancer, large cell carcinoma and adenocarcinoma) from lung histopathological images is outrageous works that shorten the chance of infected with lung cancer in the future. This research propounds an accurate diagnosis scheme using various neural network features and fusion of contourlet transform from lung histopathological image. This lesson has used several pre-train models (Alexnet, ResNet50, and VGG-16) in addition to divers scratch models while the pre-train Resnet50 model works better. The two reduction techniques (Principle Component Analysis (PCA) and Minimum Redundancy Maximum Relevance (MRMR)) have used to classify the type of lung cancer with the extraction of the most significant properties. In Convolution Neural Network (CNN) based lung cancer detection, the reduction approach PCA performs better. This proposed methodology is performed on ordinary datasets and establishes comparative better performance. The accuracy of this paper is 98.5%, sensitivity 96.50, specificity 97.00%, which is more effective than other approaches.
Keywords: Convolutional Neural Network (CNN), Principle Component Analysis (PCA), Contourlet Transform (CT), Histopathological Image, Minimum Redundancy Maximum Relevance (MRMR).