# Comparison of Accuracy in Extreme Learning Machine Based on Hidden Node Structure Variation for Lung Cancer Classification
> Tandungan S.

URL kanonis: https://discover.unhas.ac.id/publications/comparison-of-accuracy-in-extreme-learning-machine-based-on-hidden-node-structur
Jurnal / Konferensi: Iop Conference Series Materials Science and Engineering
Tahun terbit: 2019
DOI: https://doi.org/10.1088/1757-899X/676/1/012014
ISSN: 17578981
Citations: 2

## Authors
- Tandungan S.

## Abstract
Abstract This paper present Extreme Learning Machine to classify lung cancer nodules. Lung cancer is a type of lung disease that requires fast and specified treatment. Skills, facilities and multidisciplinary approach are required for diagnosing lung cancer. The use of Computed Tomography (CT) to detect lung cancer can reduce the number of deaths from lung cancer, but it increases the workload of the radiologist because CT screening process produces many medical images. Computer systems become one of the potential solutions to help radiologists solve the problem. Extreme Learning Machine is an algorithm that able to provide good generalization at fast learning time which is essential to help radiologists in analyzing lung cancer nodules images. In this paper, there were 877 nodules extracted from LIDC-IDRI dataset. All nodules used in this experiment consist of lung cancer nodules that diagnosed to four different level of malignancy and annotated by up-to four different radiologists. The result shows Extreme Learning Machine achieve 85.17%, 85.58% and 84.87% in accuracy and Matthew Correlation Coefficient 0.755, 0.762 and 0.749 using Hardlimit, Radial basis Function and Triangular Basis function, respectively.

## Keywords
- Lung cancer
- Extreme learning machine
- Malignancy
- Artificial intelligence
- Workload
- Machine learning
- Medicine
- Cancer
- Computer science
- Radiology
- Pathology
- Internal medicine
- Artificial neural network
- Operating system

---
Sumber: Discover Unhas — RIMS Universitas Hasanuddin.
Saat mengutip, gunakan DOI bila tersedia atau URL kanonis di atas.