Multi Parameterized Modified Local Binary Pattern for Lung Cancer Detection by Deep Learning Methods

Title

Multi Parameterized Modified Local Binary Pattern for Lung Cancer Detection by Deep Learning Methods

Creator

Dominic, Denny

Contributor

K, Balachandran.

Description

The research work is focusing on developing a classification model for Lung Cancer detection by integrating the image features with Modified Local Binary Pattern (MLBP), Modified Principal Component Analysis (MPCA), newlinesymptoms and Risk factors using Deep Learning methods and converting the image features into three dimensional (3D) images. The aim of this research is to identify the malignant and normal tumours from the Computer newlineTomography (CT) images with improved accuracy. The 2D CT images of Lung Cancer patients have been preprocessed with Median and Gabor filtering methods and watershed segmentation. The CT images are also newlineprocessed with the Zero Component Analysis (ZCA) whitening and Modified Local Binary Pattern. The processed image is used in the research for classification. The Lung Cancer dataset in the research are collected from newlinevarious medical colleges. The dataset contain CT images with Lung Cancer and without Lung Cancer. The research is conducted by integrating the selected Image features, Risk factor and symptoms of Lung Cancer of the newlinesame patients. The Integration using feature selections is carried out with Modified Principal Component Analysis. The Modified Principal Component Analysis is used in the research to reduce the time complexity. The results are evaluated with Gini coefficient, Confusion Matrix parameters and ROC newlinecurve. Two Dimensional (2D) CT images are converted into a Three Dimensional (3D) image for the clarity and the visibility of Lung Cancer nodules. The conversion from 2D to 3D has been using combining two methods, the orthogonality and visualization of 4D rotation. This enabled to find the location of the Lung Cancer from different angle and with different viewpoints. The 3D image shows the location of the Lung Cancer by Four Dimensional (4D) visualization and 3D rotation, thus giving clarity to the newlineexisting 2D images.

Source

Author's Submission

Date

2023-01-01

Publisher

Christ(Deemed to be University)

Subject

Computer Science and Engineering

Rights

Open Access

Relation

61000222

Format

PDF

Language

English

Type

PhD

Identifier

http://hdl.handle.net/10603/478099