Breast Cancer Classification And Visualisation Using Transposed Deep Neural Network

Citation

Ting, Fung Fung (2019) Breast Cancer Classification And Visualisation Using Transposed Deep Neural Network. PhD thesis, Multimedia University.

Full text not available from this repository.

Abstract

Malaysian woman has one of every 19 opportunities to be this dreaded disease amid her lifetime. Breast cancer remains as one of the most crucial causes of morbidity and mortality around the world. Mammography is currently the standard breast cancer medical screening option, however it is not that effective for patient under 40 years old and dense breasts, less susceptible to small tumours (less than 1 mm, approximately 100,000 cells), and gives no indication of breast cancer. Manual cancer mass delineation through medical doctors is currently referred as the standard approach, but it is suggested to be time-consuming and operator dependent. In order to counter these issues, a deep learning module had been designed to assist doctors in aspects of breast cancer detection and 3D model view of real patient MRI images. Breast Cancer Classification and Visualisation through Transposed Deep Neural Network (BCCV-TDNN) system is divided into three main sections, namely, Feature Wise Spatial Pre-processing (FWSP), Feature Wise Transposed Deep Neural Networks (FWTDNN), and Implicit Volume Ray Cast Mesh Renderer (IVRCMR). The designed system is able to detect and classify the breast cancer masses and visualize the patient breast cancer atlas overlay with the detected tumour. BCC-TDNN is designed to detect and classify the suspicious lesion from mammography to assist medical experts. The application of Feature Wise Spatial Pre-processing (FWSP) is to denoise and enhances the input medical images. The processed input medical images are separated into training set, validation set, and testing dataset. The datasets distribution is applied to minimise the chances of overfitting for the deep neural network rate and regularization parameter by using 20% of the training set for validation and evaluate them in the testing set.

Item Type: Thesis (PhD)
Additional Information: Call No.: QA76.87 .T56 2019
Uncontrolled Keywords: Neural networks (Computer science)
Subjects: Q Science > QA Mathematics > QA71-90 Instruments and machines > QA75.5-76.95 Electronic computers. Computer science
Divisions: Faculty of Engineering and Technology (FET)
Depositing User: Ms Nurul Iqtiani Ahmad
Date Deposited: 22 Sep 2020 17:51
Last Modified: 22 Sep 2020 17:51
URII: http://shdl.mmu.edu.my/id/eprint/7761

Downloads

Downloads per month over past year

View ItemEdit (login required)