Page 104 - The-5th-MCAIT2021-eProceeding

Detection of Cancer Cell and Tumor from MRI Image Using a

Hybrid Approach – A Conceptual Framework

b
a*
A F M Saifuddin Saif , Zainal Rasyid Mahayuddin
a Department of Computer Science, American International University - Bangladesh.
b Faculty of Information Science and Technology, University Kebangsaan Malaysia, Selangor, Malaysia
*Email: rashedcse25@yahoo.com

Abstract

Cancer is one of the deadliest diseases among all other diseases of human beings. Various types of methods for segmentation
and detection of cancer cell or tumor were proposed by previous research. Among these methodologies of detection and
segmentation, most of the previous methods can detect tumors or cancer cell accurately but have problems such as too
complex computations for normal computers to process. In addition, some methods are unable to detect or draw out the
region of the tumor with above 90% to 100% accuracy which is much needed for safe surgery. In this context, Magnetic
Resonance Imaging (MRI) is an effective tool for cancer detection in recent years. This research proposes a conceptual
framework for cancer cell and tumor detection from MRI images. For improved classification this research intends to use
convolutional neural network. In addition, adaptive histogram equalization (CLAHE) is proposed to provide noise free
images for improved performance of the overall methodology. Proposed conceptual framework is expected to provide robust
solution as improved segmentation and detection of cancer cell and tumor from MRI images.

Keywords: Segmentation; Convolutional neural network; MRI images

1. Introduction

Cancer is a dense and abnormal cells proliferation in the body tissue. Cancer cells do not have the longer
respond to some or many of the signals that mainly control cellular growth and death. Researchers proposed
various methods for detecting cancer, i.e. Magnetic Resonance Imaging (MRI), microwave imaging, film-screen
mammography. MRI is one of the most significant technology of cancer imaging for its accuracy. There are
also some significant deep learning and machine-learning algorithms also proposed by the researchers, i.e.
convolutional neural network (CNN) etc. This research proposes a conceptual framework where denoising
approaches are considered and convolution neural network is intended to used for classification of tumor from
input image. For pre-processing, this research intends to use adaptive thresholding for foreground extraction
and adaptive histogram equalization method for image contrasting. The histogram equalization operation is
considered to enhance the contrast of the image.
Rest of this paper is organized as follows. Section 2 presents core research background, section 3
demonstrates conceptual framework proposed by this research, section 4 illustrates proposed experimentation
evaluation approach for the proposed conceptual framework, finally concluding remarks are presented in section
5.

2. Background Study

Cancer and tumor detection is a significant research in modern biomedical and computer vision research
domain. Researchers proposed various methods for detection of cancer and tumor in modern technology, i.e.
Microwave imaging, MRI, Film-screen mammography. MRI is mostly used for tumor detection. In this context,

E- Proceedings of The 5th International Multi-Conference on Artificial Intelligence Technology (MCAIT 2021) [91]
Artificial Intelligence in the 4th Industrial Revolution