Detection of colorectal cancer using gene expression profile and artificial neural network

Fateme Asadollahzadeh shamkhal,1,* Hamidreza kobravi,2

1. 1Department of Electrical Engineering, Faculty of Engineering, Ferdowsi University Of Mashhad, Iran,
2. 2Research Center of Biomedical Engineering, Mashhad Branch, Islamic Azad University, Mashhad, Iran

Abstract

Introduction

The findings indicate that early detection of colorectal cancer and, consequently, appropriate therapeutic can be effective in reducing the morbidity of the cancer, since colorectal cancer, if detected early, is one of the most curable cancers. on the other hand, research has shown that neural networks increase the accuracy of colon cancer classification compared to other clinical pathology and statistical methods. for this reason, in the research, the method to classify the gene expression profiling data in colorectal cancer using the artificial neural network has been investigated.

Methods

The data of the colon and rectum gene expression profile in colorectal cancer patients and healthy subjects were used as input to artificial neural network (mlp). the used data include the data of 20 genes for 22 samples, that 11 are colorectal cancer tissues and 11 are normal colon tissues. the best result was obtained when the 10 hidden layers were in the neural network. the network trained ten epochs and the average was calculated from the results of these ten times of training. figure 1. shows this network. also, the gene expression profiling data got from cged (cancer gene expression database). this database has 1536 for colorectal cancer and we selected and investigated 20 genes randomly.

Results

Gene expression profiling data were given as inputs to the artificial neural network. the accuracy of classifying of the cancer data from healthy subjects was obtained using the mlp neural network, 94.77 ± 4.1764.

Conclusion

Mlp and information of gene expression profiling data, can classify and distinguish the genetic pattern of healthy subjects and colorectal cancer patients with high precision. considering that optimal cancer treatment requires accurate and timely diagnosis using a combination of histopathologic and clinical approaches, the proposed strategy can be used to increase the accuracy of the diagnosis of different types of cancer. in addition, making changes in the gene expression profile can play an important role in early detection of colorectal cancer.

Keywords

Artificial neural network, multilayer perceptron (mlp), colorectal cancer (crc), cancer detection, g