الفهرس 
صفحة العنوانOnly 14 pages are availabe for public view
 |  | from | 122 |  |  |
| | | from | 122 | | |
Abstract
ABSTRACT
Recently, Cancer has been considered one of the most dangerous diseases that leads to direct death. It spreads quickly in a frightening shape into all the body parts. Early detection of such a kind of disease may help to reduce its risks.
There are different detection methods such as collecting blood, urine, or stool sample from the patient and testing it in the laboratory. These traditional methods take a long time, require experienced examiners, and the probability of error is relatively high. On the other hand, new technologies are being used for cancer detection, such as nanotechnology and image processing.
Nanotechnology is very expensive and developing it can cost a lot of money in addition to the lack of proper knowledge about the effect of nanoparticles on biochemical pathways and processes of the human body. Image processing techniques need to expose the patient to the risk of radiation. It is sometimes difficult to interpret the results due to the low quality and resolution images or the absence of a skilled-experienced radiologist.
The main objective of this thesis is to present an effective cancer detection method that helps in improving the accuracy, minimizing the detection required time, avoiding the disadvantages of the previous detection methods, and increasing the research scope with a new point that can open ideas for many studies in cancer detection and diagnosis.
In this thesis, four tracks have been taken. Firstly, a detailed review has been explained on bioinformatics, genomic signal processing GSP, and their applications. Secondly, different power spectrum
methods (discrete Fourier transform DFT, power spectral density PSD, and PSD obtained by Welch’s averaged periodogram method) have been presented for breast cancer detection. Then the output plots have been compared at each case for normal and cancerous breast cells. Thirdly, a method for cancer classification has been discussed. It used both moment invariants and artificial neural networks ANNs. The used ANNs are the Kohonen network, Feed-forward backpropagation network, and Trainable cascade-forward backpropagation network. After that, the output results of each classifier have been compared by calculating its receiver operating characteristic ROC parameters (accuracy ACC, false-negative rate FNR, false-positive rate FPR, Specificity, and Sensitivity). Finally, a method for colon cancer detection has been applied using discrete wavelet transform DWT, statistical parameters (mean, variance, standard deviation, autocorrelation, entropy, skewness, and kurtosis), and machine learning algorithms (k-nearest neighbor KNN and support vector machine SVM). Then the classifiers’ results have been evaluated by calculating accuracy ACC, F1 score, and Matthews correlation coefficient MCC for each one of them.
The COVID-19 has appeared during the work so, at the end of the thesis, an important application has been presented. It has explained the efficiency of using GSP methods on viral disease detection and diagnosis.