Effect of obesity on electroretnography (ERG) signal

Obesity is a health problem that arises as a result of an increase in the amount of fat in the body to an extent that will impair human health. Obesity, the prevalence of which is increasing day by day, has become a worldwide public health problem, especially in developed and developing countries. Obesity is not only an aesthetic problem, but also closely related to many health problems such as cardiovascular diseases, diabetes, stroke, joint disorders, and some types of cancer. In addition, it has been revealed that obesity has negative effects on the eye and causes irreversible vision loss if necessary precautions are not taken in a timely manner. Today, ocular electrophysiological tests such as Electrooculography (EOG), Electroretinography (ERG) and Visual Evoked Potential (VEP) are widely used in the diagnosis of eye diseases in clinical studies. ERG, which is also used in this thesis, provides the opportunity to examine the massive response of the retinal layer, which includes rod, cone and ganglion cells, to a light stimulus. There are many studies in the literature on the diagnosis of retinal diseases by using ERG. However, it has been determined that there is no study examining the relationship between obesity and ERG signals. On the other hand, it has been demonstrated that the body mass index (BMI) parameter, which is frequently used to measure obesity in clinical practice, is not sufficient on its own in the evaluation of obesity-related diseases. Therefore, within the scope of this thesis, the physiological effect of obesity on flash ERG (fERG) signals was investigated. In addition, an automatic decision support system has been proposed to classify obesity and its levels using fERG signals. Rod, maximum combined and cone responses of fERG signals recorded from healthy, overweight, obese, morbid obese and super obese individuals were used in the thesis. In this direction, normal values for fERG signals were determined in the laboratory environment established primarily. Then, the effects of obesity and its levels on fERG signals were analyzed with Short Time Fourier Transform (STFT), Continuous Wavelet Transform (CWT) and Discrete Wavelet Transform (DWT) methods. Then, feature extraction was performed with DWT and Wavelet Packet Transformation (WPT) methods. Obesity levels were classified by using the obtained features as inputs in ANN models based on ANN and particle swarm optimization (PSO). The performances of the two models used in the classification process were compared statistically.