Electrical - Electronic Engineering / Elektrik - Elektronik Mühendisliği
Permanent URI for this collectionhttps://hdl.handle.net/11147/11
Browse
4 results
Search Results
Article Citation - WoS: 4Citation - Scopus: 4A New Shapley-Based Feature Selection Method in a Clinical Decision Support System for the Identification of Lung Diseases(MDPI, 2023) Kababulut, Fevzi Yasin; Kuntalp, Damla Gurkan; Düzyel, Okan; Özcan, Nermin; Kuntalp, MehmetThe aim of this study is to propose a new feature selection method based on the class-based contribution of Shapley values. For this purpose, a clinical decision support system was developed to assist doctors in their diagnosis of lung diseases from lung sounds. The developed systems, which are based on the Decision Tree Algorithm (DTA), create a classification for five different cases: healthy and disease (URTI, COPD, Pneumonia, and Bronchiolitis) states. The most important reason for using a Decision Tree Classifier instead of other high-performance classifiers such as CNN and RNN is that the class contributions of Shapley values can be seen with this classifier. The systems developed consist of either a single DTA classifier or five parallel DTA classifiers each of which is optimized to make a binary classification such as healthy vs. others, COPD vs. Others, etc. Feature sets based on Power Spectral Density (PSD), Mel Frequency Cepstral Coefficients (MFCC), and statistical characteristics extracted from lung sound recordings were used in these classifications. The results indicate that employing features selected based on the class-based contribution of Shapley values, along with utilizing an ensemble (parallel) system, leads to improved classification performance compared to performances using either raw features alone or traditional use of Shapley values.Article Citation - WoS: 6Citation - Scopus: 10Using Chemosensory-Induced Eeg Signals To Identify Patients With <i>de Novo</I> Parkinson's Disease(Elsevier Sci Ltd, 2024) Olcay, Orkan; Onay, Fatih; Ozturk, Guliz Akin; Oniz, Adile; Ozgoren, Murat; Hummel, Thomas; Guducu, CagdasObjective: Parkinson's disease (PD) patients generally exhibit an olfactory loss. Hence, psychophysical or electrophysiological tests are used for diagnosis. However, these tests are susceptible to the subjects' behavioral response bias and require advanced techniques for an accurate analysis. Proposed Approach: Using well-known feature extraction methods, we characterized chemosensory-induced EEG responses of the participants to classify whether they have PD. The classification was performed for different time intervals after chemosensory stimulation to see which temporal segment better separates healthy controls and subjects with de novo PD. Results: The performances show that entropy and connectivity features discriminate effectively PD and HC participants when olfactory-induced EEG signals were used. For these methods, discrimination is over 80% for segments 100-700 and 200-800 milliseconds after stimulus onset. Comparison with Existing Methods: We compared the performance of our framework with linear predictive coding, bispectrum, wavelet entropy-based methods, and TDI score-based classification. While the entropy- and connectivity-based methods elicited the highest classification performances for olfactory stimuli, the linear predictive coding-based method elicited slightly higher performance than our framework when the trigeminal stimuli were used. Conclusion: This is one of the first studies that use chemosensory-induced EEG signals along with different feature extraction methods to classify healthy subjects and subjects with de novo PD. Our results show that entropy and functional connectivity methods unravel the chemosensory-induced neural dynamics encapsulating critical information about the subjects' olfactory performance. Furthermore, time- and frequency-resolved feature analysis is beneficial for capturing disease-affected neural patterns.Conference Object Citation - Scopus: 2Parkinson hastalığı sınıflandırmasına yönelik ivmeölçer tabanlı zamanlama analizi(IEEE, 2023) Karaçalı, Bilge; Onay, FatihParkinson's disease is a neurodegenerative disorder caused by dopamine deficiency in the basal ganglia, resulting in cognitive and motor impairments. In this study, accelerometer signals were used to estimate the delay time between the command to start pedaling and the actual movement onset in three groups: healthy individuals (n=13), Parkinson's disease patients (n=13), and patients with freezing of gait symptoms (n=13). Features were extracted from the delay time distributions for each participant and subjected to a triple classification. Linear support vector machine achieved a classification accuracy of 69.2% for all participants. Notably, the average time to start pedaling was found to be significantly different among the three groups, and accelerometer-based timing analysis could be used as a diagnostic tool to assist clinical tests.Article Citation - WoS: 3Citation - Scopus: 3Automated Labeling of Cancer Textures in Larynx Histopathology Slides Using Quasi-Supervised Learning(Science Printers and Publishers Inc., 2014) Önder, Devrim; Sarıoğlu, Sülen; Karaçalı, BilgeOBJECTIVE: To evaluate the performance of a quasisupervised statistical learning algorithm, operating on datasets having normal and neoplastic tissues, to identify larynx squamous cell carcinomas. Furthermore, cancer texture separability measures against normal tissues are to be developed and compared either for colorectal or larynx tissues. STUDY DESIGN: Light microscopic digital images from histopathological sections were obtained from laryngectomy materials including squamous cell carcinoma and nonneoplastic regions. The texture features were calculated by using co-occurrence matrices and local histograms. The texture features were input to the quasisupervised learning algorithm. RESULTS: Larynx regions containing squamous cell carcinomas were accurately identified, having false and true positive rates up to 21% and 87%, respectively. CONCLUSION: Larynx squamous cell carcinoma versus normal tissue texture separability measures were higher than colorectal adenocarcinoma versus normal textures for the colorectal database. Furthermore, the resultant labeling performances for all larynx datasets are higher than or equal to that of colorectal datasets. The results in larynx datasets, in comparison with the former colorectal study, suggested that quasi-supervised texture classification is to be a helpful method in histopathological image classification and analysis.