Filter by type:

Sort by year:

Annealing-Based Model-Free Expectation Maximization for Multicolor Flow Cytometry Data Clustering

Journal
Başak Esin Kokturk, Bilge Karacali
International Journal of Data Mining and Bioinformatics (In Publication)

This paper proposes an optimized model-free expectation maximization method for automated clustering of high dimensional datasets. The method is based on a recursive binary division strategy that successively divides an original dataset into distinct clusters. Each binary division is carried out using a model-free expectation maximization scheme that exploits the posterior probability computation capability of the quasi-supervised learning algorithm subjected to a line-search optimization over the reference set size parameter analogous to a simulated annealing approach. The divisions are continued until a division cost exceeds an adaptively determined limit. Experiment results on synthetic as well as real multi-color flow cytometry datasets showed that the proposed method can accurately capture the prominent clusters without requiring any prior knowledge on the number of clusters or their distribution models.

Detection of vessels in a retinal image using decomposed pixel classification method

Conference Proceedings
Sargan, E.; Kokturk, B.E.
Signal Processing and Communications Applications Conference (SIU), 23th

The aim of this study, determination of blood vessels on high dimensional retinal images using decomposed pixel classification method. In this work intensity histograms are obtained by using vectoral directions. Local minimum and local maximum points are found by using these histograms to detect location of vessels. In addition to this, vessel thickness information are detected by determining a threshold point on these histograms.

Automated labeling of electroencephalography data using quasi-supervised learning

Conference Proceedings
Kokturk, B. E., and B. Karacali.
Signal Processing and Communications Applications Conference (SIU), 20th

In this study, the separation of the stimulus effects from the baseline was investigated in electroencephalography data recorded under different visual stimuli using quasi-supervised learning. The data feature vectors were constructed using independent component analysis and wavelet transform, and then, these feature vectors were separated using quasi-supervised learning. Experiment results showed that the EEG data of the stimuli can be separated using quasi-supervised learning

Quasi-supervised learning on DNA regions in colon cancer histology slides.

Conference Proceedings
Kokturk, B. E., and B. Karacali
Signal Processing and Communications Applications Conference (SIU), 2013 21st. IEEE

The aim of this study, nuclei base automatic detection of cancerous regions via determination of DNA-rich regions in high definition histology images. In the study; DNA-rich regions were determined using k-means clustering and some mathematical morphology operations, the diseased regions were diagnosed using morphological characteristics via quasi-supervised learning.It’s observed that quasi-supervised learning method successfully separates cancerous chromatin regions from others successfully with experiments of colon cross-section histology images.

Model-free expectation maximization for divisive hierarchical clustering of multicolor flow cytometry data

Conference Proceedings
Başak Esin Köktürk, Bilge Karaçalı
Bioinformatics and Biomedicine (BIBM), 2014 IEEE International Conference on. IEEE

This paper proposes a new method for automated clustering of high dimensional datasets. The method is based on a recursive binary division strategy that successively divides an original dataset into distinct clusters. Each binary division is carried out using a model-free expectation maximization scheme that exploits the posterior probability computation capability of the quasi-supervised learning algorithm. The divisions are carried out until a division cost exceeds an adaptively determined limit. Experiment results on synthetic as well as real multi-color flow cytometry datasets showed that the proposed method can accurately capture the prominent clusters without requiring any knowledge on the number of clusters or their distribution models.