Semra Icer

Comparison of multilayer perceptron training algorithms for portal venous doppler signals in the cirrhosis disease

Abstract In this study, we developed an expert diagnostic system for the interpretation of the portal vein Doppler signals belong the patients with cirrhosis and healthy subjects using signal processing and Artificial Neural Network (ANN) methods. Power spectral densities (PSD) of these signals were obtained to input of ANN using Short Time Fourier Transform (STFT) method. The four layered Multilayer Perceptron (MLP) training algorithms that we have built had given very promising results in classifying the healthy and cirrhosis. For prediction purposes, it has been presented that Levenberg Marquardt training algorithm of MLP network employing backpropagation works reasonably well. The diagnosis performance of the study shows the advantages of this system: It is rapid, easy to operate, noninvasive and not expensive. This system is of the better clinical application over others, especially for earlier survey of population. The stated results show that the proposed method can make an effective interpretation and point out the ability of design of a new intelligent assistance diagnosis system.

Quantitative Grading Using Grey Relational Analysis on Ultrasonographic Images of a Fatty Liver

A quantitative graduation system based on Grey Relational Analysis is proposed to recognize fatty livers in B-scan ultrasonic images. We evaluated ultrasonography liver images from 95 subjects having fatty livers (Grade I, II, III) and 45 normal subjects, as diagnosed by an expert radiologist. In practice, ultrasonographical findings of fatty liver are based on the brightness level of the liver in comparison to the renal parenchyma. The development of a non-invasive and accurate method would be of great clinical value as an alternative to diagnosing fatty liver based on the radiologist’s visual perception. In this study, we also evaluated AST and ALT liver enzymes for fatty liver having different grades. A high correlation between enzymes and Grey Relational Grades were found. The Receiver Operating Characteristic (ROC) curves were obtained and yielded satisfactory classification results using sensitivity, specificity and area under the curve for computing graduation and distinguishing fatty livers from healthy livers. With the proposed method based on Grey Relational Analysis, not only misdiagnosis caused by subjective differences in clinical evaluation will be reduced, but also the early diagnosis fatty liver and quantitative assessment of its degree will be achieved.

Automatic segmentation of corpus collasum using Gaussian mixture modeling and Fuzzy C means methods

This paper presents a comparative study of the success and performance of the Gaussian mixture modeling and Fuzzy C means methods to determine the volume and cross-sectionals areas of the corpus callosum (CC) using simulated and real MR brain images. The Gaussian mixture model (GMM) utilizes weighted sum of Gaussian distributions by applying statistical decision procedures to define image classes. In the Fuzzy C means (FCM), the image classes are represented by certain membership function according to fuzziness information expressing the distance from the cluster centers. In this study, automatic segmentation for midsagittal section of the CC was achieved from simulated and real brain images. The volume of CC was obtained using sagittal sections areas. To compare the success of the methods, segmentation accuracy, Jaccard similarity and time consuming for segmentation were calculated. The results show that the GMM method resulted by a small margin in more accurate segmentation (midsagittal section segmentation accuracy 98.3% and 97.01% for GMM and FCM); however the FCM method resulted in faster segmentation than GMM. With this study, an accurate and automatic segmentation system that allows opportunity for quantitative comparison to doctors in the planning of treatment and the diagnosis of diseases affecting the size of the CC was developed. This study can be adapted to perform segmentation on other regions of the brain, thus, it can be operated as practical use in the clinic.

Spectral broadening of lower extremity venous Doppler signals using STFT and AR modeling

This study researches the behaviour of spectral broadening index (SBI) obtained from spectra achived using short-time Fourier transform (STFT) analysis compared to that of SBI based on autoregressive (AR) modeling of clinical Doppler lower extremity vein signal. Doppler signals from 12 healthy subject with eight different physiologic situations were analysed. Sonograms obtained from Doppler signals were used to compare the applied methods in terms of their frequency resolution and their effectiveness for the determination of SBI. The AR based sonograms produced narrower spectra compared to STFT sonograms. Besides, the magnitude of the STFT based SBI was larger than that of the AR based SBI. Furthermore, standard deviations and coefficient of variations of STFT and AR based SBIs changed depending on each physiologic situation. The results of this research have also shown that despite the qualitative improvement in the individual frequency spectra, there was no quantitative advantage in using the AR approach over the STFT for the determination of SBI. Moreover there was also an additional computational complexity income connections with AR modeling.

Neural Correlates of Default Mode Network Connectivity in Children with Attention Deficit and Hyperactivity Disorder

The Purpose: The objective of this study is to explore neural correlates of Default Mode Network (DMN) regions in children with attention deficit and hyperactivity disorder (ADHD) using resting-state functional magnetic resonance imaging (rs-fMRI). Methods: The study included ten children with ADHD (aged between 9 and 16) and ten age-matched controls. Four DMN regions (medial prefrontal cortex (MPFC), the posterior cingulate cortex (PCC), left and right inferior parietal lobes (IPL) and the corresponding Broadmann areas in each one were used as seeds and their functional connectivity with the whole brain was explored and compared between ADHD and control groups using t-test (p<0.05). Results: We observed that when DMN regions were selected as seeds, the connected regions were different between two groups and were mostly in the right hemisphere in ADHD patients contrary to the left hemisphere in the control group. Conclusion: In conclusion, neural correlates of DMN regions differ in ADHD patients compared to healthy controls. Our findings suggest that in ADHD patients, DMN regions show more connectivity with the right hemisphere of the brain whereas the left hemisphere is more functionally connected with DMN in health controls. Further research is required to explore this atypical DMN connectivity in ADHD using larger cohort.

Determining of brain gray and white matter regions in magnetic resonance images

There are three parts in MRI projections of which are grey matter formed basically by neurons, white matter formed by axon exremities with mylelins, and cerebrospinal fluid. Changes and damages in these regions can cause various diseases. Autism, Parkinsonism, dyslexia, mental disorders, visual and audial loss can be the examples of grey matter diseases. As for the white matter diseases, MS (multiple sclerosis) and demyelinated diseases, minor cardiovascular diseases, neurologic damages, and blindness can be listed. In this study, a method has been proposed for segmentation of gray matter and white matter regions in the sections of the brain. The proposed method was used to determine the areas and calculate gray, matter matter of two hemispheres in terms of proportions of each other also to all the brain.

Obtaining resting state networks in early onset schizophrenia disease by Independent Component Analysis

Schizophrenia is a mental illness which usually begins in adolescence and can display important disorders of feelings, thoughts, and behavior. Delusion, hallucinations, and disorganized speech and behavior are some of the specific symptoms of schizophrenia. It is indicated that schizophrenia can be confused with other psychiatric disorders such as attention deficit hyperactivity disorder and mood disorder. Advanced magnetic resonance (MR) imaging methods are applied in the diagnosis of psychiatric illnesses due to the inadequacy of conventional anatomic imaging in the diagnosis of psychiatric disorders. Functional MRI which is an advanced MR imaging technique enables mapping of brain regions showing activation at rest or during task related events. This preliminary study investigated functional connectivity differences in some of resting state networks (anterior attention network, temporal network, audio network, and the default mode network) in early onset schizophrenia patients and children of control group (typically development) using resting state functional MRI method. This preliminary study aimed to contribute to the characterization of schizophrenia disease by identifying the specific changes that occur in the resting state networks of schizophrenia brain.

Comparison of lung tumor segmentation methods on PET images

Lung cancer is the most common cause of cancer-related deaths that occur all over the world. Recently, various image processing approaches have been used on PET images in order to characterize the uniformity, density, coarseness, roughness, and regularity (i.e., texture properties) of the intratumoral 18F-fluorodeoxyglucose (FDG) uptake. The first and important step of this kind of analysis is to differentiate tumor region from other structures and background, which is called segmentation. In this study, k-means, active contour (snake), and Otsus tresholding methods were applied on PET images obtained from 36 patients and the performances were compared by the nuclear medicine expert in our team. The results show that Otsu tresholding approach is more selective.

Exploring default mode connectivity using region of interest analysis in children with attention deficit and hyperactivity disorder

In this preliminary study, functional connectivities among brain default mode network (DMN) regions were aimed to investigate using resting-state functional magnetic resonance imaging. 13 children diagnosed with ADHD whose ages are between 9 and 16 and 13 healthy controls in the same ages were studied using ROI-to-ROI analysis method. Major Default Mode Networks(DMN) investigated were medial prefrontal cortex (MPFC), Posterior cingulate cortex, right parietal cingulate lobe (right IPL) and left parietal cingulate lobe (left IPL). The group level comparison was performed on functional MR images after spatial and temporal preprocessing. Functional activities of control and DEHB groups were examined and meaningful results were obtained associating with anatomical regions. In the ADHD group, increased connectivity between right inferior parietal lobe and left inferior parietal lobe and decreased connectivity between other DMN regions were obtained significantly compared to the control group.

Registration and fusion of lung tumor PET/CT images

Image fusion attracts attention in medical field due to complementary behavior and application such as diagnosis and treatment planning. In this study, first positron emission tomography (PET) and computed tomography (CT) images coming from 8 nonsmall cell lung cancer were registered then wavelet and principal component analysis methods were applied to fuse images. According to mutual information metric and nuclear medicine expert wavelet method gave better results when compared to PCA.