DSpace Arşivi :: by Yazar "Ceylan, Murat" değerine göre listeleniyor

Yazar "Ceylan, Murat" seçeneğine göre listele

Listeleniyor 1 - 20 / 39

Application of complex discrete wavelet transform in classification of Doppler signals using complex-valued artificial neural network
(ELSEVIER, 2008) Ceylan, Murat; Ceylan, Rahime; Oezbay, Yueksel; Kara, Sadik
Objective: In biomedical signal classification, due to the huge amount of data, to compress the biomedical waveform data is vital. This paper presents two different structures formed using feature extraction algorithms to decrease size of feature set in training and test data. Materials and methods: The proposed structures, named as wavelet transform-complex-valued artificial neural network (WT-CVANN) and complex wavelet transform-complex-valued artificial neural network (CWT-CVANN), use real and complex discrete wavelet transform for feature extraction. The aim of using wavelet transform is to compress data and to reduce training time of network without decreasing accuracy rate. In this study, the presented structures were applied to the problem of classification in carotid arterial Doppler ultrasound signals. Carotid arterial Doppler ultrasound signals were acquired from left carotid arteries of 38 patients and 40 healthy volunteers. The patient group included 22 mates and 16 females with an established diagnosis of the early phase of atherosclerosis through coronary or aortofemoropopliteal (tower extremity) angiographies (mean age, 59 years; range, 48-72 years). Healthy volunteers were young non-smokers who seem to not bear any risk of atherosclerosis, including 28 mates and 12 females (mean age, 23 years; range, 19-27 years). Results and conclusion: Sensitivity, specificity and average detection rate were calculated for comparison, after training and test phases of all structures finished. These parameters have demonstrated that training times of CVANN and real-valued artificial neural network (RVANN) were reduced using feature extraction algorithms without decreasing accuracy rate in accordance to our aim. (C) 2008 Elsevier B.V. All rights reserved.
An approach for tissue density classification in mammographic images using artificial neural network based on wavelet and curvelet transforms
(SPIE-INT SOC OPTICAL ENGINEERING, 2015) Yasar, Huseyin; Ceylan, Murat
Breast cancer is one of the types of cancer which is most commonly seen in women. Density of breast is an important indicator for the risk of cancer. In addition, densities of tissue may harden the diagnosis by hiding the abnormalities occurring on the breast. For this reason, during the process of diagnosis, the process of automatic classification of breast density has a significant importance. In this study, a new system with the base of Artificial Neural Network (ANN) and multiple resolution analysis is suggested. Wavelet and curvelet analyses having the most common use have been used as multi resolution analysis. 4 pieces of statistics which are minimum value, maximum value, mean value and standard deviation have been extracted from the images which have been eluted to their sub-bands via multi resolution analysis. For the purpose of testing the success of the system, 322 pieces of images which are in MIAS database have been used. The obtained results for different backgrounds are so satisfying; and the highest classification values have been obtained as 97.16 % with Wavelet transform and ANN for fatty background and 79.80 % with Wavelet transform and ANN for fatty-glanduar background. The same results have been obtained using Wavelet transform and ANN and Curvelet transform and ANN for dense background and accuracy rate of 84.82 % have been reached. The results of mean classification have been obtained, for three pieces of tissue types (fatty, fatty-glanduar, dense), in sequence as 84.47 % with the use of ANN, 85.71 % with the use of curvelet analysis and ANN; and 87.26 % with the use of wavelet analysis and ANN.
An Automatic System of Detecting Changes in Aerial Images Using ANN Based Contourlet Transform
(IEEE, 2015) Yasar, Huseyin; Hatipoglu, Ridvan Safa; Ceylan, Murat
The obtaining of the aerial images got easy thanks to technological developments in the field of unmanned aerial vehicles and these images were began to be used frequently in the field of image processing. Automatic changes detection from aerial images is among the most important study fields. An automatic system for changes detection has been proposed by using contourlet transform and artificial neural network (ANN) in this study. The contourlet transform is applied to the reference image in the first phase of the system consisting of two phases. Mean, variance, standard deviation and skewness values were calculated from the obtained sub-image matrix and seven image feature vectors are formed by using these statistical values and combinations. The numerical equivalents of the reference image were obtained by using the feature vectors by ANN. The same procedures were applied to the image that its exchange will be examined in the second phase of the system. The change between numerical provisions of the reference image and the image to be examined compared to the threshold value set by the user and automatic changes detection was performed. It was found that the changes in numerical results obtained at the end of the study overlap with the changes in aerial images.
Bilgisayar tomografili akciğer görüntülerinin değerlendirilmesinde kompleks değerli yeni bir akıllı sistem tasarımı
(Selçuk Üniversitesi, Fen Bilimleri Enstitüsü, 2009) Ceylan, Murat; Özbay, Yüksel
Akciğer kanseri, diğer kanser türleri ile karşılaştırıldığında en yüksek ölüm oranına sahip olan kanser türüdür ve akciğer kanserinin neden olduğu ölümler her geçen yıl düzgün bir artış göstermektedir. Birçok ülke, akciğer kanserinin yayılım hızını düşürebilmek amacı ile en etkili yol olarak sigara kullanımını yasaklamıştır. Bununla birlikte kanserin erken aşamalarında şüpheli nodüllerin doğru bir şekilde ayrıştırılabilmesi, yani benign (iyi huylu) ve malign (kötü huylu) nodüllerin başarı ile sınıflandırılabilmesi, hem hastalığın takibi hem de hastalığın tedavisinin şekillendirilmesi için oldukça önemlidir.Akciğer nodüllerinin görüntülenmesinde kullanılan akciğer bilgisayarlı tomografi (BT) taraması, tek dedektörlü sistemlerde tipik olarak 40 ila 100 eksenel görüntü kesiti üretirken, daha yeni versiyonlarında, yani çok dedektörlü sistemlerde 300 ile 600 görüntü kesiti elde edilmektedir. Bu çok sayıdaki görüntüyü okumak ve yorumlamak, radyologların önemli bir çaba göstermesini gerektirmekle birlikte insan hatası ve kanserli nodüllerin kaçırılması gibi dezavantajlar ile sıklıkla karşılaşılmaktadır. Bu yüzden, bilgisayar destekli teşhis (BDT) yaklaşımlarına, radyologların iş yükünü azaltmak ve sınıflama hassasiyetini artırmak amacı ile gittikçe artan bir şekilde ihtiyaç duyulmaktadır.Tez çalışmasında, akciğer BT görüntülerini yorumlayan radyologların, karşılaştıkları nodüller hakkında karar vermelerine yardımcı olabilmek amacı ile yeni bir akıllı BDT sistemi önerilmiştir. Sistem Başkent Üniversitesi Konya Uygulama ve Araştırma Hastanesinden alınan ve biyopsi sonuçlarına uygun olarak etiketlenmiş 22 malign ve 10 benign nodülü içeren 32 BT görüntüsü kullanılarak test edilmiştir.Geliştirilen sistem, iki temel işlem bloğu şeklinde tasarlanmıştır. İlk olarak segmentasyon işlemi gerçekleştirilmiştir. Bu amaçla, BT cihazından alınan akciğer görüntüleri üzerinde kompleks-değerli dalgacık dönüşümü (KDDD) kullanılarak boyut azaltımı yapılmış ve elde edilen görüntülerde akciğer bölgesi haricindeki bölgeler atılmıştır. Bu işlem, kompleks-değerli yapay sinir ağı (KDYSA) ve hücresel yapay sinir ağı (HYSA) gibi iki farklı YSA modelinin kullanıldığı yeni bir hibrit YSA modeli olan Kompleks Değerli Hibrit Hücresel Yapay Sinir Ağı (KDHHYSA) kullanılarak gerçekleştirilmiştir. Elde edilen ortalama doğruluk oranı test için % 99.71' dir.Geliştirilen BDT sisteminin ikinci bileşeni, segmente edilen akciğer BT görüntüsünde yer alan nodüllerin sınıflandırılmasıdır. Segmente edilen BT görüntülerine dört farklı seviyede KDDD uygulanmış ve ayrı ayrı her bir seviye için dört istatistik özellik (en küçük değer, en büyük değer, ortalama değer, standart sapma) çıkarılarak KDYSA ile sınıflandırılmıştır. Sınıflama geçerliliğini kanıtlamak için 10-kat çapraz geçerlilik yöntemi uygulanmıştır. En iyi sonuçlar, dördüncü seviye KDDD kullanıldığında elde edilmiş ve 22 malign nodülün 22' si malign olarak, 10 benign nodülün 9 tanesi ise benign olarak sınıflandırılmıştır. Duyarlılık % 95.2, belirlilik % 100, ortalama belirleme oranı % 97.6, doğruluk % 96.7, pozitif tahmin testi % 100 ve negatif tahmin testi % 90 olarak elde edilmiştir.
Blood Vessel Extraction From Retinal Images Using Complex Wavelet Transform and Complex-Valued Artificial Neural Network
(IEEE, 2013) Ceylan, Murat; Yasar, Huseyin
Retinal imaging in ophthalmology plays an important role for the diagnosis of diabetes, cardiovascular disease, etc. In retina images, changes of blood vessels can help the expert to detection of diseases. Manually extraction of blood vessels from retinal images is usually difficult process due to depending on the experience of physician, back-ground artifacts, different acquisition process. Therefore, the aim of this study is to purpose a novel method for automatic blood vessel extraction from retinal image. This study presents a combined structure. This structure is realized with two cascade stages: feature extraction with 4th level Complex Wavelet Transform (CWT) and Complex-Valued Artificial Neural Networks (CVANN) for the blood vessels segmentation. To check the validation of proposed method, public DRIVE database is used. Result of this study has a higher accuracy (98.56 %) than previously studies in the literature.
Classification of carotid artery Doppler signals in the early phase of atherosclerosis using complex-valued artificial neural network
(PERGAMON-ELSEVIER SCIENCE LTD, 2007) Ceylan, Murat; Ceylan, Rahime; Dirgenali, Fatma; Kara, Sadik; Ozbay, Yuksel
In this study, carotid arterial Doppler ultrasound signals were acquired from left carotid arteries of 38 patients and 40 healthy volunteers. The patient group had an established diagnosis of the early phase of atherosclerosis through coronary or aortofemoropopliteal angiographies. Results were classified using complex-valued artificial neural network (CVANN). Principal component analysis (PCA) and fuzzy c-means clustering (FCM) algorithm were used to make a CVANN system more effective. For this aim, before classifying with CVANN, PCA method was used for feature extraction in PCA-CVANN architecture and FCM algorithm was used for data set reduction in FCM-CVANN architecture. Training and test data were selected randomly using 10-fold cross validation. PCA-CVANN and FCM-CVANN architectures classified healthy and unhealthy subjects for training and test data with about 100% correct classification rate. These results shown that PCA-CVANN and FCM-CVANN classified Doppler signals successfully. (c) 2005 Elsevier Ltd. All rights reserved.
Combined Complex-Valued Artificial Neural Network (CCVANN)
(INT ASSOC ENGINEERS-IAENG, 2011) Ceylan, Murat
This study presents a new version of complex valued artificial neural networks (CVANN) for the complex valued pattern recognition and classification. Proposed new method is called as combined complex-valued artificial neural network (CCVANN) which is a combination of two complex valued artificial neural networks. To check the validation of proposed method, complex-valued XOR benchmark problem is used. The accuracy of the CCVANN model is more satisfactory as compared to the existing studies in the literature. Moreover the proposed CCVANN models' results have lower recognition error than using a single CVANN model.
Comparison of Artificial Neural Network and Extreme Learning Machine in Benign Liver Lesions Classification
(IEEE, 2015) Akin, Mustafa; Ceylan, Murat
In this study, the classification of the most common benign lesions, cysts and hemangiomas in liver was achieved using magnetic resonance (MR) images. T1 venous phase of 68 liver MR images were used for the classification, including 28 cysts and 40 hemangiomas MR images. Liver segmentation was done by expert radiologists using MR images. Then automatic windowing was applied to images to reduce the negative impact on the process of image-free areas of tissue information. The obtained images were normalized and thresholded using histogram equalization. The average, standard deviation and distortion values of the image feature matrix obtained by applying wavelet transform (WT) and complex valued wavelet transform (CVWT) onto the thresholded images were calculated. Artificial neural network (ANN), extreme learning machine (ELM), cyst and hemangiomas classification were achieved using these features as inputs. As a result of this study, 50% accuracy at the data applied CVWT, 70,5% accuracy at the data applied WT were obtained in ANN. Average processing time is 4.61 seconds. When examined the ELM application results, it can be seen that there are 55, 8% accuracy at the data applied CVWT and 62, 5% accuracy at the data applied WT. Also, the average processing time is 0,016 seconds this time. Although the classification results seem low, classification accuracy rates will increase with the development studies considering advantage of ELM processing time.
Comparison of artificial neural network and extreme learning machine in benign liver lesions classification
(Institute of Electrical and Electronics Engineers Inc., 2016) Akın, Mustafa; Ceylan, Murat
In this study, the classification of the most common benign lesions, cysts and hemangiomas in liver was achieved using magnetic resonance (MR) images. T1 venous phase of 68 liver MR images were used for the classification, including 28 cysts and 40 hemangiomas MR images. Liver segmentation was done by expert radiologists using MR images. Then automatic windowing was applied to images to reduce the negative impact on the process of image-free areas of tissue information. The obtained images were normalized and thresholded using histogram equalization. The average, standard deviation and distortion values of the image feature matrix obtained by applying wavelet transform (WT) and complex valued wavelet transform (CVWT) onto the thresholded images were calculated. Artificial neural network (ANN) ,extreme learning machine (ELM), cyst and hemangiomas classification were achieved using these features as inputs. As a result of this study,50% accuracy at the data applied CVWT, 70,5% accuracy at the data applied WT were obtained in ANN. Average processing time is 4.61 seconds. When examined the ELM application results, it can be seen that there are 55, 8% accuracy at the data applied CVWT and 62, 5% accuracy at the data applied WT. Also, the average processing time is 0,016 seconds this time. Although the classification results seem low, classification accuracy rates will increase with the development studies considering advantage of ELM processing time. © 2015 IEEE.
Comparison of complex-valued neural network and fuzzy clustering complex-valued neural network for load-flow analysis
(SPRINGER-VERLAG BERLIN, 2006) Ceylan, Murat; Cetinkaya, Nurettin; Ceylan, Rahime; Ozbay, Yuksel
Neural networks (NNs) have been widely used in the power industry for applications such as fault classification, protection, fault diagnosis, relaying schemes, load forecasting, power generation and optimal power flow etc. Most of NNs are built upon the environment of real numbers. However, it is well known that in computations related to electric power systems, such as load-flow analysis and fault level estimation etc., complex numbers are extensively involved. The reactive power drawn from a substation, the impedance, busbar voltages and currents are all expressed in complex numbers. Hence, NNs in the complex domain must be adopted for these applications. This paper proposes the complexvalued neural network (CVNN) and a new fuzzy clustering complex-valued neural network (FC-CVNN) to estimate busbar voltages in a load-flow problem. The aim of this paper is to present a comparative study of estimation busbar voltages in load-flow analysis using the conventional neural network (real-valued neural network, RVNN), the CVNN and the new FC-CVNN. The results suggest that a new proposed FC-CVNN and CVNN architecture can generalize better than ordinary RVNN and the FC-CVNN is also learn faster.
Complex-valued wavelet artificial neural network for Doppler signals classifying
(ELSEVIER SCIENCE BV, 2007) Ozbay, Yuksel; Kara, Sadik; Latifoglu, Fatma; Ceylan, Rahime; Ceylan, Murat
Objective: In this paper, the new complex-valued wavelet artificial neural network (CVWANN) was proposed for classifying Doppler signals recorded from patients and healthy volunteers. CVWANN was implemented on four different structures (CVWANN-1, -2, -3 and -4). Materials and methods: In this study, carotid arterial Doppler ultrasound signals were acquired from left carotid arteries of 38 patients and 40 healthy volunteers. The patient group had an established diagnosis of the early phase of atherosclerosis through coronary or aortofemoropopliteal angiographies. In implemented structures in this paper, Haar wavelet and Mexican hat wavelet functions were used as real and imaginary parts of activation function on different sequence in hidden layer nodes. CVWANN-1, -2 -3 and -4 were implemented by using Haar-Haar, Mexican hat-Mexican hat, Haar-Mexican hat, Mexican hat-Haar as real-imaginary parts of activation function in hidden layer nodes, respectively. Results and conclusion: In contrast to CVWANN-2, which reached classification rates of 24.5%, CVWANN-1, -3 and -4 classified 40 healthy and 38 unhealthy subjects for both training and test phases with 100% correct classification rate using leave-one-out cross-validation. These networks have 100% sensitivity, 100% specifity and average detection rate is calculated as 100%. In addition, positive predictive value and negative predictive value were obtained as 100% for these networks. These results shown that CVWANN-1, -3 and -4 succeeded to classify Doppler signals. Moreover, training time and processing complexity were decreased considerable amount by using CVWANN-3. As conclusion, using of Mexican hat wavelet function in real and imaginary parts of hidden Layer activation function (CVWANN-2) is not suitable for classifying healthy and unhealthy subjects with high accuracy rate. The cause of unsuitability (obtaining the poor results in CVWANN-2) is tack of harmony between type of activation function in hidden layer and type of input signals in neural network. (c) 2007 Elsevier B.V. All rights reserved.
Concrete compressive strength detection using image processing based new test method
(ELSEVIER SCI LTD, 2017) Dogan, Gamze; Arslan, Musa Hakan; Ceylan, Murat
Today, Artificial Neural Networks (ANN) and Image Processing (IP) are particularly used to solve engineering problems. This study uses ANN and IP together to determine the mechanical properties of concrete, such as the compressive strength, modulus of elasticity and maximum deformation, at a certain success rate. In other words, the primary objective of study is to predict the mechanical properties of concrete without causing destruction, using a new alternative method. In this context, using five distinctive parameters (water/cement ratio, curing, amount of cement, compression and additive), 96 cylindrical concrete samples were produced; images of the samples were taken before they were examined at the compression testing, and the training and testing procedures for ANN and IP were realized using the obtained pressure readings at the laboratory. In addition to 96 cylindrical concrete samples, 48 were randomly selected to verify ANN and IP. From both the training/test samples and the verification samples, there is a notably high correlation between the outcomes of ANN and IP and the actual results, which varies between 97.18% and 99.87%. When ANN and IP were used together, the described method is a good alternative to the traditional destructive and nondestructive methods that are currently used to identify the mechanical properties of concrete. (C) 2017 Elsevier Ltd. All rights reserved.
Determination of Benign and Malign Lesions by Fusion of The Different Phases of Liver MR
(IEEE, 2017) Ervural, Saim; Ceylan, Murat
In this study, different phases of T1-weighted, dynamic contrast-enhanced liver magnetic resonance (MR) images were combined with wavelet-based image fusion to support decisions of radiologists. Used images has labelled as 6 different focal lesion types which focal nodular hyperplasia (FNH), hemangioma, cyst, colangiocellular carcinoma (CCC), hepatocellular carcinoma (HCC) and liver metastases. In application used images are taken by 4 different phases called pre-contrasted, arterial, portal venous, and delay venous from 30 patient. Images registered with efficient subpixel registration by cross correlation method. Discrete wavelet transform(DWT) based image fusion algorithm used and maximum selection method applied as fusion rule. As result 180 fused images obtained The performances of fusion results compared with structural similarity index (SSIM), peak to noise ratio (PSNR) and fusion factor (FF) metrics. In the fusion of portal venous phase and delay venous phase images, 98.7% SSIM and 74.95 dB PSNR values were obtained, respectively. FF value in the fusion of pre-contrast phase & arterial phase images measured as 7.258. In comparison of lesion types were represented with 98.5% SSIM
DETERMINING THE NUMBER OF TETROMINOE ORDERS FOR DENOISING APPLICATIONS PERFORMED BY TETROLET TRANSFORM
(IEEE, 2014) Ceylan, Murat; Ozturk, Ayse Elif
While Tetrolet transform (TT) which is one of the multi-resolution analysis improved recently is applied to images, 5 different shape called tetrominoes is gathered for 4x4 pixel blocks and TT is performed by arranging pixels according to this order. Tetrominoes can be chosen with 117 different combinations. In this study, firstly benchmark images (Lena, Barbara, Boat, Mandrill, Cameraman) and liver MR (magnetic resonance) images are denoised by utilizing TT with 1 to 117 different tetrominoe combinations respectively and optimal number of orders is determined for different noise rates by comparing obtained PSNR (peak signal to noise ratio) results. After that, images are denoised by using just optimal number of tetrominoe orders and all 117 combinations seperately. The operation times are compared. It is clearly seen that using all possible combinations for denoising causes redundant processes and an optimal number of tetrminoe orders can be specified for different image sets according to noise ratio. Decreasing number of combinations shortened the operation time seriously.
Effects of Complex Wavelet Transform with Different Levels in Classification of ECG Arrhytmias using Complex-Valued ANN
(IEEE, 2009) Ceylan, Murat; Ozbay, Yueksel
In this study, a new structure formed by complex wavelet transform (CHIT) with different levels and complex-valued artificial neural network (CVANN) is proposed for classification of ECG arryhytmias. In this structure, features of ECG data are extracted using CWT and data size is reduced. After then, four statistical features (maximum value, minimum value, mean value and standard deviation) are obtained from; extracted features. These new statistical features are presented to CVANN as inputs. Data set used in this studs,, including five different arrhytmias (normal sinus rhythm, right bundle branch block, left bundle branch block, atrial fibrilation and atrial flutter), are selected from MIT-BIH ECG database. Number of samples in training and test sets for each pattern is reduced from 200 real-valued samples to 100, 50 and 25 complex-valued samples using first level CWT second level CWT and third level CWT respectively. Classificaton results shown that arrhytmias are classified with 100 % accuracy rate using CWT with third level. Classification process was done in 32.62 second.
Effects of window types on classification of carotid artery Doppler signals in the early phase of atherosclerosis using complex-valued artificial neural network
(PERGAMON-ELSEVIER SCIENCE LTD, 2007) Ozbay, Yuksel; Ceylan, Murat
In this study. carotid artery Doppler ultrasound signals were acquired from left carotid arteries of 38 patients and 40 healthy volunteers. The patient group had an established diagnosis of the early phase of atherosclerosis through coronary or aortofemoropopliteal angiographies. Doppler signals were processed using fast Fourier transform (FFT) with different window types, Hilbert transform and Welch methods. After these processes. Doppler signals were classified using complex-valued artificial neural network (CVANN). Effects of window types in classification were interpreted. Results for three methods and five window types (Bartlett, Blackman, Boxcar, Hamming, Harming) were presented as comparatively. CVANN is a new technique for solving classification problems in Doppler signals. Furthermore, examining the effects of window types in addition to CVANN in this classification problem is also the first study in literature related with this subject. Results showed that CVANN, whose input data were processed by Welch method for each window types stated above, had classified all training and test patterns, which consist of 36 healthy, 34 unhealthy and four healthy, four unhealthy subjects, respectively, with 100% classification accuracy for both training and test phases. (c) 2006 Elsevier Ltd. All rights reserved.
Estimation of flexural capacity of quadrilateral FRP-confined RC columns using combined artificial neural network
(ELSEVIER SCI LTD, 2012) Koroglu, Mehmet Alpaslan; Ceylan, Murat; Arslan, Musa Hakan; Ilki, Alper
This study presents the application of combined artificial neural networks (CANNs) for the flexural capacity estimation of quadrilateral fiber-reinforced polymer (FRP) confined reinforced concrete (RC) columns. A database on quadrilateral FRP confined RC columns subjected to axial load and moment was obtained from experimental studies in the literature; CANN models were built, trained and tested. Then the flexural capacities of quadrilateral FRP confined RC columns were determined using the developed CANN model. Single and combined ANN was used for the first time in the literature for the estimation of flexural capacities of non-circular fiber-reinforced polymer (FRP) confined reinforced concrete (RC) columns. The accuracies of the proposed ANN and CANN models were more satisfactory as compared to the existing conventional approaches in the literature. Moreover, the proposed CANN models' results had lower prediction error than those of the single ANN model. (C) 2012 Elsevier Ltd. All rights reserved.
An examination on the effect of CVNN parameters while classifying the real-valued balanced and unbalanced data
(IEEE, 2018) Acar, Yunus Emre; Ceylan, Murat; Yaldiz, Ercan
In this study, a Complex-Valued Neural Network is designed to investigate the effects of the mapping angle and the learning rate on both imbalanced and balanced data. Symmetry detection problems with 3 different lengths are handled as the imbalanced data with event rates of 0.25, 0.125 and 0.0675. In order to make the data balanced, the symmetric members of the training set are resampled. The effects of the learning rate and the mapping angle are investigated for 3 different activation functions. The performance of the CVNN is measured using confusion matrix. 4-fold cross validation is used to validate the results. The results show that the CVNN is a strong tool to classify both the real valued imbalanced and balanced data with the right mapping angle and the learning rate that suit the selected activation function.
FUSION AND ANN BASED CLASSIFICATION OF LIVER FOCAL LESIONS USING PHASES IN MAGNETIC RESONANCE IMAGING
(IEEE, 2015) Ozturk, Ayse Elif; Ceylan, Murat
Detecting and diagnosing the liver focal lesions have vital importance in planning the treatments of the patients. While there is no need to apply any treatment for benign lesions, medical treatments or surgical operations are necessary in case of existence of malign lesions. Pre-contrast, arterial, portal venous and delayed venous phases in magnetic resonance imaging help to make clear diagnosis through their different contrast material holding properties. In this study, magnetic resonance images belonging to 60 patients are classified as benign/malign by using multi-resolution analysis methods and artificial neural networks. In proposed system, the magnetic resonance images taken from four different phases for each patient are merged with three multi-resolution analyses based on fusion rules and classified by using artificial neural networks. The accuracy rate of the study is reached to 90%.
Fuzzy clustering complex-valued neural network to diagnose cirrhosis disease
(PERGAMON-ELSEVIER SCIENCE LTD, 2011) Ceylan, Rahime; Ceylan, Murat; Ozbay, Yuksel; Kara, Sadik
In this study, fuzzy clustering complex-valued neural network (FCCVNN) was proposed to classify portal vein Doppler signals recorded from 54 patients with cirrhosis and 36 healthy subjects. This proposed neural network is a new model for biomedical pattern classification. The FCCVNN was composed of three phases: fuzzy clustering, calculation of FFT values and complex-valued neural network (CVNN). In first phase, fuzzy clustering was done to reduce the number of segments in training pattern. After that, FFT values of Doppler signals were calculated for pre-processing and then obtained values, which include real and imaginary components, were used as the inputs of the CVNN for classification of Doppler signals. Classification results of FCCVNN were evaluated by the different performance evaluation criterion in literature. It shows that Doppler signals were classified successfully with 100% correct classification rate using the proposed method. Moreover, the rates of sensitivity and specificity were calculated as 100% using FCCVNN method. These results were seen to be appropriate with the expected results that are derived from physician's direct diagnosis. This method would be assisted the physician to make the final decision. (C) 2011 Elsevier Ltd. All rights reserved.

Yazar "Ceylan, Murat" seçeneğine göre listele

Sayfa Başına Sonuç

Sıralama seçenekleri