G. Nikiforidis

Spiking neural network training using evolutionary algorithms

Networks of spiking neurons can perform complex non-linear computations in fast temporal coding just as well as rate coded networks. These networks differ from previous models in that spiking neurons communicate information by the timing, rather than the rate, of spikes. To apply spiking neural networks on particular tasks, a learning process is required. Most existing training algorithms are based on unsupervised Hebbian learning. In this paper, we investigate the performance of the parallel differential evolution algorithm, as a supervised training algorithm for spiking neural networks. The approach was successfully tested on well-known and widely used classification problems.

Computer-based grading of haematoxylin-eosin stained tissue sections of urinary bladder carcinomas

PURPOSE A computer-based image analysis system was developed for assessing the malignancy of urinary bladder carcinomas in a more objective manner. Tumours characterized in accordance with the WHO grading system were classified into low-risk (grades I and II) and high-risk (grades III and IV). MATERIALS AND METHODS Images from 92 haematoxylin-eosin stained sections of urinary bladder carcinomas were digitized and analysed. An adequate number of nuclei were segmented from each image for morphologic and textural analysis. Image segmentation was performed by an efficient algorithm, which used pattern recognition methods to automatically characterize image pixels as nucleus or background. Image classification into low-risk or high-risk tumours was performed by means of the quadratic non-linear Bayesian classifier, which was designed employing 36 textural and morphological features of the nucleus. RESULTS Automatic segmentation of nuclei on all images was about 90% on average. Overall system accuracy in correctly classifying tumours into low-risk or high-risk was 88%, employing the leave-one-out method and the best combination of three textural and one morphological feature. Classification accuracy for low-risk tumours was 88.8% and for high-risk tumours 86.2%. CONCLUSION The proposed image analysis system may be of value to the objective assessment of the malignancy of urine bladder carcinomas, since it relies on nuclear parameters that are employed in visual grading and their prognostic value has been proved.

A computer-based diagnostic and prognostic system for assessing urinary bladder tumour grade and predicting cancer recurrence

Purpose : A computer-based system was designed, incorporating subjective criteria employed by pathologists in their usual microscopic observation of tissue samples and measurements of nuclear characteristics, with the purpose of automatically assessing urinary bladder tumour grade and predicting cancer recurrence. Material and Methods : Ninety-two cases with urine bladder carcinoma were diagnosed and followed-up. Forty-seven patients had cancer recurrence. Each case was represented by eight histological (subjective) features, evaluated by pathologists, and thirty-six automatically extracted nuclear features. Grading and prognosis were performed by neural-network based classifiers employing both histological and nuclear features. Results : Employing a combination of histological and nuclear features, highest classification accuracy was 82%, 80.5%, and 93.1% for tumours of grade I, II and III respectively. The prognostic-system, gave a significant prognostic assessment of 72.8% with a confidence of 74.5% that cancer might recur and of 71.1% that might not, employing two histological features and two textural nuclear features. Conclusions : The system for grading and predicting tumour recurrence may serve as a second opinion tool and features employed for designing the system may be of value to pathologists using descriptive grading systems.

An accurate, stable and efficient domain-type meshless method for the solution of MHD flow problems

The aim of the present paper is the development of an efficient numerical algorithm for the solution of magnetohydrodynamics flow problems for regular and irregular geometries subject to Dirichlet, Neumann and Robin boundary conditions. Toward this, the meshless point collocation method (MPCM) is used for MHD flow problems in channels with fully insulating or partially insulating and partially conducting walls, having rectangular, circular, elliptical or even arbitrary cross sections. MPC is a truly meshless and computationally efficient method. The maximum principle for the discrete harmonic operator in the meshfree point collocation method has been proven very recently, and the convergence proof for the numerical solution of the Poisson problem with Dirichlet boundary conditions have been attained also. Additionally, in the present work convergence is attained for Neumann and Robin boundary conditions, accordingly. The shape functions are constructed using the Moving Least Squares (MLS) approximation. The refinement procedure with meshless methods is obtained with an easily handled and fully automated manner. We present results for Hartmann number up to 10^5. The numerical evidences of the proposed meshless method demonstrate the accuracy of the solutions after comparing with the exact solution and the conventional FEM and BEM, for the Dirichlet, Neumann and Robin boundary conditions of interior problems with simple or complex boundaries.

An image-analysis system based on support vector machines for automatic grade diagnosis of brain-tumour astrocytomas in clinical routine

An image-analysis system based on the concept of Support Vector Machines (SVM) was developed to assist in grade diagnosis of brain tumour astrocytomas in clinical routine. One hundred and forty biopsies of astrocytomas were characterized according to the WHO system as grade II, III and IV. Images from biopsies were digitized, and cell nuclei regions were automatically detected by encoding texture variations in a set of wavelet, autocorrelation and parzen estimated descriptors and using an unsupervised SVM clustering methodology. Based on morphological and textural nuclear features, a decision-tree classification scheme distinguished between different grades of tumours employing an SVM classifier. The system was validated for clinical material collected from two different hospitals. On average, the SVM clustering algorithm correctly identified and accurately delineated 95% of all nuclei. Low-grade tumours were distinguished from high-grade tumours with an accuracy of 90.2% and grade III from grade IV with an accuracy of 88.3% The system was tested in a new clinical data set, and the classification rates were 87.5 and 83.8%, respectively. Segmentation and classification results are very encouraging, considering that the method was developed based on every-day clinical standards. The proposed methodology might be used in parallel with conventional grading to support the regular diagnostic procedure and reduce subjectivity in astrocytomas grading.

Estimation of artifacts induced by the Ilizarov device in quantitative computed tomographic analysis of tibiae

This study concerns the estimation of the artifacts introduced by the metallic frame of an Ilizarov apparatus during quantitative computed tomography for the assessment of neoosteogenesis. Ten cadaver tibiae were tomographed before and after the mounting of an Ilizarov device. The increase in mean density and in the relative number of pixels were used as indicators of the artifacts induced by the apparatus. No significant influence by the device was recorded in regions corresponding to a corticotomy site. In regions close to the metallic rings of the Ilizarov apparatus, measurements are distorted significantly, but with no influence on the assessment of neoosteogenesis.

Two-phase blood flow modeling and mass transport in the human aorta

In the present study we investigate the effects of both two-phase blood flow behavior and the pulsation of blood flow on the distributions of luminal surface of low density lipoproteins (LDL) concentration and oxygen flux along the wall of the human aorta. We compare the predictions of a two-phase model with those of the single phase one under both steady flow and realistic pulsatile flow conditions using a human aorta model constructed from CT images. As it has been noted, mass transfer of low-density lipoproteins (LDLs) may occur in the arterial system and is likely involved in the localization of atherogenesis. We utilized a tapered kind of the aorta in order to stabilize the flow of blood, thus delay the attenuation of the helical flow, making it move beyond the arch and into the first part of the descending aorta. The results therefore may be used to explain why the ascending aorta and the arch are relatively free of atherosclerosis. The dependence of viscosity and diffusivity on the local density is incorporated in the two-phase flow model rendering these quantities position dependent. For oxygen transport, we have compared the numerical results obtained with those utilizing the shear thinning non-Newtonian nature of blood. Finally, we examine the effect of pulsatile flow on the transport of LDLs and on the oxygen flux in the aorta.

PDA-based system with teleradiology and image analysis capabilities

The aim of the present study was to design and implement a Personal Digital Assistant (PDA)-based teleradiology system incorporating image processing and analysis facilities for use in emergency situations within a hospital environment. The system comprised a DICOM-server, connected to an MRI unit, 3 wireless access points, and 3 PDAs (HP iPaq rx3715). PDA application software was developed in MS Embedded Visual C++ 4.0. Each PDA can receive, load, process and analyze hi-quality static MR images. Image processing includes gray-scale manipulation and spatial filtering techniques while image analysis incorporates a probabilistic neural network (PNN) classifier, which was optimally designed employing a suitable combination of textural features and was evaluated using the leave-one-out method. The PNN is capable of discriminating between three major types of human brain tumors with accuracy of 86.66%. The developed application may be useful as a mobile medical teleconsultation tool.

A hierarchical decision tree classification scheme for brain tumour astrocytoma grading using support vector machines

The use of the concepts of support vector machines (SVMs) and decision tree (DT) classification as a possible methodology for the characterization of the degree of malignancy of brain tumours astrocytomas (ASTs) is proposed in this paper. A two-level hierarchical DT model was constructed for the discrimination of 87 ASTs in accordance to the WHO grading system. The first level concerned the detection of low versus high-grade tumours and the second level the detection of less aggressive as opposed to highly aggressive tumours. The decision rule at each level was based on a SVM classification methodology comprising 3 steps: i) From each biopsy, images were digitized and segmented to isolate nuclei from surrounding tissue. ii) Descriptive quantitative variables related to chromatin distribution and DNA content were generated to encode the degree of tumour malignancy. iii) Exhaustive search was performed to determine best feature combination that led to the smallest classification error. SVM classifier training was based on the leave-one-out method. Finally, SVMs were comparatively evaluated with the Bayesian classifier and the probabilistic neural network. The SVM classifier discriminated low from high-grade tumours with an accuracy of 90.8% and less from highly aggressive tumours with 85.6%. The proposed decision tree classification scheme based on SVMs and the analysis of quantitative nuclear features provide means to reduce subjectivity in grading brain tumors.

Cell-nuclear data reduction and prognostic model selection in bladder tumor recurrence

OBJECTIVE The paper aims at improving the prediction of superficial bladder recurrence. To this end, feedforward neural networks (FNNs) and a feature selection method based on unsupervised clustering, were employed. MATERIAL AND METHODS A retrospective prognostic study of 127 patients diagnosed with superficial urinary bladder cancer was performed. Images from biopsies were digitized and cell nuclei features were extracted. To design FNN classifiers, different training methods and architectures were investigated. The unsupervised k-windows (UKW) and the fuzzy c-means clustering algorithms were applied on the feature set to identify the most informative feature subsets. RESULTS UKW managed to reduce the dimensionality of the feature space significantly, and yielded prediction rates 87.95% and 91.41%, for non-recurrent and recurrent cases, respectively. The prediction rates achieved with the reduced feature set were marginally lower compared to the ones attained with the complete feature set. The training algorithm that exhibited the best performance in all cases was the adaptive on-line backpropagation algorithm. CONCLUSIONS FNNs can contribute to the accurate prognosis of bladder cancer recurrence. The proposed feature selection method can remove redundant information without a significant loss in predictive accuracy, and thereby render the prognostic model less complex, more robust, and hence suitable for clinical use.