Functional link convolutional neural network for the classification of diabetes mellitus.

Diabetes is a faction of metabolic ailments distinguished by hyperglycemia which is the consequence of a defect, in the action of insulin, insulin secretion, or both and producing various abnormalities in the human body. In recent years, the utilization of intelligent systems has been expanded in disease classification and numerous researches have been proposed. In this research article, a variant of Convolutional Neural Network (CNN) that is, Functional Link Convolutional Neural Network (FLCNN) is proposed for the diabetes classification. The main goal of this article is to find the potential of a computationally less complex deep learning network like FLCNN and applied the proposed technique on a real dataset of diabetes for classification. This article also presents the comparative studies where various other machine learning techniques are implemented and outcomes are compared with the proposed FLCNN network. The performance of each classification techniques have been evaluated based on standard measures and also validated with a non-parametric statistical test such as Friedman. Data for modeling diabetes classification is collected from Bombay Medical Hall, Upper Bazar, Ranchi, India. Accuracy achieve by the proposed classifier is more than 90% which is closer to the other state-of-the-art implemented classifiers.

Cat Swarm Optimization based Functional Link Multilayer Perceptron for Suppression of Gaussian and Impulse Noise from Computed Tomography Images.

BACKGROUND: The Gaussian and impulse noises corrupt the Computed Tomography (CT) images either individually or collectively, and the conventional fixed filters do not have the potential to suppress these noise. OBJECTIVES: These spurious noises affect the inherent features of CT image awkwardly. Hence, to handle such a situation adaptive Cat Swarm Optimization based Functional Link Multilayer Perceptron (CSO-FLMLP) has been proposed in this paper to get rid of unwanted noise from the CT images. METHODS: Here, the nature-inspired CSO technique which is an optimization algorithm has been employed to assist in updating the weights of FLMLP network. In this work, the cost function considered for CSO is the error between noisy and contextual pixels of reference images which need to minimize. For examining the efficiency of CSO-FLMLP filter, it is compared with the other six competitive adaptive filters. RESULTS: The performance of proposed approach and other state-of-the-art filters are compared on the basis of performance metrics like the structural similarity index (SSIM), peak signal to noise ratio (PSNR), computational time and convergence rate. Supremacy of CSO-FLMLP among the considered adaptive filters is validated through Friedman statistical test. CONCLUSION: The CSO-FLMLP adaptive filter could successfully re-move the dominant Gaussian, impulse or combination of both noises from the clinical CT images.

Implementation and Analysis of Classification Algorithms for Diabetes.

BACKGROUND: In this era of cutting edge research, though one of the ubiquitous facilities accessible to modern man is state of the art medical care yet diabetes has emerged as one of the major ailments afflicting the present generation. So the prime necessity of this age has transformed into providing cheap and sustainable medical care against such major diseases like diabetes. In layman's terms Diabetes may be defined as a physiological condition wherein the blood glucose level is more than the prescribed level on a regular basis. OBJECTIVES: So the prime objective of this work is to provide a novel classification technique for detection of diabetes in a timely and effective manner. METHODS: The proposed work comprises of four phases: In the first phase a "Localized Diabetes Dataset" has been compiled and collected from Bombay Medical Hall, Mahabir Chowk, Pyada Toli, Upper Bazar, Jharkhand, Ranchi, India. In the second phase various classification techniques namely RBF NN, MLP NN, NBs, and J48graft DT have been applied on the Localized Diabetes Dataset. In the third phase, Genetic algorithm (GA) has been utilized as an attribute selection technique through which six attributes among twelve attributes have been filtered. Lastly in the fourth phase RBF NN, MLP NN, NBs and J48graft DT has been utilized for classification on relevant attributes obtained by GA. RESULTS: In this study, comparative analysis of outcomes obtained by with and without the use of GA for the same set of classification technique has been done w.r.t performance assessment. It has been conclusively inferred that GA is helpful in removing insignificant attributes, reducing the cost and computation time while enhancing ROC and accuracy. CONCLUSION: The utilized strategy may likewise be executed for other medical issues.

Comparative Analysis of Classification Methods with PCA and LDA for Diabetes.

BACKGROUND: The modern society is extremely prone to many life-threatening diseases, which can be easily controlled as well as cured if diagnosed at an early stage. The development and implementation of a disease diagnostic system have gained huge popularity over the years. In the current scenario, there are certain factors such as environment, sedentary lifestyle, genetic (hereditary) are the major factors behind the life threatening diseases such as 'diabetes.' Moreover, diabetes has achieved the status of the modern man's leading chronic disease. So one of the prime needs of this generation is to develop a state-of-the-art expert system which can predict diabetes at a very early stage with a minimum of complexity and in an expedited manner. The primary objective of this work is to develop an indigenous and efficient diagnostic technique for detection of diabetes. Method & Discussion: The proposed methodology comprises of two phases: In the first phase The Pima Indian Diabetes Dataset (PIDD) has been collected from the UCI machine learning repository databases and Localized Diabetes Dataset (LDD) has been gathered from Bombay Medical Hall, Upper Bazar Ranchi, Jharkhand, India. In the second phase, the dataset has been processed through two different approaches. The first approach entails classification through Adaboost, Classification via Regression (CVR), Radial Basis Function Network (RBFN), K-Nearest Neighbor (KNN) on Pima Indian Diabetes Dataset and Localized Diabetes Dataset. In the second approach, Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) have been applied as a feature reduction method followed by using the same set of classification methods used in the first approach. Among all of the implemented classification methods, PCA_CVR achieves the maximum performance for both the above mentioned datasets. CONCLUSION: In this article, comparative analysis of outcomes obtained by with and without the use of PCA and LDA for the same set of classification method has been done w.r.t performance assessment. Finally, it has been concluded that PCA & LDA both are useful to remove the insignificant features, decreasing the expense and computation time while improving the ROC and accuracy. The used methodology may similarly be applied to other medical diseases.