Applications of Intelligent Optimization in Biology and Medicine

In this paper, we propose a new algorithm, namely genetic Nelder Mead algorithm (GNMA), for minimizing molecular potential energy function. The minimization of molecular potential energy function problem is very challenging, since the number of local minima grows exponentially with the molecular size. The new algorithm combines a global search genetic algorithm with a local search Nelder-Mead algorithm in order to search for the global minimum of molecular potential energy function. Such hybridization enhances the power of the search technique by combining the wide exploration capabilities of genetic algorithm and the deep exploitation capabilities of Nelder-Mead algorithm. The proposed algorithm can reach the global or near-global optimum for the molecular potential energy function with up to 200\(^\circ \) of freedom. We compared the proposed GNMA results with the results of 9 existing algorithms from the literature. Experimental results show efficiency of the proposed GNMA to have more accurate solutions with low computational costs.

Over the past few decades, metaheuristics methods have been applied to a large variety of bioinformatic applications. There is a growing interest in applying metaheuristics methods in the analysis of gene sequence and microarray data. Therefore, this review is intend to give a survey of some of the metaheuristics methods to analysis biological data such as gene sequence analysis, molecular 3D structure prediction, microarray analysis and multiple sequence alignment. The survey is accompanied by the presentation of the main algorithms belonging to three single solution based metaheuristics and three population based methods. These are followed by different applications along with their merits for addressing some of the mentioned tasks.

Nowadays, cryptography and steganography are considered the most famous and secured techniques used in the security field. This chapter introduces a survey on the recent DNA based steganography algorithms that use the DNA as a carrier to enable safe transfer of the critical data over the unsecure network. Some of the recent data hiding techniques based on DNA will be discussed with the main differences between them with respect to some important security parameters such as cracking probability, indicating the algorithm’s strength and weakness. However, some of these techniques do not include the computation of the cracking probability, so each of them will be reviewed and then further analyzed to compute its cracking probability. Then, the issues and the problems in these algorithms are presented and discussed with proposing some solutions and recommendations for achieving optimization in the DNA based steganography field.

In the last few decades, image registration has been established as a very active research area in computer vision. Over the years, image registration applications cover a broad range of real-world problems including remote sensing, medical imaging, artificial vision, and computer-aided design. This chapter deals with the image registration problem, in particular dental image registration using computational intelligence techniques. In the practical applications, there are many medical images needing to be registered at some time and the requirement for the time of the registration is high. Image registration is the process of transforming different sets of data into one coordinate system. Data may be multiple photographs, data from different sensors, from different times, or from different viewpoints. Registration is necessary in order to be able to compare or integrate the data obtained from these different measurements.

Location-allocation decision of special facilities in healthcare is an emergent topic that is gaining noticeable attention especially in developing countries. Given the high cost of such facilities and the need to cover as much demand as possible, systematic methods are needed to be used to handle such a problem. In this chapter, a Capacitated Maximal Covering Location Problem (CMCLP) is used in order to model this problem. A proposed modified Particle Swarm Optimization (PSO) algorithm is used to solve the CMCLP. The solutions of the modified PSO algorithm is compared to GAMS outcomes, showing much better results. The proposed algorithm denotes promised results in pinpointing good locations for the facility maximizing the covered demand.

Super-resolution (SR) concentrates on constructing a high-resolution (HR) image of a scene from two or more sets of low-resolution (LR) images of the same scene. It is the process of combining a sequence of low-resolution (LR) noisy blurred images to produce a higher-resolution image. The reconstruction of high-resolution images is computationally expensive. SR is defined to be an inverse problem that is well-known as ill-conditioned. The SR problem has been reformulated using optimization techniques to define a solution that is a close approximation of the true scene and less sensitive to errors in the observed images. This paper reviews the optimized SR reconstruction approaches and highlights its challenges and limitations. An experiment has been done to compare between bicubic, iterative back-projection (IBP), projected onto convex sets (POCS), total variation (TV) and Gradient descent via sparse representation. The experimental results show that Gradient descent via sparse representation outperforms other optimization techniques.

Early prediction of breast density is clinically significant as there is an association between the risk of breast cancer development and breast density. In the present work, the performance of two computer aided diagnostic (CAD) systems has been compared for classification of breast tissue density. The work has been carried out on MIAS dataset with 322 mammographic images consisting of 106 fatty and 216 dense images. The ROIs have been selected from densest region (i.e., the center of each image, ignoring the pectoral muscle) of each mammogram. The total dataset consisted of 322 ROIs (106 fatty ROIs and 216 dense ROIs). Five statistical texture features namely, mean, standard deviation, entropy, kurtosis and skewness are evaluated from Laws’ texture energy images resulting from Laws’ masks of length 5, 7 and 9. The texture feature vectors computed from Laws’ masks of different lengths are then subjected to principal component analysis (PCA) for reduction in feature space dimensionality. The SVM and PNN classifiers are used for the classification task. It is observed that the highest classification accuracy of 92.5 % is achieved with first four principal components derived from texture features computed with Laws’ masks of length 7 by using PNN classifier and the highest classification accuracy of 94.4 % is achieved with first four principal components derived from texture features computed with Laws’ masks of length 5 by using SVM classifier. It can be concluded that the first four principal components derived from Laws’ texture energy images resulting from Laws’ masks of length 5 are sufficient to account for textural changes exhibited by fatty and dense mammograms. The promising results obtained by the proposed CAD design indicate its usefulness to assist radiologists for breast density classification.

The diabetic retinopathy disease spreads diabetes on the retina vessels thus they lose blood supply that causes blindness in short time, so early detection of diabetes prevents blindness in more than 50 % of cases. The early detection can be achieved by automated segmentation of blood vessels in retinal images which is two-class classification problem; vessel-like or non-vessel. This chapter presents an ant colony system based approach and its improvements for the segmentation of retinal blood vessels. To minimize classification complexity, time and maximizes its accuracy, features selection is an essential step for reducing data dimensionality by removing redundant features. The performance of the presented approach on the benchmark databases of retinal images is considerable and promising as it uses features that are simple, fast in computation and needn’t to be computed at multiple scales or orientations.

This chapter covers the data mining techniques applied to the processing of clinical data to detect cardiovascular diseases. Technology evaluation and rapid development in medical diagnosis have always attracted the researchers to deliver novelty. Chronic diseases such as cancer and cardiac have been under discussion to ease their treatments using computer aided diagnosis (CAD) by optimizing their architectural complexities with better accuracy rate. To design a medical diagnostic system, raw ECG Signals, clinical and laboratory results are utilized to proceed further processing and classification. The significance of an optimized system is to give timely detection with lesser but essential clinical attributes for a patient to ensue surgical or medical follow-up. Such appropriate diagnostic systems which can detect abnormalities in clinical data and signals are truly vital and various soft computing techniques based on data mining have been applied. Hybrid approaches derived from data mining algorithms are immensely incorporated for extraction and classification of clinical records to eliminate possible redundancy and missing details which can cause worse overhead issues for the designed systems. It also extends its applications in selection, processing and ranking clinical attributes which are integral components of any medical diagnostic system. Such systems are evaluated by determining the performance measures such as system’s accuracy, sensitivity and specificity. Various supervised and unsupervised learning algorithms have been ensemble with feature processing methods to optimize in the best possible manner to detect cardiac abnormalities. This chapter analyzes all the earlier applied approaches for the cardiac disease and highlights the associated inadequacies. It also includes the architectural constraints of developing classification models. Hybrid methodologies combined with requisite clinical extraction and ranking tools to enhance system’s efficiency are also discussed. This systematic analysis of recent applied approaches for cardiac disease, aids in the domain of clinical data processing to discuss the present limitations and overcome the forthcoming complexity issues in terms of time and memory. Further, it explains that how efficient techniques for data processing and classification have not been used appropriately by considering their strengths in either phase, which leads to processing overhead and increased false alarms. Overall, the aim of this chapter is to resolve assorted concerns and challenges for designing optimized cardiac diagnostic systems with well tuned architecture.

A Fast Fuzzy C-Means (FFCM) clustering algorithm, optimized by the Particle Swarm Optimization (PSO) method, referred to as PSOFFCM, has been introduced and applied on liver CT images. Compared to FFCM, the proposed approach leads to higher values in terms of Jaccard Index and Dice Coefficient, and thus, indicating higher similarity with the ground truth provided. Based on ANOVA analysis, PSOFFCM showed better results in terms of Dice Coefficient. It also showed better mean values in terms of Jaccard Index and Dice Coefficient based on the box and whisker plots.

Predicting DNA-binding proteins computationally based on proteins features is a very challenging process. This is due to the diversity of DNA-binding patterns and classes. Therefore, the accurate prediction of DNA-binding proteins and their classes is essential. This chapter proposes efficient protein feature representations for the prediction of DNA-binding proteins and their classes. The prediction results achieved are comparable or superior to previously published results using different benchmark datasets. A protein representation of sequence, psychochemical and structural features achieved accuracy improvement of about 7 % on average for the prediction of DNA-binding proteins. Moreover, a newly proposed structure-based protein representation that takes distance and angle patterns into accounts was evaluated for DNA-binding proteins prediction. It achieved when combined with other feature representations improvement in accuracy over previously published results about 7 and 12 % on average for the prediction of DNA-binding proteins and DNA-binding protein classes, respectively. All results were evaluated using two classifiers, Random Forest and SVM.

MEDLINE is the largest biomedical literature database. It is updated daily with 200–4,000 citations. This permanent growth induces the need of a good MEDLINE abstract clustering to accelerate the procedure of research and information retrieval. Several works have been developed in this context, but clustering MEDLINE abstracts are still an area where researchers are trying to propose new approaches to better clustering. Over the last few years, evolutionary algorithms have been widely applied to clustering problems because of their ability to avoid local optimal solutions and converge to a global one. In this article, a new approach is proposed for clustering MEDLINE abstracts based on an extension of an evolutionary algorithm which is the genetic algorithm combined with a Vector Space Model and an agglomerative algorithm.

Breast cancer is the most frequently diagnosed cancer in women worldwide and the leading cause of cancer death among females. Currently the most effective method for early detection and screening of breast abnormalities is mammography. Computer aided design (CAD) systems are used to assist radiologists in better classification of tumor in a mammography as benign or malignant. Ensemble classifier construction has received considerable attention in the recent years. In the modeling of classifier ensemble, many researchers believe that the success of classifier ensembles only when classifier members present diversity among themselves. The most widely used ensemble creation techniques are focused on incorporating the concept of diversity with the construction of different features subsets or selection of the most diverse components from initial classifiers pool. Therefore the motivation of this work is to propose a CAD system using a novel classification approach based on feature selection and static classifier selection schemes.