Advances in Intelligent Analysis of Medical Data and Decision Support Systems

Frontmatter

Machine Learning Applications in Cancer Informatics

Abstract

Cancer informatics is a multidisciplinary field of research. It includes oncology, pathology, radiology, computational biology, physical chemistry, computer science, information systems, biostatistics, machine learning, artificial intelligence (AI), data mining and many others. Machine learning (ML) offers potentially powerful tools, intelligent methods, and algorithms that can help in solving many medical and biological problems. The variety of ML algorithms enable the design of a robust techniques and new methodologies for managing, representing, accumulating, changing, discovering, and updating knowledge in cancer-based systems. Moreover it supports learning and understanding the mechanisms that will help oncologists, radiologists and pathologists to induce knowledge from cancer information databases. This paper presents the research results of the author and his colleagues that have been carried out in recent years on using machine learning in cancer informatics. In addition the talk discusses several directions for future research.

Abdel-Badeeh M. Salem

Methods for Interpretation of Data in Medical Informatics

Abstract

An outline of a few methods in an emerging field of data analysis, ”data interpretation”, is given as pertaining to medical informatics and being parts of a general interpretation issue. Specifically, the following subjects are covered: measuring correlation between categories, conceptual clustering, and generalization and interpretation of empirically derived concepts in taxonomies. It will be shown that all of these can be put as parts of the same inquiry.

Boris Mirkin

Visual System of Sign Alphabet Learning for Poorly-Hearing Children

Abstract

Training visual systems have significant role for people with limited physical abilities. In this paper, the task of sign alphabet learning by poorly-hearing children was discussed using advanced recognition methods. Such intelligent system is an additional instrument for cultural development of children who can not learn alphabet in the usual way. The novelty of the method consists in proposed technique of features extraction and building vector models of outer contours for following identification of gestures which are associated with letters. The high variability of gestures in 3D space causes ambiguous segmentation, which makes the visual normalization necessary. The corresponding software has two modes: a learning mode (building of etalon models) and a testing mode (recognition of a current gesture). The Visual system of Russian sign alphabet learning is a real-time application and does not need high computer resources.

Margarita Favorskaya

Decorrelation of Sequences of Medical CT Images Based on the Hierarchical Adaptive KLT

Abstract

In this work is presented one new approach for processing of sequences of medical CT images, called Hierarchical Adaptive Karhunen-Loeve Transform (HAKLT). The aim is to achieve high decorrelation for each group of 9 consecutive CT images, obtained from the original larger sequence. In result, the main part of the energy of all images in one group is concentrated in a relatively small number of eigen images. This result could be obtained using the well-known Karhunen-Loeve Transform (KLT) with transformation matrix of size 9x9. However, for the implementation of the 2-levels HAKLT in each level are used 3 transform matrices of size 3x3, in result of which the computational complexity of the new algorithm is reduced in average 2 times, when compared to that of KLT with 9x9 matrix. One more advantage is that the algorithm permits parallel processing for each group of 3 images in every hierarchical level. In this work are also included the results of the algorithm modeling for sequences of real CT images, which confirm its ability to carry out efficient decorrelation. The HAKLT algorithm could be farther used as a basis for the creation of algorithms for efficient compression of sequences of CT images and for features space minimization in the regions of interest, which contain various classes of searched objects.

Roumen Kountchev, Peter Ivanov

Compression with Adaptive Speckle Suppression for Ultrasound Medical Images

Abstract

In the paper is presented one new approach for efficient processing of ultrasound medical images. The application of the algorithm for image compression based on the inverse difference pyramid (IDP) permits together with considerable compression ratio to achieve suppression of the specific (speckle) noise in ultrasound medical images. The paper describes the principle of image decomposition and its modification, designed for this medical application. Special attention is paid to achieve relatively low computational complexity of the used algorithms. Besides, an adaptive filtration aimed at the visual quality improvement of the restored image is also included. At the end of the paper are given experimental results and comparison with other contemporary methods for image archiving based on the JPEG and JPEG 2000 standards.

Roumen Kountchev, Vladimir Todorov, Roumiana Kountcheva

Adaptive Approach for Enhancement the Visual Quality of Low-Contrast Medical Images

Abstract

In the paper is presented one specific approach aimed at improvement of the visual quality of underexposed or low-contrast medical images. For this are developed adaptive contrast-enhancement algorithms, based on the segmentation of the image area with relatively high density of dark elements. The problem is solved changing the brightness intervals of the selected segments followed by equalization (in particular - linear stretch and skew) of the corresponding parts of the histogram. The implementation is relatively simple and permits easy adaptation of the contrasting algorithms to image contents, requiring setting of small number of parameters only. The corresponding software tools permit to change the image with consecutive steps and to evaluate the visual quality of the processed images. The original image is also available, which permits easy comparison and evaluation. The obtained results prove the efficiency of the new methods for image quality enhancement.

Vladimir Todorov, Roumiana Kountcheva

An Adaptive Enhancement of X-Ray Images

Abstract

Most of the X-ray images are no truly isotropic and its quality varies depending on penetration of X-rays in different anatomical structures and on the technologies of their obtaining. The noise problem arises from the fundamentally statistical nature of photon production. This paper presents an approach for X-ray image enhancement based on contrast limited adaptive histogram equalization (CLAHE), following by morphological processing and noise reduction, based on the Wavelet Packet Decomposition and adaptive threshold of wavelet coefficients in the high frequency sub-bands of the shrinkage decomposition. Implementation results are given to demonstrate the visual quality and to analyze some objective estimation parameters in the perspective of clinical diagnosis.

Veska Georgieva, Roumen Kountchev, Ivo Draganov

Medical Images Transform by Multistage PCA-Based Algorithm

Abstract

In this paper a novel approach for medical images transform by the Multistage Principal Component Analysis (MPCA) algorithm is presented. It consists of applying PCA over series of pixels grouped two by two in multiple stages. The process is extremely straightforward and the computation complexity is considerably reduced in comparison to the full PCA performed over the whole image. Promising results are achieved experimentally over a multitude of test images and the proposed approach is considered very perspective for both lossy and lossless compression of medical visual data.

Ivo Draganov, Roumen Kountchev, Veska Georgieva

A New Histogram-Based Descriptor for Images Retrieval from Databases

Abstract

In this paper, we propose a new approach for designing histogram-based descriptors. For demonstration purpose, we generate a descriptor based on the histogram of differential-turning angle scale space (d-TASS) function and its derived data. We then compare the proposed histogram-based descriptor with the traditional histogram descriptors in terms of retrieval performance from image databases. Experiments on three shapes databases demonstrate the efficiency and the effectiveness of the new technique: the proposed technique of histogram-based descriptor outperforms the traditional one. These experiments showed also that the proposed histogram-based descriptor using d-TASS function and the derived features performs well compared with the state-of-the-art. When applied to texture images retrieval, the proposed approach yields higher performance than the traditional histogram-based descriptors. From these results, we believe that the proposed histogram-based descriptor should perform efficiently for medical images retrieval so we will focus on this aspect in the future work.

Kidiyo Kpalma, Cong Bai, Miloud Chikr El Mezouar, Kamel Belloulata, Nasreddine Taleb, Lakhdar Belhallouche, Djamal Boukerroui

Combining Features Evaluation Approach in Content-Based Image Search for Medical Applications

Abstract

In this paper we propose an approach for a feature combination helping to distinguish searched images from databases by retrieving relevant images. The retrieval effectiveness of 11 well known image features, commonly used in Content Based Image Retrieval (CBIR) systems, is investigated. We suggest a combined features approach including features’ performance comparison of 57 various medical image categories from IRMA Database. The most informative 3 features, adaptive to image categories, are defined. Based on experiments and image similarity accuracy analysis we suggest a set of 3 low level features Color Layout, Edge Histogram and DCT Coefficients. The developed approach achieves better similar images retrieval results for more image classes. The results show an accuracy improvement of 14.49% on Mean Average Precision (MAP). The comparison is done to the same type performance measure of the best individual feature in different medical image categories.

Antoaneta A. Popova, Nikolay N. Neshov

Semi-automatic Ultrasound Medical Image Recognition for Diseases Classification in Neurology

Abstract

The main aim of this work is semi-automatic ROI positionig in transcranial medical images based on multi-agent systems (MAS) in preprocessing module. Designed approach is based on image processing and is realized by means of artifical intelligence, MAS, which has been experimentally designed in Matlab software environment. Within this processing has been worked with a set of TCS static images in grayscale and binary representation to experimental testing to positioning. This designed application is used for diseases classification in neurology.

Jiří Blahuta, Tomáš Soukup, Petr Čermák, David Novák, Michal Večerek

Classification and Detection of Diabetic Retinopathy

Abstract

Diabetic retinopathy (DR) is the leading cause of blindness in adults around the world today. Early detection (that is, screening) and timely treatment have been shown to prevent visual loss and blindness in patients with retinal complications of diabetes. The basis of the classification of different stages of diabetic retinopathy is the detection and quantification of blood vessels and hemorrhages present in the retinal image. In this paper, the four retinal abnormalities (microaneurysms, haemorrhages, exudates, and cotton wool spots) are located in 100 color retinal images, previously graded by an ophthalmologist. A new automatic algorithm has been developed and applied to 100 retinal images. Accuracy assessment of the classified output revealed the detection rate of the microaneurysms was 87% using the thresholding method, whereas the detection rate for the haemorrhages was 88%. On the other hand, the correct classification rate for microaneurysms and haemorrhages using the minimum distance classifier was 60% and 94% respectively. The thresholding method resulted in a correct detection rate for exudates and cotton wool spots of 93% and 89% respectively. The minimum distance classifier gave a correct rate for exudates and cotton wool spots of 95% and 86% respectively.

Ahmad Taher Azar, Valentina E. Balas

Principal Component Analysis Used in Estimation of Human’s Immune System, Suffered from Allergic Rhinosinusopathy Complicated with Clamidiosis or without It

Abstract

The immune system is very important in overcoming the influence of harmful factors to human organism. This is one of the three integrative systems of the organism, which provides maintenance of homeostasis together with the nervous and endocrine systems. In this paper, estimation of human’s immune system, suffered from allergic rhinosinusopathy, complicated or not complicated by clamidiosis is given. The novelty of method consists in using correlation adaptometry and principal component analysis for this estimation. These methods allowed estimation of changes of the immune system during stress adaptation.

Lyudmila Pokidysheva, Irina Ignatova

Computer-Aided Diagnosis of Laryngopathies in the LabVIEW Environment: Exemplary Implementation

Abstract

In the paper, we present a computer tool supporting a non-invasive diagnosis of selected larynx diseases. The tool is created on the basis of the LabVIEW environment. LabVIEW enables us to create, in an easy way, a user-friendly graphical interface facilitating both entering input data and visualizing results in order to make the platform ready to use directly in the medical community. Computer-aided diagnosis of laryngopathies, in the presented tool, is based on a family of coefficients reflecting spectrum disturbances around basic tones and their multiples for patients’ voice signals.

Dominika Gurdak, Krzysztof Pancerz, Jaroslaw Szkola, Jan Warchol

Analysis and Development of Techniques and Methods in Medical Practice in Cochlear Implant Systems

Abstract

Cochlear implant methods are the area of interest and intensive medical and technical efforts for investigations, analysis, implementation, testing and finally producing more and more precise and effective cochlear implant prosthesis. The goals of this article are first to present the cochlear implant system in historical plan, then to analyze some of the existing signal processing strategies in cochlear implant systems, which leads to the decision of the importance of filter bank design for the precision of the signal processing in the cochlear implant algorithms and practical implementations. It is presented a detailed description and critical comparison of the most useful types of cochlear filter banks and as the results from the analysis some useful conclusion are presented for the cochlear filter banks characteristics, time consuming in calculation, the importance for overall cochlear implants quality of speech sound processing, etc. The defined in conclusion assertions are accepted as the basis of future investigations in area of new cochlear implant model development.

Svetlin Antonov, Snejana Pleshkova-Bekiarska

On Confidentially Connected-Free Graphs

Abstract

Graph theory provides algorithms and tools to handle models for important applications in biology and medicine, such as drug design, diagnosis, or visualization. This paper deals with some theoretical results concerning the relationship between two classes of graphs which may be susceptible of applications in medicine and intelligent systems. The class of Confidentially Connected-free graphs is introduced and related to the class of Asteroidal Triple-free graphs, as well as to the graphs that have a star-cutset. We give a characterization of Confidentially Connected-free graphs using neighborhoods and weakly decomposition.

Mihai Talmaciu, Elena Nechita, Barna Iantovics

Route Search Algorithm in Timetable Graphs and Extension for Block Agents

Abstract

This paper describes an algorithm that determines routes using three graphs: the railway graph, the train timetable graph and the summary timetable graph. The search in the timetable graphs is guided by a subgraph of the railway graph, which is defined by the nodes that form an ellipse around the minimum distance path from departure to arrival. We also present some performance evaluations of our proposed algorithm. Finally we describe an extension of this algorithm that can be used in conjunction with block agents to find routes in large timetable graphs, and some applications for medical domain.

Ion Cozac

From Individual EHR Maintenance to Generalised Findings: Experiments for Application of NLP to Patient-Related Texts

Abstract

Experiments in automatic analysis of free texts in Bulgarian hospital discharge letters are presented. Natural Language Processing (NLP) has been applied to medical texts since decades but high-quality results have been demonstrated only recently. The progress in automatic text analysis opens new directions for secondary use of Electronic Health Records (EHR). It enables also the design and development of software systems which provide better patient access to his/her health records as well as better maintenance of large EHR archives. We report about successful extraction of important patient-related entities from hospital EHR texts and consider several scenarios for application of NLP modules in healthcare software systems.

Galia Angelova, Dimitar Tcharaktchiev, Svetla Boytcheva, Ivelina Nikolova, Hristo Dimitrov, Zhivko Angelov

A Terminology Indexing Based on Heuristics Using Linguistic Resources for Medical Textual Corpus

Abstract

The term extraction is an important step in building a resource of indexing and many strong tools are available for many languages. This complex process, which identifies candidate terms may become indexes for annotations, is often subject to the problem of lack of relevance of calculated terms. As a result, extractor terms must be strong to handle the errors and suggest better results, without encumbering the user with too many proposed index. In this respect, we are suggesting a new indexing approach based on a hybrid of terminologies extraction using a filter by removing terms and operates upon corpus of medical texts.

Ali Benafia, Ramdane Maamri, Zaidi Sahnoun

Architecture for Medical Image Processing

Abstract

In this paper, software architecture for medical image processing, analysis and archiving is presented. On the basis of the considered architecture a new task-oriented medical image processing system, which allows imitating of the human visual system, is developed. The basic functions include input/output of halftone images, pre- and post-processing, filtration, compression, enhancement, 2D linear transforms, pseudo-color transforms, analysis and interpolations. Using the system features, various image processing tasks are semantically described in the experimental part. The main advantages of the proposed architecture are the use of adaptive algorithms for processing of medical images, tailored to their specific features.

Rumen Mironov, Roumen Kountchev

A New Generation of Biomedical Equipment: FPGA

Abstract

This chapter aims to broaden the bridge that covers the gap between the biomedical science community and engineers, by encouraging the developers and the users of biomedical equipment to apply at a large scale and to promote the Field-Programmable Gate Array technology. One provides a brief recall of this technology and of its key advantages: high electrical performances (great complexity, high speed, low energy consumption, etc.), extremely short time-to-market, high reliability even in field conditions, flexibility, portability, standardization, etc. A tutorial description of the FPGA development methodology is also provided. Some of the most successful FPGA applications in this field are mentioned: medical imaging, minimally-invasive surgery platforms, radio-frequency identification, etc.

Marius M. Balas

Backmatter