Showing papers in "Journal of Medical Imaging and Health Informatics in 2012"

White Blood Cell Segmentation for Acute Leukemia Bone Marrow Images

Abstract: Morphological diagnosis of the blood and bone marrow smear under the microscope is an important preliminary step in the diagnosis of acute leukemia. The features and differential counts of the cells provide valuable information to the specialist to confirm the diagnosis and begin treatment, thus increasing the chance of survival of the patient. Manual diagnosis procedures are often tedious, labor intensive and time consuming. A computerised system can help accelerating the morphological diagnosis process. The proposed method consists of gradient magnitude, thresholding, morphological operations and watershed transform to perform cells segmentation. 50 images from subtypes M2, M5 and M6 were used to test the proposed method and the result showed that the method managed to obtain qualitatively good segmentation results. The segmentation accuracy for the tested image is 94.5% while the overage accuracies for the other subtypes are 94.58%, 95.06% and 95.65% respectively.

Measuring Pulse Rate with a Webcam

Abstract: In this paper a simple method of measuring the pulse rate is presented. Elaborated algorithm allows for efficient pulse rate registration directly from face images captured from a webcam. The desired signal is obtained by proper channel selection and principal component analysis. To determine the accuracy of the method an ECG signal is collected together with a video recordings. The effectiveness of the algorithm is considered for different regions of interest, several combinations of color channels and different lightening conditions. The time of principal components computation is compared with that obtained for independent component analysis. The performed analysis shows that the accuracy of the method is dependent on the face illumination but when considering home health care monitoring the proposed technique may be very useful due to its simplicity.

Hybrid Total Variation and Wavelet Thresholding Speckle Reduction for Medical Ultrasound Imaging

Abstract: Ultrasound imaging is a widely used and safe medical diagnostic technique, due to its noninvasive nature, low cost, capability of forming real time imaging, and the continuing improvements in image quality. However, the usefulness of ultrasound imaging is degraded by the presence of signal dependent noise known as speckle. In this paper, we propose a new method for speckle reduction and coherence enhancement of ultrasound images based on a hybrid of total variation (TV) method and wavelet thresholding. In this model, a noisy image is decomposed into four subbands in wavelet domain. The low frequency subband contains the low frequency coefficients with less noise that can be easily eliminated using TV-based method. More edges and other detailed information like textures are contained in the other three subbands the wavelet based soft thresholding is applied on these three subbands. In the last step we use TV method to get the final denoised image since the TV is the ability of preserving edge is smoothening by wavelet thresholding. The proposed method is compared with previous methods as applied to simulated and real data using quantitative quality evaluation metrics to show the advantage of the new method.

A Technology Platform for a Novel Home Care Delivery Service to Patients with Dementia

Abstract: Chronic illnesses impose a great burden on the lives of citizens worldwide. In modern health-care, decentralisation, dehospitalisation and self management of diseases at home are crucial factors for improving the every-day life of the patients and the people close to them. People in general tend to dislike obtrusive monitoring on their daily activities, so the challenge for home care solutions is to implement systems that provide clinicians with adequate and concise information on their patients’ health status while at the same time be unobtrusive and easy to use. Moreover, such systems must ensure that they produce high impact warnings on the patient’s status only when it is needed, in order to relieve clinicians from unnecessary workload and become a real tool for decision making and efficient patient follow-up. ALADDIN’s objective is to develop a trustworthy and reliable system supporting patients with dementia and their informal carers in the management of the disease from home. Based on a set of monitoring parameters and measuring scales feeding a reconfigurable Event Detection mechanism used for Risk Assessment and Analysis, the system aims to early detect symptoms that predict decline, avoid emergencies and secondary effects and, ultimately, prolong the period that patients can remain safely cared at home. Informal carers are also closely monitored by the system whereas additional features supporting networking, education and cognitive stimulation are also integrated along with decision support and patient management tools for the treating clinicians. The platform has been built based on credible methodologies for efficient patient follow-up, risk detection and adaptive care. It is an open, secure, interoperable, integrated IT-solution designed according to Service Oriented Architecture principles. The benefits of this platform are expected to lie in the prevention of emergencies, in reduction of carer burden and in maintenance of the patient’s and carer’s quality of life.

Monitoring of a bathing person

Abstract: W pracy przedstawiono koncepcje i wstepne wyniki ukladu pozwalającego na monitorowanie stanu i aktywności osoby kąpiącej sie. Zaprezentowany system pozwala na wykrycie osoby kąpiącej sie w wannie, analize jej aktywności oraz detekcje stanow potencjalnie niebezpiecznych. W artykule pokazano metode pomiaru, dokonano analizy czulości, zaprezentowano prototyp ukladu pomiarowego i wyniki wstepnych pomiarow.

A review of fundus image analysis for the automated detection of diabetic retinopathy

Abstract: Diabetic related eye diseases like Diabetic retinopathy (DR), Diabetic maculopathy (DM) and Glaucoma are a major cause of blindness worldwide. The early detection of these diseases plays a very significant role in the prevention of vision loss. In the last few years medical image processing of the retinal digital fundus images has emerged as a very important research area to aid an ophthalmologist in clinical diagnosis. The detection of retinal landmarks of fundus image like the Optic disk (OD), fovea and the retinal vessels is very significant in the automated detection of Diabetic retinopathy (DR), Diabetic maculopathy (DM) and Glaucoma. This review outlines the principles, methods and algorithms used in the automated detection of diabetic retinopathy. The recent methods used to detect the retinal landmarks and pathologies like hemorrhages, micro aneurysms (HMA), cotton wool spots and retinal exudates are discussed. We present the quantitative evaluation of various methods used for the automated detection of DR. The methodologies used by the researches in analyzing their results were also discussed

Wavelet Domain Bilateral Filtering with Subband Mixing for Magnetic Resonance Image Enhancement

Abstract: A method for magnetic resonance image denoising based on wavelet domain bilateral filtering (WDBF) is proposed. The main problem in bilateral filtering based methods is that the choice of filtration parameters has a trade-off between preserving edges and noise removal. In this work, a solution that would allow different components of the image to be filtered using different parameters is presented. The bilateral filtering is applied in a customized manner to different wavelet subbands and followed by subband mixing to form the final image. The proposed method is implemented to filter magnetic resonance images and verified both qualitatively and quantitatively. Verification of the new method was carried out on synthetic as well as real data sets. Qualitative and quantitative comparisons with present techniques indicate that the proposed method produces superior denoising results and suggesting potential for clinical application to boost the signal-to-noise ratio of low magnetic field scanners.

Early Detection of Cancer by Regression Analysis and Computer Simulation of Gene Regulatory Rules

Abstract: Cellular signaling and dynamic interaction among genes result in stable phenotype structures such as tumor or non tumor cells. Tumor and non-tumor cellular cells often contain some identical cancer causing genes but due to differences in their regulatory networks they evolve differently. As a result, if such regulatory networks are discovered one could predict whether a cell containing particular cancer genes will in fact end up to become a cancerous cell. This paper utilizes a mathematical approach to determine such regulatory rules for a set of cells containing cancer causing genes and uses computer simulation to predict whether in a long run a particular cell will evolve into a cancerous cell. The proposed process utilizes Probabilistic Boolean Networks (PBN) on two gene regulatory networks; one for tumor and one for non-tumor producing structures. The process uses a regression analysis to identify the regulatory networks and a computer simulation model to predict long term cancer potential.

Performance Evaluation of Classifiers for Detection of Alcoholics Using Electroencephalograms (EEG)

Abstract: This paper suggests an automated detection of alcoholics and controls using Electroencephalograms (EEG). EEG signals are recorded from healthy and alcoholic subjects during visual object recognition task. Threshold of 100 �V is used to remove the eye blink artifact and the gamma sub band (30–50 Hz) is extracted using elliptic band pass filter of 6th order. Different classical (spectral entropy), model based (power spectral density (PSD) estimation by Burg autoregressive (AR) model and eigenvector PSD estimation methods (MUSIC) in frequency domain are used as feature selection parameters. Three classification models such as feed forward Back propagation neural network (BPNN), Probabilistic neural network (PNN) and Support Vector Machine (SVM) networks are used to classify alcoholics and control subjects. The performances of BPNN and PNN classifiers are evaluated using 10-fold cross validation and holdout cross validation is performed to evaluate SVM classifier. Results show that model based and eigenvector based feature extraction methods are more accurate and sensitive than the conventional method such as spectral entropy estimation method to discriminate alcoholics from healthy controls. The SVM and PNN classifiers perform better than the BPNN classifier in terms of classification accuracy and computational complexity. However to achieve high classification accuracy, SVM does not require either the preprocessing of the input signal as in BPNN classifier or the tuning of the spread factor as in PNN classifier. Hence among the three classifiers used in this work, SVM seems to be an optimal classifier for classifying alcoholics and healthy subjects.