Economic considerations make the conventional chest radiograph (X-ray) film an important ingredient in the diagnostic process. An initial clinical investigation for patients with suspected lung ailments is the study of the chest X-rays. The problem of detection for diseases in their early stages are well known using X-ray. A technique involving wavelets coefficient as the feature vector and Andrew's Curve has been proposed for detection of Mycobacterium Tuberculosis (MTB). This paper presents new and important results whereby lung cancer (LC) may be detected and differentiated from MTB. A method to calculate misclassification probabilities is given.
The development of a craniofacial database is a multidisciplinary initiative that will provide an important reference for community, security, social and medical applications. A method of landmark identifications and measurements in 3d on craniofacial patients is described. anatomical regions such as mandible, orbits, zygoma and maxilla are located, created and stored as templates of 3D CAD files for subsequent analysis. Data from these images were tested for accuracy and repeatability by comparing with direct measurements using caliper and CMM. The landmark points are reproducible in CAD system for further analysis. it was found that the approach provides a fast, accurate and efficient method for landmarks identification of the craniofacial areas in database development.
A common practice in medical diagnosis and patient management is the comparison of two chest radiographs images. The difference between two digital images at two time points is a measure of the effect of treatment on the patient. Two measures of similarity, the ordinary regression coefficients, R(s)(2) and coefficients of determination for the Unreplicated linear functional relationship model (ULFR), R(f) (2), are used to compare images for the particular case of Mycobacterium Tuberculosis (MTB). Our results suggest that a series of R2 values indicates gradual declining trends with values falling within a band. New patients with a series of R2 values falling within this band may be consider as making a good or acceptable recovery.
The purpose of this paper is to investigate the auditory discrimination skill of Malay children using computer-based method. Currently, most of the auditory discrimination assessments are conducted manually by Speech-Language Pathologist. These conventional tests are actually general tests of sound discrimination, which do not reflect the client's specific speech sound errors. Thus, we propose computer-based Malay auditory discrimination test to automate the whole process of assessment as well as to customize the test according to the specific speech error sounds of the client. The ability in discriminating voiced and unvoiced Malay speech sounds was studied for the Malay children aged between 7 and 10 years old. The study showed no major difficulty for the children in discriminating the Malay speech sounds except differentiating /g/-/k/ sounds. Averagely the children of 7 years old failed to discriminate /g/-/k/ sounds.
The treatment and therapy to be administered on breast cancer patients are dependent on the stage of the disease at time of diagnosis. It is therefore crucial to determine the stage at the earliest time possible. Tumor dissemination to axillary lymph nodes has been regarded as an indication of tumor aggression, thus the stage of the disease. Neural networks have been employed in many applications including breast cancer prognosis. The performance of the networks have often been quoted based on accuracy and mean squared error. In this paper, the performance of hybrid networks based on Multilayer Perceptron and Radial Basis Function networks to predict axillary lymph node involvement have been investigated. A measurement of how confident the networks are with respect to the results produced is also proposed. The input layer of the networks include four image cytometry features extracted from fine needle aspiration of breast lesions. The highest accuracy achieved by the hybrid networks was 69% only. However, most of the correctly predicted cases had a high confidence level.
A Barker sequence is employed for the synchronization of two photoplethysmogram (PPG) channels. The correctness of this technique is demonstrated by recording a PPG signal, injecting a Barker sequence at the start of this trace and producing a delayed version of it. After preprocessing, cross-correlation techniques are utilized for accurate time alignment of the two traces. The algorithm can correct for any time misalignment as long as the synchronization sequence appears on both channels.
Genetic algorithm is used to perform variable selection to determine the ranges of wavelengths in NIR spectral data suitable to be used as predictors in multivariate calibration model via partial least squares. The NIR spectral data consists of three components of active substances, namely human serum albumin (HSA), γ-globulin and glucose. The wavelength selection is able to improve the calibration model by selecting the wavelengths that contains information or correlated with the concentration of substances, while others non-chosen wavelengths, which contribute no information or contain noises, are excluded from the calibration model.
This paper discussed security of medical images and reviewed some work done regarding them. A fragile watermarking scheme was then proposed that could detect tamper and subsequently recover the image. Our scheme required a secret key and a public chaotic mixing algorithm to embed and recover a tampered image. The scheme was also resilient to VQ attack. The purposes were to verify the integrity and authenticity of medical images. We used 800 x 600 x 8 bits ultrasound (US) greyscale images in our experiment. We tested our algorithm for up to 50% tampered block and obtained 100% recovery for spread-tampered block.
In fetal heart monitoring using Doppler ultrasound signals the cardiac information is commonly extracted from non-directional signals. As a consequence often some of the cardiac events cannot be observed clearly which may lead to the incorrect detection of the valve and wall motions. Here, directional signals were simulated to investigate their enhancement of cardiac events, and hence provide clearer information regarding the cardiac activities. First, fetal Doppler ultrasound signals were simulated with signals encoding forward and reverse motion then obtained using a pilot frequency. The simulation results demonstrate that the model has the ability to produce realistic Doppler ultrasound signals and a pilot frequency can be used in the mixing process to produce directional signals that allow the simulated cardiac events to be distinguished clearly and correctly.
Digital watermarking medical images provides security to the images. The purpose of this study was to see whether digitally watermarked images changed clinical diagnoses when assessed by radiologists. We embedded 256 bits watermark to various medical images in the region of non-interest (RONI) and 480K bits in both region of interest (ROI) and RONI. Our results showed that watermarking medical images did not alter clinical diagnoses. In addition, there was no difference in image quality when visually assessed by the medical radiologists. We therefore concluded that digital watermarking medical images were safe in terms of preserving image quality for clinical purposes.
The Principal Component Analysis (PCA) is proposed as feature selection method in choosing a subset of channels for Visual Evoked Potentials (VEP). The selected channels are to preserve as much information present as compared to the full set of 61 channels as possible. The method is applied to classify two categories of subjects: alcoholics and non-alcoholics. The electroencephalogram (EEG) was recorded when the subjects were presented with single trial visual stimuli. The proposed method is successful in selecting the a subset of channels that contribute to high accuracy in the classification of alcoholics and non-alcoholics.
This paper describes the water compartments in healthy subjects and dengue patients on the day of defervescence of fever using bioelectrical impedance analysis. A total of 223 healthy subjects (65 males and 158 females) and 210 dengue patients (119 males and 91 females) in Hospital Universiti Kebangsaan Malaysia (HUKM), were studied. The ages for healthy subjects vary between 14 and 60 years old with mean age of 26.05 years while the ages for the dengue patients vary between 12 and 83 years old with mean age of 30.14 years. The parameters of water compartments investigated were total body water (TBW), extracellular water (ECW) and intracellular water (ICW). There were significant difference between healthy subjects and dengue patients for both male (p<0.05) and female (p<0.001) beginning on fever day 0 till fever day 4. The mean TBW and ECW values of dengue patients obtained were found to be higher than the normal healthy subjects while the mean ICW was lower. The mean TBW and ICW for male were higher than female while the mean ECW for male was lower than female for healthy subjects and dengue patients.
Classification of EEG signals extracted during mental tasks is a technique for designing brain computer interfaces (BCI). We classify EEG signals that were extracted during mental tasks using fixed autoregressive (FAR) and adaptive AR (AAR) models. Five different mental tasks from 4 subjects were used in the experimental study and combinations of 2 different mental tasks are studied for each subject. Four different feature extraction methods were used to extract features from these EEG signals: FAR coefficients computed with Burg's algorithm using 125 data points, without segmentation and with segmentation of 25 data points, AAR coefficients computed with least-mean-square (LMS) algorithm using 125 data points, without segmentation and with segmentation of 25 data points. Multilayer perception (MLP) neural network (NN) trained by the backpropagation (BP) algorithm is used to classify these features into the different categories representing the mental tasks. The best results for FAR was 92.70% while for AAR was only 81.80%. The results obtained here indicated that FAR using 125 data points without segmentation gave better classification performance as compared to AAR, with all other parameters constant.
Our objective is to automate the detection of apnea and hypopnea events in obstructive sleep apnea hypopnea (OSAH) syndrome based on analysis of arterial oxygen saturation signal alone. This is the first attempt where wavelet is used to detect OSAH events. Detection of OSAH events through wavelet depends on the fluctuations in the magnitude of the transformed coefficients, thus circumventing the problem of variability in the criteria on the magnitude and duration of the signal. Our work evaluates the performance of the wavelet transform to detect OSAH events against three conventional amplitude and duration algorithms. High performance in the detection of OSAH events can be achieved through the wavelet algorithm (score 96.55%, sensitivity 95.74% and specificity 97.02%) if the threshold on wavelet coefficients is individually tuned for each study. However, this is impossible in clinical practice. It is interesting to observe that the conventional methods based on amplitude and duration are able to attain a performance as close as this. The Nervus algorithm obtains the best result (score 96.66%, sensitivity 95.26% and specificity 97.46%) compared to the amplitude duration algorithm, the drop duration algorithm and the wavelet algorithm with global threshold, in descending order of performance.
This paper proposes a method to remove the noise in the ECG (Electrocardiogram) signals using Legendre moments. Noise is removed in the reconstructed ECG signals when lower order Legendre moments are used. RMSE (Root Mean Square Error) is used as the distortion measure for the reconstructed ECG signals. With sampling rate of 256 Hz and number of moments used is 13% of the data in each interval, experimental results show that reconstruction of ECG signal using Legendre moments can produce a smoother signal without noise while maintaining signal quality that is acceptable to cardiologist.
This study presents a data registration method for craniofacial spatial data of different modalities. The data consists of three dimensional (3D) vector and raster data models. The data is stored in object relational database. The data capture devices are Laser scanner, CT (Computed Tomography) scan and CR (Close Range) Photogrammetry. The objective of the registration is to transform the data from various coordinate systems into a single 3-D Cartesian coordinate system. The standard error of the registration obtained from multimodal imaging devices using 3D affine transformation is in the ranged of 1-2 mm. This study is a step forward for storing the spatial craniofacial data in one reference system in database.
A parametric approach is proposed for the characterization of the upper vascular tree. The novelty of the approach resides in it being differential: the processed signals are the photoplethysmograms obtained non-invasively from the left and right indexes. The investigated model is an ARX (auto-regressive with exogenous input) where the order has been estimated utilizing standard techniques. The model coefficients show a relatively high degree of stability.