Monitoring the depth of anesthesia (DOA) during surgery is very important in order to avoid patients' interoperative awareness. Since the traditional methods of assessing DOA which involve monitoring the heart rate, pupil size, sweating etc, may vary from patient to patient depending on the type of surgery and the type of drug administered, modern methods based on electroencephalogram (EEG) are preferred. EEG being a nonlinear signal, it is appropriate to use nonlinear chaotic parameters to identify the anesthetic depth levels. This paper discusses an automated detection method of anesthetic depth levels based on EEG recordings using non-linear chaotic features and neural network classifiers. Three nonlinear parameters, namely, correlation dimension (CD), Lyapunov exponent (LE) and Hurst exponent (HE) are used as features and two neural network models, namely, multi-layer perceptron network (feed forward model) and Elman network (feedback model) are used for classification. The neural network models are trained and tested with single and multiple features derived from chaotic parameters and the performances are evaluated in terms of sensitivity, specificity and overall accuracy. It is found from the experimental results that the Lyapunov exponent feature with Elman network yields an overall accuracy of 99% in detecting the anesthetic depth levels.
Two-stage lossless data compression methods involving predictors and encoders are well known. This paper discusses the application of context based error modeling techniques for neural network predictors used for the compression of EEG signals. Error modeling improves the performance of a compression algorithm by removing the statistical redundancy that exists among the error signals after the prediction stage. In this paper experiments are carried out by using human EEG signals recorded under various physiological conditions to evaluate the effect of context based error modeling in the EEG compression. It is found that the compression efficiency of the neural network based predictive techniques is significantly improved by using the error modeling schemes. It is shown that the bits per sample required for EEG compression with error modeling and entropy coding lie in the range of 2.92 to 6.62 which indicates a saving of 0.3 to 0.7 bits compared to the compression scheme without error modeling.
Diabetic retinopathy (DR) has become a serious threat in our society, which causes 45% of the legal blindness in diabetes patients. Early detection as well as the periodic screening of DR helps in reducing the progress of this disease and in preventing the subsequent loss of visual capability. This paper provides an automated diagnosis system for DR integrated with a user-friendly interface. The grading of the severity level of DR is based on detecting and analyzing the early clinical signs associated with the disease, such as microaneurysms (MAs) and hemorrhages (HAs). The system extracts some retinal features, such as optic disc, fovea, and retinal tissue for easier segmentation of dark spot lesions in the fundus images. That is followed by the classification of the correctly segmented spots into MAs and HAs. Based on the number and location of MAs and HAs, the system quantifies the severity level of DR. A database of 98 color images is used in order to evaluate the performance of the developed system. From the experimental results, it is found that the proposed system achieves 84.31% and 87.53% values in terms of sensitivity for the detection of MAs and HAs respectively. In terms of specificity, the system achieves 93.63% and 95.08% values for the detection of MAs and HAs respectively. Also, the proposed system achieves 68.98% and 74.91% values in terms of kappa coefficient for the detection of MAs and HAs respectively. Moreover, the system yields sensitivity and specificity values of 89.47% and 95.65% for the classification of DR versus normal.
Retinal blood vessel detection and analysis play vital roles in early diagnosis and prevention of several diseases, such as hypertension, diabetes, arteriosclerosis, cardiovascular disease and stroke. This paper presents an automated algorithm for retinal blood vessel segmentation. The proposed algorithm takes advantage of powerful image processing techniques such as contrast enhancement, filtration and thresholding for more efficient segmentation. To evaluate the performance of the proposed algorithm, experiments were conducted on 40 images collected from DRIVE database. The results show that the proposed algorithm yields an accuracy rate of 96.5%, which is higher than the results achieved by other known algorithms.
The increasing number of diabetic retinopathy (DR) cases world wide demands the development of an automated decision support system for quick and cost-effective screening of DR. We present an automatic screening system for detecting the early stage of DR, which is known as non-proliferative diabetic retinopathy (NPDR). The proposed system involves processing of fundus images for extraction of abnormal signs, such as hard exudates, cotton wool spots, and large plaque of hard exudates. A rule based classifier is used for classifying the DR into two classes, namely, normal and abnormal. The abnormal NPDR is further classified into three levels, namely, mild, moderate, and severe. To evaluate the performance of the proposed decision support framework, the algorithms have been tested on the images of STARE database. The results obtained from this study show that the proposed system can detect the bright lesions with an average accuracy of about 97%. The study further shows promising results in classifying the bright lesions correctly according to NPDR severity levels.
This paper presents the results obtained for medical image compression using autoencoder neural networks. Since mammograms (medical images) are usually of big sizes, training of autoencoders becomes extremely tedious and difficult if the whole image is used for training. We show in this paper that the autoencoders can be trained successfully by using image patches instead of the whole image. The compression performances of different types of autoencoders are compared based on two parameters, namely mean square error and structural similarity index. It is found from the experimental results that the autoencoder which does not use Restricted Boltzmann Machine pre-training yields better results than those which use this pre-training method.
The volume of patient monitoring video acquired in hospitals is very huge and hence there is a need for better compression of the same for effective storage and transmission. This paper presents a new motion segmentation technique, which improves the compression of patient monitoring video. The proposed motion segmentation technique makes use of a binary mask, which is obtained by thresholding the standard deviation values of the pixels along the temporal axis. Two compression methods, which make use of the proposed motion segmentation technique, are presented. The first method uses MPEG-4 coder and 9/7-biorthogonal wavelet for compressing the moving and stationary portions of the video respectively. The second method uses 5/3-biorthogonal wavelet for compressing both the moving and the stationary portions of the video. The performances of these compression algorithms are evaluated in terms of PSNR and bitrate. From the experimental results, it is found that the proposed motion technique improves the performance of the MPEG-4 coder. Among the two compression methods presented, the MPEG-4 based method performs better for bitrates less than 767 Kbps whereas for bitrates above 767 Kbps the performance of the wavelet based method is found superior.
Many medical examinations involve acquisition of a large series of slice images for 3D reconstruction of the organ of interest. With the paperless hospital concept and telemedicine, there is very heavy utilization of limited electronic storage and transmission bandwidth. This paper proposes model-based compression to reduce the load on such resources, as well as aid diagnosis through the 3D reconstruction of the structures of interest, for images acquired by various modalities, such as MRI, Ultrasound, CT, PET etc. and stored in the DICOM file format. An example implementation for the biliary track in MRCP images is illustrated in the paper. Significant compression gains may be derived from the proposed method, and a suitable mixture of the models and raw images would enhance the patient medical history archives as the models may be stored in the DICOM file format used in most medical archiving systems.
Stones in the biliary tract are routinely identified using MRCP (magnetic resonance cholangiopancreatography). The noisy nature of the images, as well as varying intensity, size and location of the stones, defeat most automatic detection algorithms, making computer-aided diagnosis difficult. This paper proposes a multi-stage segment-based scheme for semi-automated detection of choledocholithiasis and cholelithiasis in the MRCP images, producing good performance in tests, differentiating them from "normal" MRCP images. With the high success rate of over 90%, refinement of the scheme could be applicable in the clinical environment as a tool in aiding diagnosis, with possible applications in telemedicine.
Tumors are generally difficult to detect in Magnetic Resonance (MR) images as they can be of varying intensities and do not appear as clear structures on these images. This difficulty is more prominent in MR Cholangiopancreatography (MRCP), which is the MR technology using a special sequence of T2-weighted imaging to identify the biliary tract, pancreatic duct, and gallbladder in the liver region, as MRCP images are more noisy in nature and are acquired for a more focused area with too much flexibility in position orientation for convenient computer-aided diagnosis. Based on the principle that the tumor mass manifests itself as blockage of the biliary tree structure, this paper introduces a technique that uses a region growing algorithm to identify discontinuities in the biliary tree as a means to preliminary detection of a possible tumor, in a fashion similar to the visual observation used by most radiologists in making their preliminary diagnosis. Through the use of appropriate image normalization, watershed segmentation, thresholding, rule-based region growing, and region analysis, the proposed technique is shown in this paper to be successful in identifying MRCP images with liver carcinoma from those with normal liver. Acquisition standardization, interactive image selection, and optimum image orientation will further enhance the accuracy of this proposed scheme for use in aiding clinical diagnosis at medical institutions.
Electroencephalogram (EEG) signal plays an important role in the diagnosis of epilepsy. The long-term EEG recordings of an epileptic patient obtained from the ambulatory recording systems contain a large volume of EEG data. Detection of the epileptic activity requires a time consuming analysis of the entire length of the EEG data by an expert. The traditional methods of analysis being tedious, many automated diagnostic systems for epilepsy have emerged in recent years. This paper discusses an automated diagnostic method for epileptic detection using a special type of recurrent neural network known as Elman network. The experiments are carried out by using time-domain as well as frequency-domain features of the EEG signal. Experimental results show that Elman network yields epileptic detection accuracy rates as high as 99.6% with a single input feature which is better than the results obtained by using other types of neural networks with two and more input features.
In this paper, 3-D discrete Hartley transform is applied for the compression of two medical modalities, namely, magnetic resonance images and X-ray angiograms and the performance results are compared with those of 3-D discrete cosine and Fourier transforms using the parameters such as PSNR and bit rate. It is shown that the 3-D discrete Hartley transform is better than the other two transforms for magnetic resonance brain images whereas for the X-ray angiograms, the 3-D discrete cosine transform is found to be superior.
This paper focuses on the detection of retinal blood vessels which play a vital role in reducing the proliferative diabetic retinopathy and for preventing the loss of visual capability. The proposed algorithm which takes advantage of the powerful preprocessing techniques such as the contrast enhancement and thresholding offers an automated segmentation procedure for retinal blood vessels. To evaluate the performance of the new algorithm, experiments are conducted on 40 images collected from DRIVE database. The results show that the proposed algorithm performs better than the other known algorithms in terms of accuracy. Furthermore, the proposed algorithm being simple and easy to implement, is best suited for fast processing applications.
Due to increasing number of diabetic retinopathy cases, ophthalmologists are experiencing serious problem to automatically extract the features from the retinal images. Optic disc (OD), exudates, and cotton wool spots are the main features of fundus images which are used for diagnosing eye diseases, such as diabetic retinopathy and glaucoma. In this paper, a new algorithm for the extraction of these bright objects from fundus images based on marker-controlled watershed segmentation is presented. The proposed algorithm makes use of average filtering and contrast adjustment as preprocessing steps. The concept of the markers is used to modify the gradient before the watershed transformation is applied. The performance of the proposed algorithm is evaluated using the test images of STARE and DRIVE databases. It is shown that the proposed method can yield an average sensitivity value of about 95%, which is comparable to those obtained by the known methods.
The detection of bright objects such as optic disc (OD) and exudates in color fundus images is an important step in the diagnosis of eye diseases such as diabetic retinopathy and glaucoma. In this paper, a novel approach to automatically segment the OD and exudates is proposed. The proposed algorithm makes use of the green component of the image and preprocessing steps such as average filtering, contrast adjustment, and thresholding. The other processing techniques used are morphological opening, extended maxima operator, minima imposition, and watershed transformation. The proposed algorithm is evaluated using the test images of STARE and DRIVE databases with fixed and variable thresholds. The images drawn by human expert are taken as the reference images. The proposed method yields sensitivity values as high as 96.7%, which are better than the results reported in the literature.
Blood vessel detection in retinal images is a fundamental step for feature extraction and interpretation of image content. This paper proposes a novel computational paradigm for detection of blood vessels in fundus images based on RGB components and quadtree decomposition. The proposed algorithm employs median filtering, quadtree decomposition, post filtration of detected edges, and morphological reconstruction on retinal images. The application of preprocessing algorithm helps in enhancing the image to make it better fit for the subsequent analysis and it is a vital phase before decomposing the image. Quadtree decomposition provides information on the different types of blocks and intensities of the pixels within the blocks. The post filtration and morphological reconstruction assist in filling the edges of the blood vessels and removing the false alarms and unwanted objects from the background, while restoring the original shape of the connected vessels. The proposed method which makes use of the three color components (RGB) is tested on various images of publicly available database. The results are compared with those obtained by other known methods as well as with the results obtained by using the proposed method with the green color component only. It is shown that the proposed method can yield true positive fraction values as high as 0.77, which are comparable to or somewhat higher than the results obtained by other known methods. It is also shown that the effect of noise can be reduced if the proposed method is implemented using only the green color component.
Cardiovascular disease (CVD) is the major cause of death globally. More people die of CVDs each year than from any other disease. Over 80% of CVD deaths occur in low and middle income countries and occur almost equally in male and female. In this paper, different computational models based on Bayesian Networks, Multilayer Perceptron,Radial Basis Function and Logistic Regression methods are presented to predict early risk detection of the cardiovascular event. A total of 929 (626 male and 303 female) heart attack data are used to construct the models.The models are tested using combined as well as separate male and female data. Among the models used, it is found that the Multilayer Perceptron model yields the best accuracy result.