The demand for automatically classification of medical X-ray images is rising faster than ever. In this paper, an approach is presented to gain high accuracy rate for those classes of medical database with high ratio of intraclass variability and interclass similarities. The classification framework was constructed via annotation using the following three techniques: annotation by binary classification, annotation by probabilistic latent semantic analysis, and annotation using top similar images. Next, final annotation was constructed by applying ranking similarity on annotated keywords made by each technique. The final annotation keywords were then divided into three levels according to the body region, specific bone structure in body region as well as imaging direction. Different weights were given to each level of the keywords; they are then used to calculate the weightage for each category of medical images based on their ground truth annotation. The weightage computed from the generated annotation of query image was compared with the weightage of each category of medical images, and then the query image would be assigned to the category with closest weightage to the query image. The average accuracy rate reported is 87.5 %.
Breast cancer mostly arises from the glandular (dense) region of the breast. Consequently, breast density has been found to be a strong indicator for breast cancer risk. Therefore, there is a need to develop a system which can segment or classify dense breast areas. In a dense breast, the sensitivity of mammography for the early detection of breast cancer is reduced. It is difficult to detect a mass in a breast that is dense. Therefore, a computerized method to separate the existence of a mass from the glandular tissues becomes an important task. Moreover, if the segmentation results provide more precise demarcation enabling the visualization of the breast anatomical regions, it could also assist in the detection of architectural distortion or asymmetry. This study attempts to segment the dense areas of the breast and the existence of a mass and to visualize other breast regions (skin-air interface, uncompressed fat, compressed fat, and glandular) in a system. The graph cuts (GC) segmentation technique is proposed. Multiselection of seed labels has been chosen to provide the hard constraint for segmentation of the different parts. The results are promising. A strong correlation (r = 0.93) was observed between the segmented dense breast areas detected and radiological ground truth.
Echocardiogram is an ultrasound image of the heart that demonstrates the size, motion and composition of cardiac structures and is also used to diagnose various abnormalities of the heart including abnormal chamber size, shape and congenital heart disease. Echocardiography provides important morphological and functional details of the heart. Most of the presented automatic cardiac disease recognition systems that use echocardiograms based on defective anatomical region detection. In this paper we present a simple technique for cardiac geometry detection via echocardiogram images which conquer these borders and exploits cues from cardiac structure. To demonstrate the effectiveness of this technique, we present results for cardiac geometry detection through difference intensity of echocardiogram images. We have developed a simple program code for the prediction of cardiac geometry using difference intensity of echocardiogram images. With this code, users can generate node or point for detection of cardiac geometry as ventricle and atrium in size, shape and location.
INTRODUCTION: This paper presents a comprehensive review of acne grading and measurement. Acne is a chronic disorder of the pilosebaceous units, with excess sebum production, follicular epidermal hyperproliferation, inflammation and Propionibacterium acnes activity. Most patients are affected with acne vulgaris, which is the prevalent type of acne. Acne vulgaris consists of comedones (whitehead and blackhead), papules, pustules, nodules and cysts.
OBJECTIVES: To review and identify the issues for acne vulgaris grading and computational assessment methods. To determine the future direction for addressing the identified issues.
METHODS: There are two main methods of assessment for acne severity grading, namely, lesion counting and comparison of patient with a photographic standard. For the computational assessment method, the emphasis is on computational imaging techniques.
RESULTS: Current acne grading methods are very time consuming and tedious. Generally, they rely on approximation for counting lesions and hence the assessment is quite subjective, with both inter and intra-observer variability. It is important to accurately assess acne grade to evaluate its severity as this influences treatment selection and assessment of response to therapy. This will further help in better disease management and more efficacious treatment.
CONCLUSION: Semi-automated or automated methods based on computational imaging techniques should be devised for acne grade assessment.
Diabetic retinopathy has become an increasingly important cause of blindness. Nevertheless, vision loss can be prevented from early detection of diabetic retinopathy and monitor with regular examination. Common automatic detection of retinal abnormalities is for microaneurysms, hemorrhages, hard exudates, and cotton wool spot. However, there is a worse case of retinal abnormality, but not much research was done to detect it. It is neovascularization where new blood vessels grow due to extensive lack of oxygen in the retinal capillaries. This paper shows that various combination of techniques such as image normalization, compactness classifier, morphology-based operator, Gaussian filtering, and thresholding techniques were used in developing of neovascularization detection. A function matrix box was added in order to classify the neovascularization from natural blood vessel. A region-based neovascularization classification was attempted as a diagnostic accuracy. The developed method was tested on images from different database sources with varying quality and image resolution. It shows that specificity and sensitivity results were 89.4% and 63.9%, respectively. The proposed approach yield encouraging results for future development.
This paper introduces an approach to perform segmentation of regions in computed tomography (CT) images that exhibit intra-region intensity variations and at the same time have similar intensity distributions with surrounding/adjacent regions. In this work, we adapt a feature computed from wavelet transform called wavelet energy to represent the region information. The wavelet energy is embedded into a level set model to formulate the segmentation model called wavelet energy-guided level set-based active contour (WELSAC). The WELSAC model is evaluated using several synthetic and CT images focusing on tumour cases, which contain regions demonstrating the characteristics of intra-region intensity variations and having high similarity in intensity distributions with the adjacent regions. The obtained results show that the proposed WELSAC model is able to segment regions of interest in close correspondence with the manual delineation provided by the medical experts and to provide a solution for tumour detection.
Diabetic retinopathy (DR) is a sight threatening complication due to diabetes mellitus that affects the retina. In this article, a computerised DR grading system, which digitally analyses retinal fundus image, is used to measure foveal avascular zone. A v-fold cross-validation method is applied to the FINDeRS database to evaluate the performance of the DR system. It is shown that the system achieved sensitivity of >84%, specificity of >97% and accuracy of >95% for all DR stages. At high values of sensitivity (>95%), specificity (>97%) and accuracy (>98%) obtained for No DR and severe NPDR/PDR stages, the computerised DR grading system is suitable for early detection of DR and for effective treatment of severe cases.
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.
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.
This research investigates the use of charge coupled device (abbreviated as CCD) linear image sensors in an optical tomographic instrumentation system used for sizing particles. The measurement system, consisting of four CCD linear image sensors are configured around an octagonal shaped flow pipe for a four projections system is explained. The four linear image sensors provide 2,048 pixel imaging with a pixel size of 14 micron × 14 micron, hence constituting a high-resolution system. Image reconstruction for a four-projection optical tomography system is also discussed, where a simple optical model is used to relate attenuation due to variations in optical density, [R], within the measurement section. Expressed in matrix form this represents the forward problem in tomography [S] [R] = [M]. In practice, measurements [M] are used to estimate the optical density distribution by solving the inverse problem [R] = [S](-1)[M]. Direct inversion of the sensitivity matrix, [S], is not possible and two approximations are considered and compared-the transpose and the pseudo inverse sensitivity matrices.
The motivation of this paper is to analyse the efficiency and reliability of our proposed algorithm of femur length (FL) measurement for the estimation of gestational age. The automated methods are divided into the following components: threshold, segmentation and extraction. Each component is examined, and improvements are made with the objective of finding the optimal result for FL measurement. The methods are tested with a total of 200 different digitized ultrasound images from our database collection. Overall, the study shows that the watershed-based segmentation method combined with enhanced femur extraction algorithm and a 12 x 12 block averaging seed-point threshold method perform identically well with the expert measurements for every image tested and superior as compared to a previous method.
This paper presents an approach for breast cancer diagnosis in digital mammogram using curvelet transform. After decomposing the mammogram images in curvelet basis, a special set of the biggest coefficients is extracted as feature vector. The Euclidean distance is then used to construct a supervised classifier. The experimental results gave a 98.59% classification accuracy rate, which indicate that curvelet transformation is a promising tool for analysis and classification of digital mammograms.
We present an efficient method for the fusion of medical captured images using different modalities that enhances the original images and combines the complementary information of the various modalities. The contourlet transform has mainly been employed as a fusion technique for images obtained from equal or different modalities. The limitation of directional information of dual-tree complex wavelet (DT-CWT) is rectified in dual-tree complex contourlet transform (DT-CCT) by incorporating directional filter banks (DFB) into the DT-CWT. The DT-CCT produces images with improved contours and textures, while the property of shift invariance is retained. To improve the fused image quality, we propose a new method for fusion rules based on principle component analysis (PCA) which depend on frequency component of DT-CCT coefficients (contourlet domain). For low frequency components, PCA method is adopted and for high frequency components, the salient features are picked up based on local energy. The final fusion image is obtained by directly applying inverse dual tree complex contourlet transform (IDT-CCT) to the fused low and high frequency components. The experimental results showed that the proposed method produces fixed image with extensive features on multimodality.
Skin colour is vital information in dermatological diagnosis as it reflects the pathological condition beneath the skin. It is commonly used to indicate the extent of diseases such as psoriasis, which is indicated by the appearance of red plaques. Although there is no cure for psoriasis, there are many treatment modalities to help control the disease. To evaluate treatment efficacy, the current gold standard method, PASI (Psoriasis Area and Severity Index), is used to determine severity of psoriasis lesion. Erythema (redness) is one parameter in PASI and this condition is assessed visually, thus leading to subjective and inconsistent results. Current methods or instruments that assess erythema have limitations, such as being able to measure erythema well for low pigmented skin (fair skin) but not for highly pigmented skin (dark skin) or vice versa. In this work, we proposed an objective assessment of psoriasis erythema for PASI scoring for different (low to highly pigmented) skin types. The colour of psoriasis lesions are initially obtained by using a chromameter giving the values L*, a*, and b* of CIELAB colour space. The L* value is used to classify skin into three categories: low, medium and highly pigmented skin. The lightness difference (DeltaL*), hue difference (Deltah(ab)), chroma (DeltaC*(ab)) between lesions and the surrounding normal skin are calculated and analysed. It is found that the erythema score of a lesion can be distinguished by their Deltah(ab) value within a particular skin type group. References of lesion with different scores are obtained from the selected lesions by two dermatologists. Results based on 38 lesions from 22 patients with various level of skin pigmentation show that PASI erythema score for different skin types i.e. low (fair skin) to highly pigmented (dark skin) skin types can be determined objectively and consistent with dermatology scoring.
Chromatin morphologies in human breast cancer cells treated with an anti-cancer agent are analyzed at their early stage of programmed cell death or apoptosis. The gray-level images of nuclear chromatin are modelled as random fields. We used two-dimensional isotropic generalized Cauchy field to characterize local self-similarity and global long-range dependence behaviors in the image spatial data. Generalized Cauchy field allows the description of fractal behavior inferred from fractal dimension and the long-range dependence inferred from correlation exponent to be carried out independently. We demonstrated the usefulness of locally self-similar random fields with long-range dependence for modelling chromatin condensation.
Automated computer analysis of magnetic resonance cholangiopancreatography (MRCP) (a focused magnetic resonance imaging sequence for the pancreatobiliary region of the abdomen) images for biliary diseases is a difficult problem because of the large inter- and intrapatient variations in the images, varying acquisition settings, and characteristics of the images, defeating most attempts to produce computer-aided diagnosis systems. This paper proposes a system capable of automated preliminary diagnosis of several diseases affecting the bile ducts in the liver, namely, dilation, stones, tumor, and cyst. The system first identifies the biliary ductal structure present in the MRCP images, and then proceeds to determine the presence or absence of the diseases. Tested on a database of 593 clinical images, the system, which uses visual-based features, has shown to be successful in delivering good performance of 70-90% even in the presence of multiple diseases, and may be useful in aiding medical practitioners in routine MRCP examinations.
Image retrieval at the semantic level mostly depends on image annotation or image classification. Image annotation performance largely depends on three issues: (1) automatic image feature extraction; (2) a semantic image concept modeling; (3) algorithm for semantic image annotation. To address first issue, multilevel features are extracted to construct the feature vector, which represents the contents of the image. To address second issue, domain-dependent concept hierarchy is constructed for interpretation of image semantic concepts. To address third issue, automatic multilevel code generation is proposed for image classification and multilevel image annotation. We make use of the existing image annotation to address second and third issues. Our experiments on a specific domain of X-ray images have given encouraging results.
Information about retinal vasculature morphology is used in grading the severity and progression of diabetic retinopathy. An image analysis system can help ophthalmologists make accurate and efficient diagnoses. This paper presents the development of an image processing algorithm for detecting and reconstructing retinal vasculature. The detection of the vascular structure is achieved by image enhancement using contrast limited adaptive histogram equalization followed by the extraction of the vessels using bottom-hat morphological transformation. For reconstruction of the complete retinal vasculature, a region growing technique based on first-order Gaussian derivative is developed. The technique incorporates both gradient magnitude change and average intensity as the homogeneity criteria that enable the process to adapt to intensity changes and intensity spread over the vasculature region. The reconstruction technique reduces the required number of seeds to near optimal for the region growing process. It also overcomes poor performance of current seed-based methods, especially with low and inconsistent contrast images as normally seen in vasculature regions of fundus images. Simulations of the algorithm on 20 test images from the DRIVE database show that it outperforms many other published methods and achieved an accuracy range (ability to detect both vessel and non-vessel pixels) of 0.91 - 0.95, a sensitivity range (ability to detect vessel pixels) of 0.91 - 0.95 and a specificity range (ability to detect non-vessel pixels) of 0.88 - 0.94.
The data distribution system of this project is divided into two types, which are a Two-PC Image Reconstruction System and a Two-PC Velocity Measurement System. Each data distribution system is investigated to see whether the results' refreshing rate of the corresponding measurement can be greater than the rate obtained by using a single computer in the same measurement system for each application. Each system has its own flow control protocol for controlling how data is distributed within the system in order to speed up the data processing time. This can be done if two PCs work in parallel. The challenge of this project is to define the data flow process and critical timing during data packaging, transferring and extracting in between PCs. If a single computer is used as a data processing unit, a longer time is needed to produce a measurement result. This insufficient real-time result will cause problems in a feedback control process when applying the system in industrial plants. To increase the refreshing rate of the measurement result, an investigation on a data distribution system is performed to replace the existing data processing unit.