MyMedR

Displaying publications 1 - 20 of 113 in total

Abstract:

Sort:

Speckle Noise Diffusion in Knee Articular Cartilage Ultrasound Images

Shoaib MA, Hossain MB, Hum YC, Chuah JH, Mohd Salim MI, Lai KW

Curr Med Imaging, 2020;16(6):739-751.
PMID: 32723246 DOI: 10.2174/1573405615666190903143330

BACKGROUND: Ultrasound (US) imaging can be a convenient and reliable substitute for magnetic resonance imaging in the investigation or screening of articular cartilage injury. However, US images suffer from two main impediments, i.e., low contrast ratio and presence of speckle noise.
AIMS: A variation of anisotropic diffusion is proposed that can reduce speckle noise without compromising the image quality of the edges and other important details.
METHODS: For this technique, four gradient thresholds were adopted instead of one. A new diffusivity function that preserves the edge of the resultant image is also proposed. To automatically terminate the iterative procedures, the Mean Absolute Error as its stopping criterion was implemented.
RESULTS: Numerical results obtained by simulations unanimously indicate that the proposed method outperforms conventional speckle reduction techniques. Nevertheless, this preliminary study has been conducted based on a small number of asymptomatic subjects.
CONCLUSION: Future work must investigate the feasibility of this method in a large cohort and its clinical validity through testing subjects with a symptomatic cartilage injury.

Matched MeSH terms: Image Interpretation, Computer-Assisted/methods*
Automated approach for detection of ischemic stroke using Delaunay Triangulation in brain MRI images

Subudhi A, Acharya UR, Dash M, Jena S, Sabut S

Comput Biol Med, 2018 12 01;103:116-129.
PMID: 30359807 DOI: 10.1016/j.compbiomed.2018.10.016

It is difficult to develop an accurate algorithm to detect the stroke lesions using magnetic resonance imaging (MRI) images due to variation in different lesion sizes, variation in morphological structure, and similarity in intensity of lesion with normal brain in three types of stroke, namely partial anterior circulation syndrome (PACS), lacunar syndrome (LACS) and total anterior circulation stroke (TACS). In this paper, we have integrated the advantages of Delaunay triangulation (DT) and fractional order Darwinian particle swarm optimization (FODPSO), called DT-FODPSO technique for automatic segmentation of the structure of the stroke lesion. The approach was validated on 192 MRI images obtained from different stroke subjects. Statistical and morphological features were extracted and classified according to the Oxfordshire community stroke project (OCSP) using support vector machine (SVM) and random forest (RF) classifiers. The method effectively detected the stroke lesions and achieved promising results with an average sensitivity of 0.93, accuracy of 0.95, JI of 0.89 and Dice similarity index of 0.93 using RF classifier. These promising results indicates the DT based optimized approach is efficient in detecting ischemic stroke and it can aid the neuro-radiologists to validate their routine screening.

Matched MeSH terms: Image Interpretation, Computer-Assisted/methods*
Automated retinal health diagnosis using pyramid histogram of visual words and Fisher vector techniques

Koh JEW, Ng EYK, Bhandary SV, Hagiwara Y, Laude A, Acharya UR

Comput Biol Med, 2018 01 01;92:204-209.
PMID: 29227822 DOI: 10.1016/j.compbiomed.2017.11.019

Untreated age-related macular degeneration (AMD), diabetic retinopathy (DR), and glaucoma may lead to irreversible vision loss. Hence, it is essential to have regular eye screening to detect these eye diseases at an early stage and to offer treatment where appropriate. One of the simplest, non-invasive and cost-effective techniques to screen the eyes is by using fundus photo imaging. But, the manual evaluation of fundus images is tedious and challenging. Further, the diagnosis made by ophthalmologists may be subjective. Therefore, an objective and novel algorithm using the pyramid histogram of visual words (PHOW) and Fisher vectors is proposed for the classification of fundus images into their respective eye conditions (normal, AMD, DR, and glaucoma). The proposed algorithm extracts features which are represented as words. These features are built and encoded into a Fisher vector for classification using random forest classifier. This proposed algorithm is validated with both blindfold and ten-fold cross-validation techniques. An accuracy of 90.06% is achieved with the blindfold method, and highest accuracy of 96.79% is obtained with ten-fold cross-validation. The highest classification performance of our system shows the potential of deploying it in polyclinics to assist healthcare professionals in their initial diagnosis of the eye. Our developed system can reduce the workload of ophthalmologists significantly.

Matched MeSH terms: Image Interpretation, Computer-Assisted/methods*
CT Reconstruction Algorithm and Low Contrast Detectability of Phantom Study: A Systematic Review and Meta-Analysis

Adibah Yusof NA, Abdul Karim MK, Asikin NM, Paiman S, Awang Kechik MM, Abdul Rahman MA, et al.

Curr Med Imaging, 2023;19(10):1105-1113.
PMID: 35975862 DOI: 10.2174/1573405618666220816160544

BACKGROUND: For almost three decades, computed tomography (CT) has been extensively used in medical diagnosis, which led researchers to conduct linking of CT dose exposure with image quality.
METHODS: In this study, a systematic review and a meta-analysis study were conducted on CT phantom for resolution study especially based on the low contrast detectability (LCD). Furthermore, the association between the CT parameter such as tube voltage and the type of reconstruction algorithm, the amount of phantom scanning affecting the image quality and the exposure dose were also investigated in this study. We utilize PubMed, ScienceDirect, Google Scholar and Scopus databases to search related published articles from the year 2011 until 2020. The notable keywords comprise "computed tomography", "CT phantom", and "low contrast detectability". Of 52 articles, 20 articles are within the inclusion criteria in this systematic review.
RESULTS: The dichotomous outcomes were chosen to represent the results in terms of risk ratio as per meta-analysis study. Notably, the noise in iterative reconstruction (IR) reduced by 24%, 33% and 36% with the use of smooth, medium and sharp filters, respectively. Furthermore, adaptive iterative dose reduction (AIDR 3D) improved image quality and the visibility of smaller less dense objects compared to filtered back-projection. Most of the researchers used 120 kVp tube voltage to scan phantom for quality assurance study.
CONCLUSION: Hence, optimizing primary factors such as tube potential reduces the dose exposure significantly, and the optimized IR technique could substantially reduce the radiation dose while maintaining the image quality.

Matched MeSH terms: Radiographic Image Interpretation, Computer-Assisted/methods
ExHiF: Alzheimer's disease detection using exemplar histogram-based features with CT and MR images

Kaplan E, Baygin M, Barua PD, Dogan S, Tuncer T, Altunisik E, et al.

Med Eng Phys, 2023 May;115:103971.
PMID: 37120169 DOI: 10.1016/j.medengphy.2023.103971

PURPOSE: The classification of medical images is an important priority for clinical research and helps to improve the diagnosis of various disorders. This work aims to classify the neuroradiological features of patients with Alzheimer's disease (AD) using an automatic hand-modeled method with high accuracy.
MATERIALS AND METHOD: This work uses two (private and public) datasets. The private dataset consists of 3807 magnetic resonance imaging (MRI) and computer tomography (CT) images belonging to two (normal and AD) classes. The second public (Kaggle AD) dataset contains 6400 MR images. The presented classification model comprises three fundamental phases: feature extraction using an exemplar hybrid feature extractor, neighborhood component analysis-based feature selection, and classification utilizing eight different classifiers. The novelty of this model is feature extraction. Vision transformers inspire this phase, and hence 16 exemplars are generated. Histogram-oriented gradients (HOG), local binary pattern (LBP) and local phase quantization (LPQ) feature extraction functions have been applied to each exemplar/patch and raw brain image. Finally, the created features are merged, and the best features are selected using neighborhood component analysis (NCA). These features are fed to eight classifiers to obtain highest classification performance using our proposed method. The presented image classification model uses exemplar histogram-based features; hence, it is called ExHiF.
RESULTS: We have developed the ExHiF model with a ten-fold cross-validation strategy using two (private and public) datasets with shallow classifiers. We have obtained 100% classification accuracy using cubic support vector machine (CSVM) and fine k nearest neighbor (FkNN) classifiers for both datasets.
CONCLUSIONS: Our developed model is ready to be validated with more datasets and has the potential to be employed in mental hospitals to assist neurologists in confirming their manual screening of AD using MRI/CT images.

Matched MeSH terms: Image Interpretation, Computer-Assisted/methods
Fulltext Cervical Cancer Classification From Pap Smear Images Using Deep Convolutional Neural Network Models

Tan SL, Selvachandran G, Ding W, Paramesran R, Kotecha K

Interdiscip Sci, 2024 Mar;16(1):16-38.
PMID: 37962777 DOI: 10.1007/s12539-023-00589-5

As one of the most common female cancers, cervical cancer often develops years after a prolonged and reversible pre-cancerous stage. Traditional classification algorithms used for detection of cervical cancer often require cell segmentation and feature extraction techniques, while convolutional neural network (CNN) models demand a large dataset to mitigate over-fitting and poor generalization problems. To this end, this study aims to develop deep learning models for automated cervical cancer detection that do not rely on segmentation methods or custom features. Due to limited data availability, transfer learning was employed with pre-trained CNN models to directly operate on Pap smear images for a seven-class classification task. Thorough evaluation and comparison of 13 pre-trained deep CNN models were performed using the publicly available Herlev dataset and the Keras package in Google Collaboratory. In terms of accuracy and performance, DenseNet-201 is the best-performing model. The pre-trained CNN models studied in this paper produced good experimental results and required little computing time.

Matched MeSH terms: Image Interpretation, Computer-Assisted/methods
Combined Spline and B-spline for an improved automatic skin lesion segmentation in dermoscopic images using optimal color channel

Abbas AA, Guo X, Tan WH, Jalab HA

J Med Syst, 2014 Aug;38(8):80.
PMID: 24957396 DOI: 10.1007/s10916-014-0080-7

In a computerized image analysis environment, the irregularity of a lesion border has been used to differentiate between malignant melanoma and other pigmented skin lesions. The accuracy of the automated lesion border detection is a significant step towards accurate classification at a later stage. In this paper, we propose the use of a combined Spline and B-spline in order to enhance the quality of dermoscopic images before segmentation. In this paper, morphological operations and median filter were used first to remove noise from the original image during pre-processing. Then we proceeded to adjust image RGB values to the optimal color channel (green channel). The combined Spline and B-spline method was subsequently adopted to enhance the image before segmentation. The lesion segmentation was completed based on threshold value empirically obtained using the optimal color channel. Finally, morphological operations were utilized to merge the smaller regions with the main lesion region. Improvement on the average segmentation accuracy was observed in the experimental results conducted on 70 dermoscopic images. The average accuracy of segmentation achieved in this paper was 97.21 % (where, the average sensitivity and specificity were 94 % and 98.05 % respectively).

Matched MeSH terms: Image Interpretation, Computer-Assisted/methods*
Fulltext A deformable generic 3D model of haptoral anchor of Monogenean

Teo BG, Dhillon SK, Lim LH

PLoS One, 2013;8(10):e77650.
PMID: 24204903 DOI: 10.1371/journal.pone.0077650

In this paper, a digital 3D model which allows for visualisation in three dimensions and interactive manipulation is explored as a tool to help us understand the structural morphology and elucidate the functions of morphological structures of fragile microorganisms which defy live studies. We developed a deformable generic 3D model of haptoral anchor of dactylogyridean monogeneans that can subsequently be deformed into different desired anchor shapes by using direct manipulation deformation technique. We used point primitives to construct the rectangular building blocks to develop our deformable 3D model. Point primitives are manually marked on a 2D illustration of an anchor on a Cartesian graph paper and a set of Cartesian coordinates for each point primitive is manually extracted from the graph paper. A Python script is then written in Blender to construct 3D rectangular building blocks based on the Cartesian coordinates. The rectangular building blocks are stacked on top or by the side of each other following their respective Cartesian coordinates of point primitive. More point primitives are added at the sites in the 3D model where more structural variations are likely to occur, in order to generate complex anchor structures. We used Catmull-Clark subdivision surface modifier to smoothen the surface and edge of the generic 3D model to obtain a smoother and more natural 3D shape and antialiasing option to reduce the jagged edges of the 3D model. This deformable generic 3D model can be deformed into different desired 3D anchor shapes through direct manipulation deformation technique by aligning the vertices (pilot points) of the newly developed deformable generic 3D model onto the 2D illustrations of the desired shapes and moving the vertices until the desire 3D shapes are formed. In this generic 3D model all the vertices present are deployed for displacement during deformation.

Matched MeSH terms: Image Interpretation, Computer-Assisted/methods*
Concentration measurements of bubbles in a water column using an optical tomography system

Ibrahim S, Yunus MA, Green RG, Dutton K

ISA Trans, 2012 Nov;51(6):821-6.
PMID: 22624831 DOI: 10.1016/j.isatra.2012.04.010

Optical tomography provides a means for the determination of the spatial distribution of materials with different optical density in a volume by non-intrusive means. This paper presents results of concentration measurements of gas bubbles in a water column using an optical tomography system. A hydraulic flow rig is used to generate vertical air-water two-phase flows with controllable bubble flow rate. Two approaches are investigated. The first aims to obtain an average gas concentration at the measurement section, the second aims to obtain a gas distribution profile by using tomographic imaging. A hybrid back-projection algorithm is used to calculate concentration profiles from measured sensor values to provide a tomographic image of the measurement cross-section. The algorithm combines the characteristic of an optical sensor as a hard field sensor and the linear back projection algorithm.

Matched MeSH terms: Image Interpretation, Computer-Assisted/methods*
On the use of collinear and triangle equation for automatic segmentation and boundary detection of cardiac cavity images

Sigit R, Mustafa MM, Hussain A, Maskon O, Nor IF

Adv Exp Med Biol, 2011;696:481-8.
PMID: 21431588 DOI: 10.1007/978-1-4419-7046-6_48

In this chapter, the computational biology of cardiac cavity images is proposed. The method uses collinear and triangle equation algorithms to detect and reconstruct the boundary of the cardiac cavity. The first step involves high boost filter to enhance the high frequency component without affecting the low frequency component. Second, the morphological and thresholding operators are applied to the image to eliminate noise and convert the image into a binary image. Next, the edge detection is performed using the negative Laplacian filter and followed by region filtering. Finally, the collinear and triangle equations are used to detect and reconstruct the more precise cavity boundary. Results obtained have proved that this technique is able to perform better segmentation and detection of the boundary of cardiac cavity from echocardiographic images.

Matched MeSH terms: Image Interpretation, Computer-Assisted/methods
An image processing application for the localization and segmentation of lymphoblast cell using peripheral blood images

Madhloom HT, Kareem SA, Ariffin H

J Med Syst, 2012 Aug;36(4):2149-58.
PMID: 21399912 DOI: 10.1007/s10916-011-9679-0

An important preliminary step in the diagnosis of leukemia is the visual examination of the patient's peripheral blood smear under the microscope. Morphological changes in the white blood cells can be an indicator of the nature and severity of the disease. Manual techniques are labor intensive, slow, error prone and costly. A computerized system can be used as a supportive tool for the specialist in order to enhance and accelerate the morphological analysis process. This research present a new method that integrates color features with the morphological reconstruction to localize and isolate lymphoblast cells from a microscope image that contains many cells. The localization and segmentation are conducted using a proposed method that consists of an integration of several digital image processing techniques. 180 microscopic blood images were tested, and the proposed framework managed to obtain 100% accuracy for the localization of the lymphoblast cells and separate it from the image scene. The results obtained indicate that the proposed method can be safely used for the purpose of lymphoblast cells localization and segmentation and subsequently, aiding the diagnosis of leukemia.

Matched MeSH terms: Image Interpretation, Computer-Assisted/methods*
Vascular intersection detection in retina fundus images using a new hybrid approach

Aibinu AM, Iqbal MI, Shafie AA, Salami MJ, Nilsson M

Comput Biol Med, 2010 Jan;40(1):81-9.
PMID: 20022595 DOI: 10.1016/j.compbiomed.2009.11.004

The use of vascular intersection aberration as one of the signs when monitoring and diagnosing diabetic retinopathy from retina fundus images (FIs) has been widely reported in the literature. In this paper, a new hybrid approach called the combined cross-point number (CCN) method able to detect the vascular bifurcation and intersection points in FIs is proposed. The CCN method makes use of two vascular intersection detection techniques, namely the modified cross-point number (MCN) method and the simple cross-point number (SCN) method. Our proposed approach was tested on images obtained from two different and publicly available fundus image databases. The results show a very high precision, accuracy, sensitivity and low false rate in detecting both bifurcation and crossover points compared with both the MCN and the SCN methods.

Matched MeSH terms: Image Interpretation, Computer-Assisted/methods*
A comparison of wavelet and curvelet for breast cancer diagnosis in digital mammogram

Meselhy Eltoukhy M, Faye I, Belhaouari Samir B

Comput Biol Med, 2010 Apr;40(4):384-91.
PMID: 20163793 DOI: 10.1016/j.compbiomed.2010.02.002

This paper presents a comparative study between wavelet and curvelet transform for breast cancer diagnosis in digital mammogram. Using multiresolution analysis, mammogram images are decomposed into different resolution levels, which are sensitive to different frequency bands. A set of the biggest coefficients from each decomposition level is extracted. Then a supervised classifier system based on Euclidian distance is constructed. The performance of the classifier is evaluated using a 2 x 5-fold cross validation followed by a statistical analysis. The experimental results suggest that curvelet transform outperforms wavelet transform and the difference is statistically significant.

Matched MeSH terms: Image Interpretation, Computer-Assisted/methods*
Area assessment of psoriasis lesions for PASI scoring

Fadzil MH, Ihtatho D, Affandi AM, Hussein SH

J Med Eng Technol, 2009;33(6):426-36.
PMID: 19557605 DOI: 10.1080/07434610902744066

Psoriasis is a skin disorder which is caused by a genetic fault. 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 measure psoriasis severity by evaluating the area, erythema, scaliness and thickness of the plaques. However, the determination of PASI can be tedious and subjective. In this work, we develop a computer vision method that determines one of the PASI parameters, the lesion area. The method isolates healthy and healed skin areas from lesion areas by analysing the hue and chroma information in the CIE L*a*b* colour space. Centroids of healthy skin and psoriasis in the hue-chroma space are determined from selected sample. The Euclidean distance of all pixels from each centroid is calculated. Pixels are assigned to either healthy skin or psorasis lesion classes based on the minimum Euclidean distance. The study involves patients from different ethnic origins having three different skin tones. Results obtained show that the proposed method is able to determine lesion areas with accuracy higher than 90% for 28 out of 30 cases.

Study site: Dermatology Clinic, Hospital Kuala Lumpur

Matched MeSH terms: Image Interpretation, Computer-Assisted/methods*
Liver isolation in abdominal MRI

Rajasvaran L, Haw TW, Sarker SZ

J Med Syst, 2008 Aug;32(4):259-68.
PMID: 18619090

This work presents a method for liver isolation in magnetic resonance imaging (MRI) abdomen images. It is based on a priori statistical information about the shape of the liver obtained from a training set using the segmentation approach. Morphological watershed algorithm is used as a key technique as it is a simple and intuitive method, producing a complete division of the image in separated regions even if the contrast is poor, and it is fast, with possibility for parallel implementation. To overcome the over-segmentation problem of the watershed process, image preprocessing and postprocessing are applied. Morphological smoothing, Gaussian smoothing, intensity thresholding, gradient computation and gradient thresholding are proposed for preprocessing with morphological and graph based region adjacent list constructed for region merging. A new integrated region similarity function is also defined for region merging control. The proposed method produces good isolation of liver in axial MRI images of the abdomen, as is shown in this paper.

Matched MeSH terms: Image Interpretation, Computer-Assisted/methods*
Time-domain reconstruction using sensitivity coefficients for limited view ultrawide band tomography

Abdullah MZ, Yin W, Bilal M, Armitage DW, Mackin R, Peyton AJ

Rev Sci Instrum, 2007 Aug;78(8):084703.
PMID: 17764343

This article addresses time-domain ultrawide band (UWB) electromagnetic tomography for reconstructing the unknown spatial characteristic of an object from observations of the arrivals of short electromagnetic (EM) pulses. Here, the determination of the first peak arrival of the EM traces constitutes the forward problem, and the inverse problem aims to reconstruct the EM property distribution of the media. In this article, the finite-difference time-domain method implementing a perfectly matched layer is used to solve the forward problem from which the system sensitivity maps are determined. Image reconstruction is based on the combination of a linearized update and regularized Landweber minimization algorithm. Experimental data from a laboratory UWB system using targets of different contrasts, sizes, and shapes in an aqueous media are presented. The results show that this technique can accurately detect and locate unknown targets in spite of the presence of significant levels of noise in the data.

Matched MeSH terms: Image Interpretation, Computer-Assisted/methods*
Deblurring Computed Tomography Medical Images Using a Novel Amended Landweber Algorithm

Al-Ameen Z, Sulong G

Interdiscip Sci, 2015 Sep;7(3):319-25.
PMID: 26199211 DOI: 10.1007/s12539-015-0022-1

In computed tomography (CT), blurring occurs due to different hardware or software errors and hides certain medical details that are present in an image. Image blur is difficult to avoid in many circumstances and can frequently ruin an image. For this, many methods have been developed to reduce the blurring artifact from CT images. The problems with these methods are the high implementation time, noise amplification and boundary artifacts. Hence, this article presents an amended version of the iterative Landweber algorithm to attain artifact-free boundaries and less noise amplification in a faster application time. In this study, both synthetic and real blurred CT images are used to validate the proposed method properly. Similarly, the quality of the processed synthetic images is measured using the feature similarity index, structural similarity and visual information fidelity in pixel domain metrics. Finally, the results obtained from intensive experiments and performance evaluations show the efficiency of the proposed algorithm, which has potential as a new approach in medical image processing.

Matched MeSH terms: Radiographic Image Interpretation, Computer-Assisted/methods*
Fulltext Invariant domain watermarking using heaviside function of order alpha and fractional Gaussian field

Abbasi A, Woo CS, Ibrahim RW, Islam S

PLoS One, 2015;10(4):e0123427.
PMID: 25884854 DOI: 10.1371/journal.pone.0123427

Digital image watermarking is an important technique for the authentication of multimedia content and copyright protection. Conventional digital image watermarking techniques are often vulnerable to geometric distortions such as Rotation, Scaling, and Translation (RST). These distortions desynchronize the watermark information embedded in an image and thus disable watermark detection. To solve this problem, we propose an RST invariant domain watermarking technique based on fractional calculus. We have constructed a domain using Heaviside function of order alpha (HFOA). The HFOA models the signal as a polynomial for watermark embedding. The watermark is embedded in all the coefficients of the image. We have also constructed a fractional variance formula using fractional Gaussian field. A cross correlation method based on the fractional Gaussian field is used for watermark detection. Furthermore the proposed method enables blind watermark detection where the original image is not required during the watermark detection thereby making it more practical than non-blind watermarking techniques. Experimental results confirmed that the proposed technique has a high level of robustness.

Matched MeSH terms: Image Interpretation, Computer-Assisted/methods*
Automated diabetic macular edema (DME) grading system using DWT, DCT Features and maculopathy index

Acharya UR, Mookiah MRK, Koh JEW, Tan JH, Bhandary SV, Rao AK, et al.

Comput Biol Med, 2017 05 01;84:59-68.
PMID: 28343061 DOI: 10.1016/j.compbiomed.2017.03.016

The cause of diabetic macular edema (DME) is due to prolonged and uncontrolled diabetes mellitus (DM) which affects the vision of diabetic subjects. DME is graded based on the exudate location from the macula. It is clinically diagnosed using fundus images which is tedious and time-consuming. Regular eye screening and subsequent treatment may prevent the vision loss. Hence, in this work, a hybrid system based on Radon transform (RT), discrete wavelet transform (DWT) and discrete cosine transform (DCT) are proposed for an automated detection of DME. The fundus images are subjected to RT to obtain sinograms and DWT is applied on these sinograms to extract wavelet coefficients (approximate, horizontal, vertical and diagonal). DCT is applied on approximate coefficients to obtain 2D-DCT coefficients. Further, these coefficients are converted into 1D vector by arranging the coefficients in zig-zag manner. This 1D signal is subjected to locality sensitive discriminant analysis (LSDA). Finally, various supervised classifiers are used to classify the three classes using significant features. Our proposed technique yielded a classification accuracy of 100% and 97.01% using two and seven significant features for private and public (MESSIDOR) databases respectively. Also, a maculopathy index is formulated with two significant parameters to discriminate the three groups distinctly using a single integer. Hence, our obtained results suggest that this system can be used as an eye screening tool for diabetic subjects for DME.

Matched MeSH terms: Image Interpretation, Computer-Assisted/methods*
A novel algorithm to detect glaucoma risk using texton and local configuration pattern features extracted from fundus images

Acharya UR, Bhat S, Koh JEW, Bhandary SV, Adeli H

Comput Biol Med, 2017 Sep 01;88:72-83.
PMID: 28700902 DOI: 10.1016/j.compbiomed.2017.06.022

Glaucoma is an optic neuropathy defined by characteristic damage to the optic nerve and accompanying visual field deficits. Early diagnosis and treatment are critical to prevent irreversible vision loss and ultimate blindness. Current techniques for computer-aided analysis of the optic nerve and retinal nerve fiber layer (RNFL) are expensive and require keen interpretation by trained specialists. Hence, an automated system is highly desirable for a cost-effective and accurate screening for the diagnosis of glaucoma. This paper presents a new methodology and a computerized diagnostic system. Adaptive histogram equalization is used to convert color images to grayscale images followed by convolution of these images with Leung-Malik (LM), Schmid (S), and maximum response (MR4 and MR8) filter banks. The basic microstructures in typical images are called textons. The convolution process produces textons. Local configuration pattern (LCP) features are extracted from these textons. The significant features are selected using a sequential floating forward search (SFFS) method and ranked using the statistical t-test. Finally, various classifiers are used for classification of images into normal and glaucomatous classes. A high classification accuracy of 95.8% is achieved using six features obtained from the LM filter bank and the k-nearest neighbor (kNN) classifier. A glaucoma integrative index (GRI) is also formulated to obtain a reliable and effective system.

Matched MeSH terms: Image Interpretation, Computer-Assisted/methods*

Filters

Please provide feedback to Administrator (afdal@afpm.org.my)

External Links