Genetic algorithm based feature selection combined with dual classification for the automated detection of proliferative diabetic retinopathy.

Proliferative diabetic retinopathy (PDR) is a condition that carries a high risk of severe visual impairment. The hallmark of PDR is the growth of abnormal new vessels. In this paper, an automated method for the detection of new vessels from retinal images is presented. This method is based on a dual classification approach. Two vessel segmentation approaches are applied to create two separate binary vessel map which each hold vital information. Local morphology features are measured from each binary vessel map to produce two separate 4-D feature vectors. Independent classification is performed for each feature vector using a support vector machine (SVM) classifier. The system then combines these individual outcomes to produce a final decision. This is followed by the creation of additional features to generate 21-D feature vectors, which feed into a genetic algorithm based feature selection approach with the objective of finding feature subsets that improve the performance of the classification. Sensitivity and specificity results using a dataset of 60 images are 0.9138 and 0.9600, respectively, on a per patch basis and 1.000 and 0.975, respectively, on a per image basis.

Automated detection of proliferative diabetic retinopathy using a modified line operator and dual classification.

Proliferative diabetic retinopathy (PDR) is a condition that carries a high risk of severe visual impairment. The hallmark of PDR is neovascularisation, the growth of abnormal new vessels. This paper describes an automated method for the detection of new vessels in retinal images. Two vessel segmentation approaches are applied, using the standard line operator and a novel modified line operator. The latter is designed to reduce false responses to non-vessel edges. Both generated binary vessel maps hold vital information which must be processed separately. This is achieved with a dual classification system. Local morphology features are measured from each binary vessel map to produce two separate feature sets. Independent classification is performed for each feature set using a support vector machine (SVM) classifier. The system then combines these individual classification outcomes to produce a final decision. Sensitivity and specificity results using a dataset of 60 images are 0.862 and 0.944 respectively on a per patch basis and 1.00 and 0.90 respectively on a per image basis.

Segmentation of pulmonary nodules in thoracic CT scans: a region growing approach.

This paper presents an efficient algorithm for segmenting different types of pulmonary nodules including high and low contrast nodules, nodules with vasculature attachment, and nodules in the close vicinity of the lung wall or diaphragm. The algorithm performs an adaptive sphericity oriented contrast region growing on the fuzzy connectivity map of the object of interest. This region growing is operated within a volumetric mask which is created by first applying a local adaptive segmentation algorithm that identifies foreground and background regions within a certain window size. The foreground objects are then filled to remove any holes, and a spatial connectivity map is generated to create a 3-D mask. The mask is then enlarged to contain the background while excluding unwanted foreground regions. Apart from generating a confined search volume, the mask is also used to estimate the parameters for the subsequent region growing, as well as for repositioning the seed point in order to ensure reproducibility. The method was run on 815 pulmonary nodules. By using randomly placed seed points, the approach was shown to be fully reproducible. As for acceptability, the segmentation results were visually inspected by a qualified radiologist to search for any gross miss-segmentation. 84% of the first results of the segmentation were accepted by the radiologist while for the remaining 16% nodules, alternative segmentation solutions that were provided by the method were selected.

Computer-assisted detection for CT colonography: external validation.

AIM: To externally validate a computer-assisted detection (CAD) system for computed tomography (CT) colonography, using data from a single centre uninvolved with the software development. MATERIALS AND METHODS: Twenty-five multi-detector CT colonography examinations of patients with validated polyps accumulated at a single centre were examined by two readers who used endoscopic and histopathological data to identify polyp coordinates. A CAD system that had been developed using data from elsewhere, and had not previously encountered the present data, was then applied to the data at sphericity filter settings of 0.75 and 0.50 and identified potential polyps. True-positive, false-negative, and false-positive counts were determined by comparison with the known polyp coordinates. RESULTS: Twenty-five patients had 57 polyps, median size 6mm (range 1-15mm). Per-patient sensitivity for the CAD system was 96% (24 of 25). The CAD system detected 44 (77%) polyps at sphericity setting 0.75 and 49 (86%) polyps at sphericity 0.50: the additional five polyps detected all measured 5mm or less. Sphericity of 0.75 resulted in a median of 10 (one to 34) easily dismissed false-positive prompts per patient and a median of 4 (zero to 15) that needed three-dimensional rendering before dismissal. This rose to 32 (16 to 99) and 11 (three to 35), respectively, at sphericity 0.5. CONCLUSIONS: A per-patient sensitivity of 96% was found for the CAD system (in patients with a median polyp diameter of 6mm) using external validation, a more stringent test than either internal cross-validation or temporal validation. Decreasing sphericity increases sensitivity for small polyps at the expense of decreased specificity.

Disability in multiple sclerosis is related to normal appearing brain tissue MTR histogram abnormalities.

BACKGROUND: Magnetization transfer ratio (MTR) histogram analysis provides a global measure of disease burden in multiple sclerosis (MS). MTR abnormalities in normal appearing brain tissue (NABT) provide quantitative information on the extent of tissue damage undetected by conventional T2-weighted (T2W) magnetic resonance imaging (MRI). AIMS: 1) To compare the MTR histograms from NABT across a broad spectrum of relapse onset MS patients, including relapsing-remitting (RR) MS (including newly diagnosed and benign subgroups) and secondary progressive (SP) MS. 2) To determine the relationship between clinical disability and NABT MTR histograms. METHODS: 2D spin echo magnetization transfer imaging was performed on 70 RRMS and 25 SPMS patients and compared with 63 controls. MTR histograms were acquired for NABT after extracting lesions and cerebrospinal fluid (CSF). T2W images were used to measure the brain parenchymal fraction (BPF) and T2 lesion load. RESULTS: MS patients had a disease duration ranging from 0.5 to 37 years and an Expanded Disability Status Scale (EDSS) score ranging from 0 to 8.5. There was a significant decrease in NABT mean MTR (+/- standard deviation) compared with controls (33.07 pu +/- 1.06 versus 34.26 pu +/- 0.47; P < 0.001) with an effect size of 2.56. The reduction in NABT mean MTR varied among patient groups from 4.9% for SPMS, 3% for all RRMS, 2.7% for early RRMS and 2.5% for benign MS, compared with controls. NABT mean MTR correlated significantly with T2 lesion load (r = -0.82) and BPF (r = 0.58). EDSS score correlated with NABT mean MTR (r = -0.43), BPF (r = -0.33) and with T2 lesion load (r = 0.59). Multivariate analysis using NABT MTR peak height, T2 lesion load and BPF combined only accounted for 38% of the variance in the EDSS (r = 0.62; P < 0.001). Disease duration accounted for an additional 14% of variance in the EDSS (r = 0.72; P < 0.001). CONCLUSIONS: There is evidence of diffuse abnormalities in NABT in addition to global brain atrophy in relapse onset MS patients, including those with recently diagnosed RRMS and benign MS. The abnormalities are greatest in patients with the more disabling SPMS. Atrophy, NABT and lesion abnormalities are all partly correlated; the processes marked by these MR measures all contribute to disability in MS, providing complementary information relevant to the complex pathological processes that occur in MS.

The normal appearing grey matter in primary progressive multiple sclerosis: a magnetisation transfer imaging study.

BACKGROUND: In 10-15 % of patients with multiple sclerosis (MS), the clinical course is characterized by slow progression in disability without relapses (primary progressive (PP) MS). The mechanism of disability in this form of MS is poorly understood. Using magnetization transfer ratio (MTR) imaging, we investigated normal appearing white matter (NAWM) and normal appearing grey matter (NAGM) in PPMS and explored the relationship of MTR measures with disability. METHODS: Thirty patients with PPMS and 30 age matched controls had spin echo based MTR imaging to study lesions and normal appearing tissues. The brain was segmented into NAWM and NAGM using SPM99 with lesions segmented using a semiautomated local thresholding technique. A 75% probability threshold for classification of NAWM and NAGM was used to diminish partial volume effects. From normalized histograms of MTR intensity values, six MTR parameters were measured. Mean lesion MTR and T2 lesion volume were also measured. Disability was assessed using Kurtzke's expanded disability status scale (EDSS). RESULTS: Compared with controls, patients exhibited a significant reduction in mean NAWM (p = 0.001) and NAGM (p = 0.004) MTR. Spearman's rank correlation of EDSS with the six MTR parameters in NAWM and NAGM, mean lesion MTR, and T2 lesion volume, was only significant with mean NAGM MTR (r = -0.41, p = 0.02), the 25th percentile of NAGM MTR intensity (r = -0.37, p = 0.05), and T2 lesion volume (r = 0.39, p = 0.04). Multiple regression analysis of the relationship between EDSS and 4 MR parameters representing each tissue type (mean NAWM MTR, mean NAGM MTR, mean lesion MTR, T2 lesion volume) showed that the association of EDSS with mean NAGM MTR remained significant. CONCLUSIONS: There appear to be significant abnormalities in the NAGM in PP MS. Further investigation of the pathological basis and functional significance of grey matter abnormality in PPMS is warranted.

T1 histograms of normal-appearing brain tissue are abnormal in early relapsing-remitting multiple sclerosis.

OBJECTIVE: To use both whole-brain and normal-appearing brain tissue (NABT) T1 relaxation time histograms to investigate abnormalities in early relapsing-remitting (RR) multiple sclerosis (MS). BACKGROUND: In patients with established MS, both lesions and NABT exhibit an increase in T1 relaxation time. By using T1 histogram analysis, it is hoped that such changes in early disease can be detected. METHOD: Twenty-seven patients and 14 age- and sex-matched controls underwent magnetic resonance imaging (MRI) of the brain, which included the following sequences: 1) proton density (PD)- and T2-weighted fast spin echo (FSE) to measure T2 lesion load, 2) PD- and T1-weighted gradient echos from which T1 relaxation was calculated, and 3) T1-weighted SE imaging pre- and post-triple dose (0.3 mmol/kg) gadolinium (Gd-DTPA) to measure T1 hypointense and gadolinium-enhancing lesion loads, respectively. All patients had RR MS with disease duration <3 years (median 1.7 years). Statistical parametric mapping (SPM) 99 was used to segment brain from cerebrospinal fluid (CSF), and lesions were segmented using a local thresholding technique. RESULTS: Both whole-brain and NABT histograms were abnormal for all six T1 histogram parameters that were measured. For NABT, the mean T1 was 1,027 (+/- 74) ms in patients and 969 (+/- 41) ms in controls (p=0.003). There was little difference between the global and NABT histograms, which indicates that most of the whole-brain histogram abnormality derives from normal-appearing tissues. There was a correlation between the Nine-Hole Peg Test and NABT T1 measures. CONCLUSION: There are widespread abnormalities of NABT in early RR MS, which were sensitively detected by T1 relaxation time histogram analysis. As such, T1 histogram analysis appears promising for studying the natural history of early RR MS, and in the monitoring of response to treatment

Normal-appearing brain tissue MTR histograms in clinically isolated syndromes suggestive of MS.

Segmented normal-appearing brain tissue (NABT) was investigated in 40 patients with a recent onset and 13 patients with a remote onset of a clinically isolated syndrome (CIS) using magnetization transfer ratio (MTR) histograms. Abnormalities were present in patients with a high risk for MS (recent onset and T2-weighted lesions present) and in those with a low risk for relapse (recent onset without T2-weighted lesions). Similar mild NABT abnormality was present with CIS and no further disease activity 14 years later. NABT MTR abnormality in CIS may indicate susceptibility to demyelination but not to disease progression.

Classification of disease subgroup and correlation with disease severity using magnetic resonance imaging whole-brain histograms: application to magnetization transfer ratios and multiple sclerosis.

This paper presents a new approach to characterize subtle diffuse changes in multiple sclerosis (MS) using histograms derived from magnetization transfer ratio (MTR) images. Two major parts dominate our histogram analysis; 1) Classification of MTR histograms into control and MS subgroups; 2) Correlation with current disability, as measured by the EDSS scale (a measure of disease severity). Two data reduction schemes are used to reduce the complexity of the analysis: linear discriminant analysis (LDA) and principal component analysis (PCA). LDA is better for the classification of MTR histograms as it takes into account the between-class variation. By using LDA, the space of MTR histograms is transformed to the optimal discriminant space for a nearest mean classifier. In contrast, PCA is useful for correlation with current disability as it takes into account the variation within each subgroup in its process. A multiple regression analysis is used to evaluate the multiple correlation of those principal components with the degree of disability in MS. This is the first application of such classification and correlation techniques to magnetic resonance imaging histogram data. Our MTR histogram analysis approach give improved classification success and improved correlation compared with methods that use traditional histogram features such as peak height and peak location.

Analysis of MTR histograms in multiple sclerosis using principal components and multiple discriminant analysis.

Magnetization transfer ratio (MTR) histograms have the potential to characterize subtle diffuse changes in multiple sclerosis (MS) and other white matter disease. A new method is described which gives improved correlation with the Expanded Disability Status Scale (EDSS). Classification of individual subjects into normal and MS subgroups is shown. Principal component analysis (PCA) and multiple discriminant analysis (MDA) are shown to give results superior to methods of MTR histogram analysis using traditional features such as peak height and peak location. Scatterplots confirm the improved separation between groups achieved using the MDA score. The histogram analysis provides a comparison of two classification approaches, based on PCA and MDA, to recognize differences between normal controls and the four different subgroups of MS disease (and all MS patients). Multiple linear regression of these PCs vs. EDSS established an MR-based measure of disease. Using a central 60-mm slab of brain tissue, the success rate of binary classification between control and MS subgroups using MDA was 75-95%, depending on which two groups were being compared. Multiple regression analysis of EDSS with the first three PCs as independent variables was significant (r = 0.83 for secondary progressive MS, and r = 0.80 for all MS patients).

Magnetisation transfer ratio histogram analysis of primary progressive and other multiple sclerosis subgroups.

INTRODUCTION: Global magnetisation transfer ration (MTR) histogram analysis in the brain offers a method for evaluating pathological change both as a result of lesions and microscopic changes in normal appearing tissues. METHODS: 39 controls and 83 MS patients (46 primary progressive, 11 benign, 10 relapsing-remitting, 16 secondary progressive) were studied to explore the relationship of six conventional MTR histogram parameters with MS clinical subgroups and disability. Principal component (PC) analysis, which makes use of all the histogram data, was also used to examine the relationship between the MTR histogram and disability. RESULTS: When primary progressive patients were compared to controls, there were abnormalities of average MTR, and MTR at the 25th, 50th and 75th percentile. Disabled relapsing onset patients exhibited abnormalities in the same four parameters. Benign and nondisabled relapsing onset patients exhibited no significant abnormalities. Modest correlations were observed between disability and individual MTR parameters in relapse onset but not primary progressive patients--PC analysis revealed stronger and significant associations with disability in both subgroups. (r=0.40 for primary progressive and r=0.51 for relapsing onset). CONCLUSION: A number of MTR parameters are abnormal in primary progressive MS. MTR abnormalities are seen in disabled patients, whether of relapsing or primary progressive onset. The improved correlation with disability obtained by PC analysis suggests a useful role of this method for following clinically relevant pathological changes depicted in the MTR histogram.