Average Sensitivity Values Research Articles

Automatic Optic Disc Center and Boundary Detection in Color Fundus Images

Accurately detection of retinal landmarks, like optic disc, is an important step in the computer aided diagnosis frameworks. This paper presents an efficient method for automatic detection of the optic disc’s center and estimating its boundary. The center and initial diameter of optic disc are estimated by employing an ANN classifier. The ANN classifier employs visual features of vessels and their background tissue to classify extracted main vessels of retina into two groups: the vessels inside the optic disc and the vessels outside the optic disc. To this end, average intensity values and standard deviation of RGB channels, average width and orientation of the vessels and density of the detected vessels their junction points in a window around each central pixel of main vessels are employed. The center of detected vessels, which are belonging to the inside of the optic disc region, is adopted as the optic disc center and the average length of them in vertical and horizontal directions is selected as initial diameter of the optic disc circle. Then exact boundary of the optic disc is extracted using radial analysis of the initial circle. The performance of the proposed method is measured on the publicly available DRIONS, DRIVE and DIARETDB1 databases and compared with several state-of-the-art methods. The proposed method shows much higher mean overlap (70.6%) in the same range of detection accuracy (97.7%) and center distance (12 pixels). The average sensitivity and predictive values of the proposed optic disc detection method are 80.3% and 84.6% respectively.

The Cascaded Enhanced k-Means and Fuzzy c-Means Clustering Algorithms for Automated Segmentation of Malaria Parasites

Malaria continues to be one of the leading causes of death in the world, despite the massive efforts put forth by World Health Organization (WHO) in eradicating it, worldwide. Efficient control and proper treatment of this disease requires early detection and accurate diagnosis due to the large number of cases reported yearly. To achieve this aim, this paper proposes a malaria parasite segmentation approach via cascaded clustering algorithms to automate the malaria diagnosis process. The comparisons among the cascaded clustering algorithms have been made by considering the accuracy, sensitivity and specificity of the segmented malaria images. Based on the qualitative and quantitative findings, the results show that by using the final centres that have been generated by enhanced k-means (EKM) clustering as the initial centres for fuzzy c-means (FCM) clustering, has led to the production of good segmented malaria image. The proposed cascaded EKM and FCM clustering has successfully segmented 100 malaria images of Plasmodium Vivax species with average segmentation accuracy, sensitivity and specificity values of 99.22%, 88.84% and 99.56%, respectively. Therefore, the EKM algorithm has given the best performance compared to k-means (KM) and moving k-means (MKM) algorithms when all the three clustering algorithms are cascaded with FCM algorithm.

Comparison of Saccadic Vector Optokinetic Perimetry and Standard Automated Perimetry in Glaucoma. Part I: Threshold Values and Repeatability.

PurposeWe evaluated threshold saccadic vector optokinetic perimetry (SVOP) and compared results to standard automated perimetry (SAP).MethodsA cross-sectional study was done including 162 subjects (103 with glaucoma and 59 healthy subjects) recruited at a university hospital. All subjects underwent SAP and threshold SVOP. SVOP uses an eye tracker to monitor eye movement responses to stimuli and determines if stimuli have been perceived based on the vector of the gaze response. The test pattern used was equivalent to SAP 24-2 and stimuli were presented at Goldmann III. Average and pointwise sensitivity values obtained from both tests were compared using Pearson's correlation coefficient. Two versions of SVOP were evaluated.ResultsA total of 124 tests were performed with SAP and SVOP version 2. There was excellent agreement between mean threshold values obtained using SVOP and SAP (r = 0.95, P < 0.001). Excluding the blind spot, correlation between SVOP and SAP individual test point sensitivity ranged from 0.61 to 0.90, with 48 of 54 (89%) test points > 0.70. Overall SVOP showed good repeatability with a Pearson correlation of 0.88. The repeatability on a point-by-point basis ranged from 0.66 to 0.98, with 45 of 54 points (83%) > 0.80. Repeatability of SAP was 0.87, ranging from 0.69 to 0.96, with 47 of 54 (87%) points > 0.80.ConclusionEye-tracking perimetry is repeatable and compares well with the current gold standard of SAP. The technique has advantages over conventional perimetry and could be useful for evaluating glaucomatous visual field loss, particularly in patients who may struggle with conventional perimetry.Translational RelevanceSuprathreshold SVOP already is in the field. To our knowledge, this is the first report of threshold SVOP and provides a benchmark for future iterations.

QRS complex detection in ECG signals using locally adaptive weighted total variation denoising

The QRS complex is the most prominent feature in the electrocardiogram (ECG), therefore, its detection is required for delineation of other waves and segments in the ECG and derivation of additional clinically useful information. QRS detection is complicated by factors like varying QRS morphologies, noise, artefacts and interference from tall and pointed P- and T-waves. In this paper, we propose a novel technique for QRS detection by preprocessing the ECG using weighted total variation (WTV) denoising. A local estimate of noise in the signal block under consideration is used to determine the regularization parameter in WTV minimization, which determines the amount of smoothing applied. This makes the denoising locally adaptive. The weights are chosen so as to give preference to preservation of QRS complexes over P- and T-waves while smoothing. Thus, the technique can simultaneously reduce the higher frequency noise as well as the lower frequency interference from P- and T-waves, in spite of the fact that they have overlapping spectra with the QRS complexes. The proposed method is evaluated on the MIT-BIH arrhythmia database and gives improved detection accuracy over established and state-of-the-art techniques. The technique has low computational load, therefore, it can be used for fast offline QRS detection in long duration ECG records, as well as real-time QRS detection in block-by-block processing mode. The average values of sensitivity, positive predictivity and detection error rate are 99.90%, 99.88% and 0.23%, for the offline implementation, respectively, and 99.86%, 99.85% and 0.29%, for the real-time mode, respectively.

Classification of ECG heartbeats using nonlinear decomposition methods and support vector machine

Classifying electrocardiogram (ECG) heartbeats for arrhythmic risk prediction is a challenging task due to minute variations in the amplitude, duration and morphology of the ECG signal. In this paper, we propose two feature extraction approaches to classify five types of heartbeats: normal, premature ventricular contraction, atrial premature contraction, left bundle branch block and right bundle branch block. In the first approach, ECG beats are decomposed into intrinsic mode functions (IMFs) using ensemble empirical mode decomposition (EEMD). Later four parameters, namely the sample entropy, coefficient of variation, singular values, and band power of IMFs are extracted as features. In the second approach, the same features are computed from IMFs extracted using an empirical mode decomposition (EMD) algorithm. The features obtained from the two approaches are independently fed to the sequential minimal optimization-support vector machine (SMO-SVM) for classification. We used two arrhythmia databases for our evaluation: MIT-BIH and INCART. We compare the proposed approaches with existing methods using the performance measures given by the average values of (i) specificity, (ii) sensitivity, and (iii) accuracy. The first approach demonstrates significant performance with 98.01% sensitivity, 99.49% specificity, and 99.20% accuracy for the MIT-BIH database and 95.15% sensitivity, 98.37% specificity, and 97.57% accuracy for the INCART database.

Evolving strategies for the development and evaluation of a computerised melanoma image analysis system

The rising incidence of melanoma has increased demand for early detection tools, which enable an objective decision and prevent the need for biopsies, the current gold standard in lesion diagnosis. In this study, we report the development of such a tool, a melanoma image analysis system (MIAS), designed to improve the standard of care and classify suspicious skin lesions as benign or malignant. In addition, our methodological approach to curate an image data-set (107 benign lesions and 94 melanomas) and subsequently divide into five balanced sets in order to create a quadratic discriminant analysis classifier, allowed us to determine the causes of weaknesses within our classifier. We identified the classifier’s weaknesses by analysing for which benign images the system failed and the reasons for being so. Further analysis also determined which benign categories were more difficult to classify and identified which features were useful for classification. Our classification system resulted in average sensitivity and specificity values of 79.8 and 74.8%, respectively, with an overall average accuracy of 76.6%. Developing an objective approach to visual inspection may be augmented by a methodological approach in creating a MIAS. Interestingly, the predominant (48.1%) benign misclassifications were determined to be Reed–Spitz nevi, also suggesting that in MIAS development, certain benign categories may affect future development and evaluation strategies.

Retail Order-Taking Strategies Under Competing Trade Credit Policies With Varying Demands

This paper discusses optimal order-taking strategies under competing trade credit policies with varying demands. This study examines three decision-making scenarios, namely, (i) a centralized supply chain, (ii) a decentralized supply chain, and (iii) a coordinated supply chain with a buyback contract. Optimal decisions are obtained for each scenario. We find that if the average forecast value of credit sensitivity is higher than the real value, then the supply chain is at risk of having an overestimated performance; if the average forecast value of credit sensitivity is lower than the actual value, then the supply chain is at risk of having an underestimated performance. As the sensitivity of the credit competition increases, the supply chain’s performance decreases. The supply chain can be coordinated by a buyback contract.

A Multivariate Approach for Patient-Specific EEG Seizure Detection Using Empirical Wavelet Transform.

This paper investigates the multivariate oscillatory nature of electroencephalogram (EEG) signals in adaptive frequency scales for epileptic seizure detection. The empirical wavelet transform (EWT) has been explored for the multivariate signals in order to determine the joint instantaneous amplitudes and frequencies in signal adaptive frequency scales. The proposed multivariate extension of EWT has been studied on multivariate multicomponent synthetic signal, as well as on multivariate EEG signals of Children's Hospital Boston-Massachusetts Institute of Technology (CHB-MIT) scalp EEG database. In a moving-window-based analysis, 2-s-duration multivariate EEG signal epochs containing five automatically selected channels have been decomposed and three features have been extracted from each 1-s part of the 2-s-duration joint instantaneous amplitudes of multivariate EEG signals. The extracted features from each oscillatory level have been processed using a proposed feature processing step and joint features have been computed in order to achieve better discrimination of seizure and seizure-free EEG signal epochs. The proposed detection method has been evaluated over 177 h of EEG records using six classifiers. We have achieved average sensitivity, specificity, and accuracy values as 97.91%, 99.57%, and 99.41%, respectively, using tenfold cross-validation method, which are higher than the compared state of art methods studied on this database. Efficient detection of epileptic seizure is achieved when seizure events appear for long duration in hours long EEG recordings. The proposed method develops time-frequency plane for multivariate signals and builds patient-specific models for EEG seizure detection.

Online Monitoring of Superoxide Anions Released from Skeletal Muscle Cells Using an Electrochemical Biosensor Based on Thick-Film Nanoporous Gold

Online detection and accurate quantification of superoxide anions released from skeletal muscle tissue is important in both physiological and pathological contexts. Above certain physiologically redundant levels, superoxides may exert toxic effects. Here we present design, fabrication, and successful testing of a highly sensitive electrochemical superoxide biosensor based on nanoporous gold (NPG) immobilized with cytochrome-c (cyt-c). A significant 14-fold enhancement in the biosensor sensitivity was achieved using NPG instead of nonporous gold, enabling the device to quantify minuscule levels of superoxides. Such improvement was attributed to the very large surface-to-volume ratio of the NPG network. The average values of superoxide sensitivity and analytical limit of detection (LOD) were 1.90 ± 0.492 nA nM–1 cm–2 and 3.7 nM, respectively. The sensor was employed to measure the rates of superoxide release from C2C12 myoblasts and differentiated myotubes upon stimulation with an endogenous superoxide-prod...

Role of MRI with hepatospecific contrast agent in the identification and characterization of focal liver lesions: pathological correlation in explanted livers.

To assess the diagnostic performance of magnetic resonance imaging (MRI) with gadoxetic acid in the identification of hepatocellular carcinoma (HCC) nodules by comparison with histological findings. In a cohort of patients suffering from cirrhosis of various etiologies (chronic hepatitis C virus (HCV) or hepatitis B virus (HBV), alcohol abuse, cryptogenic forms), we selected 17 patients affected by HCC who were eligible for liver transplantation on the basis of a computed-tomography (CT) total-body examination. Such patients also underwent an MRI examination under basal conditions, and with four dynamic phases, as well as a hepatobiliary phase acquired after at least 20min and recognized by the excretion of contrast agent into the bile duct, following intravenous administration of 0.05mol/kg of gadoxetic acid (gadoxetate disodium, Primovist(®); Bayer, Osaka, Japan). The MRI images were then evaluated in a double-blinded experimental setup by two radiologists experienced in imaging of the liver. The diagnosis of HCC was made in the presence of nodular lesions that showed typical or atypical enhancement patterns. The liver was subsequently explanted (on average 47.4days after MRI evaluation), dissected into 1-cm samples, and histologically evaluated according to the classification of Edmondson-Steiner. At the histopathological examination, 46 nodules were identified, on average 2.7 nodules for each patient. Of these, 37 were hepatocellular carcinomas, 3 were characterized by histologically unrecognizable complete necrosis, and 6 showed high-grade dysplasia. MRI with hepatospecific contrast medium showed inter-observer average values of sensitivity, specificity, and diagnostic accuracy of 94.6, 90, and 93.6%, respectively. In one case, a nodule was not identified by MRI with gadoxetic acid, even in the hepatospecific phase (false negative (FN)). This result could be implicated to the long time interval between the analysis and the explant (88days). In another case, there was an overdiagnosis of a HCC with a typical nodular pattern (false positive (FP)), but which most likely should have been attributed to a previous echinococcus cyst. MRI analysis, in combination with the study of the hepatobiliary phase, also showed a greater sensitivity, the same specificity, and a greater diagnostic accuracy compared to MRI evaluated only in the dynamic phases (with an average percentage between the two operators, respectively, of 75.7, 90, and 78%). MRI with gadoxetic acid shows a diagnostic accuracy superior to contrast-enhanced MRI, allowing for the diagnosis of additional lesions, and it could be considered as an imaging method to carry out a more appropriate management of waiting lists for liver transplants.

An approach for automatic segmentation of thermal imaging in Computer Aided Diagnosis

Breast cancer is major cause of the high mortality rates among young women in developing countries. In Latin America, it is a great health problem, as well. For example in Brazil and Ecuador, this is the leading cause of cancer among women around 35 years old. Early detection is important to improve the chance of cure. Thermal imaging has the ability to show regions where there is any potential cancer by indicating areas of the body where there is an abnormal temperature variation. Moreover, thermography can detect suspicious regions in patients of any age, even in cases of dense breasts, where the detection of an abnormality can not be achieved by mammography. An essential step in the use of thermal imaging systems is the development of computer-aided diagnosis (CAD). However, any development towards a CAD system or even an examination guided by computer should consider adequate extraction of the region of interest (ROI). This paper proposes a methodology for automatic segmentation of thermal imaging breast and validation of the results by generating a Ground Truth (GT). The automatic method proposed in this paper consists of several image-processing techniques such as thresholding, clustering, edge detection and refinement, among others. For the evaluation of the results the developed GT are presently available on the Internet, in order to allow proper verification of the results. Finally, the results obtained by the proposed methodology for the 328 images used in this study showed average values of accuracy and sensitivity around 96% and 97%, respectively.

An evolutionary algorithm for the segmentation of muscles and bones of the lower limb.

In the field of medical image segmentation, muscles segmentation is a problem that has not been fully resolved yet. This is due to the fact that the basic assumption of image segmentation, which asserts that a visual distinction should ex- ist between the different structures to be identified, is infringed. As the tissue composition of two different muscles is the same, it becomes extremely difficult to distinguish one another if they are near. We have developed an evolutionary algorithm which selects the set and the sequence of morphological operators that better segments muscles and bones from an MRI image. The achieved results shows that the developed algorithm presents average sensitivity values close to 75% in the segmentation of the different processed muscles and bones. It also presents average specificity values close to 93% for the same structures. Furthermore, the algorithm can identify muscles that are closely located through the path from their origin point to their insertions, with very low error values (below 7%) .

Strain-rate sensitivity of scratch hardness and deformation mechanism in nanocrystalline Ni under micro-scratch testing

To investigate the strain-rate sensitivity in nanocrystalline (NC) materials using a single experimental technique over a wide range of strain rate, the micro-scratch testing technique was selected to obtain consistent and systematic data of the strain-rate sensitivity in a fully dense, high purity, and well-characterized electrodeposited NC Ni sample. The scratch characterizations and mechanical properties of the sample under the different scratch velocities were discussed in details. First, some critical parameters under the micro-scratch testing, such as strain rates, scratch ditch widths, were confirmed. Then, the scratch hardness under the different scratch velocity was investigated. From the results, the sample exhibits strain-rate sensitive mechanical properties. Further, the average values of strain-rate sensitivity of the sample were calculated. The values increase with increasing scratch velocity, and exhibit higher values that are of the order of 0.03–0.1. The greater strain-rate sensitivity exponents indicate that the NC Ni samples have different fundamental physical deformation mechanisms. Finally, the fundamental physical deformation mechanism under scratch testing was inspected using SEM and TEM technique. From the SEM and TEM morphologies, the grain boundary dislocation pile-ups should be a carrier of plastic flow under scratch testing in the NC Ni sample.

Measurement of sensitivity distribution map of a ferroelectret polymer film

Electromechanical Film (EMFi) is an electroactive ferroelectret polymer film with a special cellular structure. Sensitivity distribution maps of EMFi sensors were measured in this paper. Seven EMFi sensors with evaporated gold electrodes, a size of $55~\mathrm{mm}\times55~\mathrm{mm}$ , were measured with a measurement grid spacing of 5 mm. The sensitivity is defined here as the charge generated by the EMFi sensor divided by the force applied to excite the sensor, and this is closely related to piezoelectric $\text{d}_{33}$ coefficient. The sensitivity of EMFi was found to vary in different places of the sensors and between the measured sensors. This is presumably due to the relatively large gas voids and local corona breakdowns. The maximum and minimum sensitivities were found to vary around 30% from the measured average sensitivity value. A thermally stimulated current analysis confirmed the results obtained in the sensitivity distribution map measurements.

CORTICAL BONE SEGMENTATION USING WATERSHED AND REGION MERGING BASED ON STATISTICAL FEATURES

Research on biomedical image is a subject that attracted many researchers’ interest. This is because the biomedical image could contain important information to help analyze a disease. One of the existing researches in his field uses dental panoramic radiographs image to detect osteoporosis. The analyzed area is the width of cortical bone. To analyze that area, however, we need to determine the width of the cortical bone. This requires proper segmentation on the dental panoramic radiographs image. This study proposed the integration of watershed and region merging method based on statistical features for cortical bone segmentation on dental panoramic radiographs. Watershed segmentation process was performed using gradient magnitude value from the input image. The watershed image that still has excess segmentation could be solved by region merging based on statistical features. Statistical features used in this study are mean, standard deviation, and variance. The similarity of adjacent regions is measured using weighted Euclidean distance from the statistical feature of the regions. Merging process was executed by incorporating the background regions as many as possible, while keeping the object regions from being merged. The segmentation result has succeeded in forming the contours of the cortical bone. The average value of accuracy is 93.211%, while the average value of sensitivity and specificity is 93.858% and respectively.

Achieving high superplasticity of a new Al–Mg–Sc–Zr alloy sheet prepared by a simple thermal–mechanical process

A new fine-grained Al–Mg–Sc–Zr alloy sheet with the grain size of 2.5μm was prepared by simple rolling and thermal treatment process. Superplastic investigations in the temperature range of 450–550°C and initial strain rate range of 1×10−3s−1–1×10−1s−1 show that a maximum elongation of 3250% is achieved at 525°C and 5×10−3s−1. The alloy also exhibits excellent superplastic ductility (>2390%) in the temperature interval 475–500°C at 5×10−2s−1. Analyses on the superplastic data reveal that the average values of the strain rate sensitivity and activation energy of the Al–Mg–Sc–Zr alloy are about 0.45 and 83KJ/mol, respectively. The microstructure results show that the studied alloy is characterized by a high fraction of low angle grain boundaries and strong Copper, Brass and Cube texture. During superplastic deformation, low angle grain boundaries gradually transfer into high angle grain boundaries and the grains gradually orientate randomly. The fine coherent Al3(Sc, Zr) particles play an important role in obtaining excellent superplasticity. Grain boundary sliding is the predominant deformation mechanism.

Abstract 4472: Plitidepsin shows antitumor activity in patient-derived tumor xenografts and hematologic malignancies

Abstract Plitidepsin is a new marine-derived anti-tumor agent, originally isolated from the tunicate Aplidium albicans that is now fully obtained by total chemical synthesis. The compound is currently being evaluated in a pivotal phase III trial in patients with refractory/relapsed multiple myeloma. Previous in vitro studies in 26 hematological cell lines demonstrated that plitidepsin was more active in multiple myeloma and non-Hodgkin lymphoma compared to CML, AML and ALL derived cell lines as determined in a 4 day proliferation assay in suspension cultures. We have recently shown that plitidepsin interacts selectively with eEF1A2 (eukaryotic translation elongation factor 1 alpha 2) in tumor cells, which results in apoptotic cell death. To identify target tumor types for future clinical studies, the inhibition of colony formation of plitidepsin was assessed in vitro in 116 established patient derived xenografts tumor models from 23 different tumor types, including solid tumors and hematological malignancies such as myeloma, leukemia and lymphoma. The anti-tumor activity of plitidepsin varied depending on the concentration used, achieving a mean IC70 (inhibitory concentration of 70%) value of 3 nM (range between 0.03 nM and &amp;gt;1 μM), which can be also achieved in patients treated using the clinical recommended dose or less. 39 out of 116 tumors showed above average sensitivity (IC70 &amp;lt; 1nM) and IC70 values in these tumors were about 9-fold lower than the mean of all tumors as tested. Sensitive tumor types were seen among different tumor types, such as hematological malignancies, pleuramesothelioma, melanoma, small cell and non-small cell lung cancer (both adeno and large cell), as well as renal cancer. In conclusion plitidepsin showed a broad differential antitumor activity in vitro in hematological malignancies and in solid tumors. Bioinformatic analyses are currently underway to identify biomarkers and/or gene signatures that can predict sensitivity or resistance to plitidepsin for selecting patients being responsive to the drug. Citation Format: Armin Maier, Gerhard Kelter, Vincent Vuaroquaux, Pablo M. Avilés Marin, Carmen Cuevas, Carlos M. Galmarini, Heinz H. Fiebig. Plitidepsin shows antitumor activity in patient-derived tumor xenografts and hematologic malignancies. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 4472. doi:10.1158/1538-7445.AM2015-4472

Robust detection of heart beats in multimodal records using slope- and peak-sensitive band-pass filters

In this work, we present the development, architecture and evaluation of a new and robust heart beat detector in multimodal records. The detector uses electrocardiogram (ECG) signals, and/or pulsatile (P) signals, such as: blood pressure, artery blood pressure and pulmonary artery pressure, if present. The base approach behind the architecture of the detector is collecting signal energy (differentiating and low-pass filtering, squaring, integrating). To calculate the detection and noise functions, simple and fast slope- and peak-sensitive band-pass digital filters were designed. By using morphological smoothing, the detection functions were further improved and noise intervals were estimated. The detector looks for possible pacemaker heart rate patterns and repairs the ECG signals and detection functions. Heart beats are detected in each of the ECG and P signals in two steps: a repetitive learning phase and a follow-up detecting phase. The detected heart beat positions from the ECG signals are merged into a single stream of detected ECG heart beat positions. The merged ECG heart beat positions and detected heart beat positions from the P signals are verified for their regularity regarding the expected heart rate. The detected heart beat positions of a P signal with the best match to the merged ECG heart beat positions are selected for mapping into the noise and no-signal intervals of the record. The overall evaluation scores in terms of average sensitivity and positive predictive values obtained on databases that are freely available on the Physionet website were as follows: the MIT–BIH Arrhythmia database (99.91%), the MGH/MF Waveform database (95.14%), the augmented training set of the follow-up phase of the PhysioNet/Computing in Cardiology Challenge 2014 (97.67%), and the Challenge test set (93.64%).

Retinal vessel segmentation using multi-scale textons derived from keypoints

This paper presents a retinal vessel segmentation algorithm which uses a texton dictionary to classify vessel/non-vessel pixels. However, in contrast to previous work where filter parameters are learnt from manually labelled image pixels our filter parameters are derived from a smaller set of image features that we call keypoints. A Gabor filter bank, parameterised empirically by ROC analysis, is used to extract keypoints representing significant scale specific vessel features using an approach inspired by the SIFT algorithm. We first determine keypoints using a validation set and then derive seeds from these points to initialise a k-means clustering algorithm which builds a texton dictionary from another training set. During testing we use a simple 1-NN classifier to identify vessel/non-vessel pixels and evaluate our system using the DRIVE database. We achieve average values of sensitivity, specificity and accuracy of 78.12%, 96.68% and 95.05%, respectively. We find that clusters of filter responses from keypoints are more robust than those derived from hand-labelled pixels. This, in turn yields textons more representative of vessel/non-vessel classes and mitigates problems arising due to intra and inter-observer variability.

Test-Retest Repeatability of Microperimetry at the Border of Deep Scotomas.

The purpose of this study was to examine the test-retest repeatability of microperimetric sensitivity at the border of deep scotomas. Thirty normal participants underwent two examinations, each on the Macular Integrity Assessment (MAIA) microperimeter and on the MP-1 microperimeter (four examinations in total). A customized stimulus pattern allowed microperimetric sensitivity to be measured at the border of the optic nerve head (ONH), which acted as a model for the border of a deep scotoma-and also at the macular and peripapillary region. There were no significant changes in average point-wise sensitivity (PWS) values between the two examinations for all three regions using the MAIA microperimeter (P ≥ 0.262). The PWS coefficient of repeatability (CoR) was ±12.99 dB at the border of the ONH, which was significantly larger than points in the macular and peripapillary regions (P > 0.001). A significant decrease in average PWS, using the MP-1 microperimeter at the macular and peripapillary region (P < 0.001), meant that the PWS CoR could not be determined in these regions. No significant changes in average PWS were observed at the border of the ONH (P = 0.223), and the PWS CoR was ±7.52 dB in this region. Microperimetric test-retest repeatability at the border of a deep scotoma was worse than at other areas of normal retina, and this highlights the limitation of applying a single estimate of test-retest repeatability to determine whether significant functional decline has occurred at the border of a deep scotoma.