Expert Classification Research Articles

Automatic and Probabilistic Foehn Diagnosis with a Statistical Mixture Model

AbstractDiagnosing foehn winds from weather station data downwind of topographic obstacles requires distinguishing them from other downslope winds, particularly nocturnal ones driven by radiative cooling. An automatic classification scheme to obtain reproducible results that include information about the (un)certainty of the diagnosis is presented. A statistical mixture model separates foehn and no-foehn winds in a measured time series of wind. In addition to wind speed and direction, it accommodates other physically meaningful classifiers such as the (potential) temperature difference to an upwind station (e.g., near the crest) or relative humidity. The algorithm was tested for Wipp Valley in the central Alps against human expert classification and a previous objective method (Drechsel and Mayr 2008), which the new method outperforms. Climatologically, using only wind information gives nearly identical foehn frequencies as when using additional covariables. A data record length of at least one year is required for satisfactory results. The suitability of mixture models for objective classification of foehn at other locations will have to be tested in further studies.

Analysing spatio-temporal footprints of urbanization on environment of Surat city using satellite-derived bio-physical parameters

Our study examines the relationships among various environmental variables in Surat city using remote sensing. Landsat Thematic Mapper satellite data were used in conjugation with geospatial techniques to study urbanization and correlation among satellite-derived biophysical parameters namely, normalized difference vegetation index (NDVI), normalized difference built-up index (NDBI), normalized difference water index (NDWI), normalized difference bareness index (NDBaI) and land surface temperature (LST). A modified NDWI (MNDWI) was used for extracting areas under water. Land use/land cover classification was performed using hierarchical decision tree classification technique using ERDAS IMAGINE Expert classifier with an accuracy of 90.4% for 1990 and 85% for 2009. It was found that city has expanded over 42.75 sq.km within two decades. Built-up, fallow and sediment land use classes exhibited high dynamics with increase of nearly 200% and 50% and decrease of 55% respectively from 1990 to 2009. Vegetation and water classes were less dynamic with 20% decrease and 15% increase. The transformation of land parcels from vegetation to built-up, vegetation to fallow and fallow to built-up has resulted in increase of LST by 5.5 ± 2.6°C, 6.7 ± 3°C and 3.5 ± 2.9°C, respectively.

THU0409 Mixed connective tissue disease and undifferentiated connective tissue disease: Do doctors classify correctly?

BackgroundMixed connective tissue disease (MCTD) and undifferentiated connective tissue disease (UCTD) are heterogeneous diseases which clinical manifestations are considerable overlapped with other connective tissue disease. Clinical diagnosis is usually depended...

FRI0296 Level of confidence in the diagnosis of spondyloarthritis by expert readers in sacroiliac joint MRI: a new gold standard criterion for defining a positive MRI?

BackgroundDefining a positive MRI in early axial spondyloarthritis (SpA) requires a gold standard criterion for the diagnosis of SpA and is typically based on clinical findings because structural changes on...

SAT0466 PED-BD COHORT 2013: Expert Consensus Classification Gives Higher Sensitivity than the International Study Group Criteria to Define Behcet’s Disease in Children

BackgroundBD is rarely encountered in children where the disease is very difficult to recognize. The outcome of patients with few symptoms is currently unknownObjectivesTo define the outcome of paediatric patients...

Health Monitoring for Elderly: An Application Using Case-Based Reasoning and Cluster Analysis

This paper presents a framework to process and analyze data from a pulse oximeter which remotely measures pulse rate and blood oxygen saturation from a set of individuals. Using case-based reasoning (CBR) as the backbone to the framework, records are analyzed and categorized according to their similarity. Record collection has been performed using a personalized health profiling approach in which participants wore a pulse oximeter sensor for a fixed period of time and performed specific activities for pre-determined intervals. Using a variety of feature extraction methods in time, frequency, and time-frequency domains, as well as data processing techniques, the data is fed into a CBR system which retrieves most similar cases and generates an alarm according to the case outcomes. The system has been compared with an expert's classification, and a 90% match is achieved between the expert's and CBR classification. Again, considering the clustered measurements, the CBR approach classifies 93% correctly both for the pulse rate and oxygen saturation. Along with the proposed methodology, this paper provides a basis for which the system can be used in the analysis of continuous health monitoring and can be used as a suitable method in home/remote monitoring systems.

Greater sexual reproduction contributes to differences in demography of invasive plants and their noninvasive relatives

An understanding of the demographic processes contributing to invasions would improve our mechanistic understanding of the invasion process and improve the efficiency of prevention and control efforts. However, field comparisons of the demography of invasive and noninvasive species have not previously been conducted. We compared the in situ demography of 17 introduced plant species in St. Louis, Missouri, USA, to contrast the demographic patterns of invasive species with their less invasive relatives across a broad sample of angiosperms. Using herbarium records to estimate spread rates, we found higher maximum spread rates in the landscape for species classified a priori as invasive than for noninvasive introduced species, suggesting that expert classifications are an accurate reflection of invasion rate. Across 17 species, projected population growth was not significantly greater in invasive than in noninvasive introduced species. Among five taxonomic pairs of close relatives, however, four of the invasive species had higher projected population growth rates compared with their noninvasive relative. A Life Table Response Experiment suggested that the greater projected population growth rate of some invasive species relative to their noninvasive relatives was primarily a result of sexual reproduction. The greater sexual reproduction of invasive species is consistent with invaders having a life history strategy more reliant on fecundity than survival and is consistent with a large role of propagule pressure in invasion. Sexual reproduction is a key demographic correlate of invasiveness, suggesting that local processes influencing sexual reproduction, such as enemy escape, might be of general importance. However, the weak correlation of projected population growth with spread rates in the landscape suggests that regional processes, such as dispersal, may be equally important in determining invasion rate.

The pachyderm and the pixel: an assessment of elephant habitat suitability in Sonitpur, India

Remote sensing and geographic information systems (GISs) are increasingly being used in protected area monitoring and habitat suitability studies. In this article, Erdas Imagine's Expert Classifier tool was used to assess the specific trajectories of habitat suitability change during 1994–2007 in the Sonitpur elephant habitat, India. Sonitpur has been witness to increasing human–elephant conflict in the past decade. The suitability analysis took into account information relating to forest type and density, elevation, slope, source of water, human activities in terms of settlement, agriculture, tea plantations, roads, and railways. Satellite imagery, data from topographical maps, digital elevation data, and global positioning system readings formed the major data inputs that were incorporated into a GIS. Various decision rules were created and confidence levels assigned to the input layers to generate high, medium, and low habitat suitability. The area witnessed a sharp decline in suitability from 63% in 1994 to less than 38% by 2007. The high-suitability area declined by more than 50% during this period. The unique natural protected areas of Sonitpur, which are a mix of reserved forests, wildlife sanctuaries, and a national park, urgently need to be protected from further habitat degradation.

Assessing Mild Cognitive Impairment with Amyloid and Dopamine Terminal Molecular Imaging

We evaluated PET-based classification of neurodegenerative pathology in mild cognitive impairment (MCI). Our study was a cross-sectional and prospective evaluation of a cohort of 27 MCI subjects drawn from a university-based cognitive disorders clinic. We compared expert clinical consensus classification of MCI at entry and possible dementia at follow-up with molecular imaging-based classification using (11)C-dihydotetrabenazine PET measurement of striatal dopamine terminal integrity and (11)C-Pittsburgh compound B ((11)C-PiB) PET measurement of cerebral amyloid burden. Eleven subjects were initially classified clinically as amnestic MCI, 7 as multidomain MCI, and 9 as nonamnestic MCI. At a mean follow-up of 3 y, 18 subjects converted to dementia. PET imaging evidence of significant cerebral amyloid deposition or nigrostriatal denervation was a strong predictor of conversion to dementia. There was only moderate concordance between expert clinical classifications and PET-based classifications of dementia subtypes. Combined PET molecular imaging of cerebral amyloid burden and striatal dopamine terminal integrity may be useful for identifying subjects at high risk for progression to dementia and in defining neurochemically differentiated subsets of MCI subjects.

Assessment of digital image classification algorithms for forest and land-use classification in the eastern Himalayas using the IRS LISS III sensor

The article describes the assessment and discusses the potential of different methods of classifying Linear Imaging Self-Scanning Sensor (LISS) III sensor data for vegetation-cover type and land-use mapping in the hilly terrain of the eastern Arunachal Pradesh, in the northeastern part of India. The forest cover types and their distribution in the study area are governed by climatic conditions and topographical features, which change along the gradient from tropical in the south to alpine in the Greater Himalayas in the north. Arunachal Pradesh is part of the Himalayan Biodiversity Hotspot and is at the tri-junction of Indian, Sino-Japanese, and Indo-Malayan floristic regions. The evergreen forests in the area represent climatic climaxes, which are partly original virgin and partly affected by anthropogenic pressures. Due to phenological and physiognomic similarities in ecotone regions, the differences in the spectral reflectance between adjacent forest types are not well pronounced. The age of the forests, terrain characteristics, and the nature of the vegetation itself could be other reasons for the near-similar reflectance. It is for these reasons that conventional classification algorithms for supervised and unsupervised classification do not perform well. Therefore, there is a need to find a suitable method for vegetation and land-use mapping with high mapping accuracy in this region, which is a biodiversity hotspot. A suitable classification technique is important to characterize vegetation-cover type in this complex terrain. We tested unsupervised, supervised, hybrid, object-oriented, and expert classification techniques for vegetation and land-use mapping on Indian Remote Sensing (IRS)-1C LISS III data. The expert classifier produced the highest accuracy (93.39%) followed by object-oriented, hybrid, and supervised and unsupervised classification techniques.

Critical assessment of automated flow cytometry data analysis techniques

Traditional methods for flow cytometry (FCM) data processing rely on subjective manual gating. Recently, several groups have developed computational methods for identifying cell populations in multidimensional FCM data. The Flow Cytometry: Critical Assessment of Population Identification Methods (FlowCAP) challenges were established to compare the performance of these methods on two tasks: (i) mammalian cell population identification, to determine whether automated algorithms can reproduce expert manual gating and (ii) sample classification, to determine whether analysis pipelines can identify characteristics that correlate with external variables (such as clinical outcome). This analysis presents the results of the first FlowCAP challenges. Several methods performed well as compared to manual gating or external variables using statistical performance measures, which suggests that automated methods have reached a sufficient level of maturity and accuracy for reliable use in FCM data analysis.

Early results of real-time qualitative sonoelastography in the evaluation of parotid gland masses: A study with histopathological correlation

Sonoelastography has been used to differentiate malignant from benign lesions in numerous types of tissues including breast, prostate, liver, blood vessels, thyroid, musculoskeletal structures, and salivary glands. To evaluate the efficacy and application of real-time qualitative sonoelastography in the differentiation of benign and malignant focal parotid gland lesions. A total of 75 patients (36 boys/men, 39 girls/women; age range, 10-83 years) with 81 lesions were evaluated prospectively by sonoelastography performed and interpreted by two expert radiologists. The results of these experts classification and scoring of lesions according to relative stiffness of the mass were compared with each other and with histopathological findings. The interpretation of sonoelastography scores of 1-4 were as follows: 1, soft; 2, mostly soft; 3, mostly stiff; and 4, stiff. The kappa statistic of 0.508 (P < 0.001) indicated moderate agreement between the two radiologists. The sonoelastography scores correctly diagnosed 30 of 49 benign tumors (sensitivity, 61.2%) and 19 of 32 malignant tumors (specificity, 59.4%). The area under the receiver-operating characteristic curve was 0.603. The diagnostic value of sonoelastography for evaluating pleomorphic adenomas, Warthin tumors, adenoid cystic carcinoma, and high-grade tumors was low, whereas the diagnostic rates for low-grade tumors such as mucoepidermoid carcinoma, acinic cell carcinoma, and metastases of basal cell carcinoma were better with sonoelastography. Although sonoelastography seems to be promising in the differentiating of low-grade malignancies, the primary role of radiology is currently limited to determination of localization, size, and morphology of parotid tumors.

Хемометрический и ИК спектроскопический анализ клейких лент

Chemometric methods for processing spectroscopic data are commonly used methods for a wide variety of chemical tasks that are gaining increasing popularity. In this work we showed the possibility of applying chemometric algorithms for the analysis of adhesive tapes - very complex objects in terms of their identification and classification. Using principal component analysis (PCA) expert classification and identification tasks in relation to fragments of biaxially oriented films of isotactic polypropylene adhesive tapes were solved, allowing the definition of objects from their FT-IR spectra. Regression methods (projection to latent structures ( PLS ), ridge regression, multiply linear regression (MLR) and principal component regression (PCR)) were used to construct models for determining the angles between the projection axes of the refractive index ellipsoid in the plane of the geometric boundary and polypropylene tape. The best prediction model was obtained using PLS ( RMSE = 1.1, R 2 = 0.93 for cross validation) indicated that the application of chemometric methods is promising in the adhesive IR spectra tapes analysis, in particularly in expert applications. Keywords: chemometrics, the principal component analysis (PCA), projection on latent structures (PLS), IR spectroscopy, polypropylene films, expertise (Russian) DOI: http://dx.doi.org/10.15826/analitika.2013.17.3.011 A.M. Tsikin, Y.B. Monakhova, S.P. Kurchatkin, S.P. Mushtakova Saratov State University named after N.G. Chernyshevsky, Saratov, Russian Federation

Phylogenetic Inference from Word Lists Using Weighted Alignment with Empirically Determined Weights

The paper investigates the task of inferring a phylogenetic tree of languages from the collection of word lists made available by the Automated Similarity Judgment Project. This task involves three steps: (1) computing pairwise word distances, (2) aggregating word distances to a distance measure between languages and inferring a phylogenetic tree from these distances, and (3) evaluating the result by comparing it to expert classifications. For the first task, weighted alignment will be used, and a method to determine weights empirically will be presented. For the second task, a novel method will be developed that attempts to minimize the bias resulting from missing data. For the third task, several methods from the literature will be applied to a large collection of language samples to enable statistical testing. It will be shown that the language distance measure proposed here leads to substantially more accurate phylogenies than a method relying on unweighted Levenshtein distances between words.

An expert classification system of pollen of Onopordum using a rough set approach

Although pollen grains have a complicated 3-dimensional structure, they can be distinguished from one another by their specific and distinctive characteristics. Using microscopic differences between the pollen grains, it may be possible to identify them by family or even at the genus level. However for the identification of pollen grains at the taxon level, we require expert computer systems. For this purpose, we used 20 different pollen types, obtained from the genus Onopordum L. (Asteraceae). For each pollen grain, 30 different images were photographed by microscope system and 11 different characteristic features (polar axis, equatorial axis, P/E ratio, colpus length, colpus weight, exine, intine, tectum, nexine, columellea, and echinae length) were measured for the analysis. The data set was formed from 600 samples, obtained from 20 different taxa, with 30 different images. The 440 samples were used for training and the remaining 160 samples were used for testing. The proposed method, a rough set-based expert system, has properly identified 145 of 160 pollen grains correctly. The success of the method for the identification of pollen grains was obtained at 90.625% (145/160). We can expect to achieve more efficient results with different genuses and families, considering the successful results in the same genus. Moreover, using computer-based systems in revision studies will lead us to more accurate and efficient results, and will identify which characters will be more effective for pollen identification. According to the literature, this is the first study for the identification and comparison of pollen of the same genus by using the measurements of distinctive characteristics with computer systems.

Feasibility analysis of the value of Q method in the classification and understanding of expert experience

The Q method, combining qualitative and quantitative methods, refers to the qualitative analysis of Q-sorts based on quantitative techniques. It is used to research individual subjective experience, analyzing consensus and divergence to identify and categorize subjects' viewpoints. The sorting process is completely performed by the subjects, independent of study researchers. The Q method in medical research has been applied in many fields, including nursing care, clinical studies, doctor and patient's perceptions, health evaluation and decision making. The authors used the Q method to research Chinese medicine (CM) group decision making, exploring its practical feasibility in this important field. Four primary domains are addressed: (1) integration of expert opinion; (2) expert classification; (3) ascertaining the entire viewpoint orientation of a certain type of expert; and (4) comparison of expert opinion using an additional perspective. The essence of the Q method caters to the CM thinking model and should be introduced into CM and explored more deeply.

Security-level classification for confidential documents by using adaptive neuro-fuzzy inference systems

The security-level detection of a confidential document is a vital task for organizations to protect their confidential information. Diverse classification rules and techniques are being applied by human experts. Increasing number of confidential information in organizations is making difficult to classify all the documents carefully with human effort. The recommended frameworks in this study classify the internal documents of TUBITAK UEKAE (National Research Institute of Electronics and Cryptology of Turkey) by using classification algorithms naive Bayes, support vector machines (SVMs) and adaptive neuro-fuzzy inference systems (ANFISs). A hybrid approach involving support vector classifiers and adaptive neuro-fuzzy classifiers exposes the most successful accuracy rates of expert system classification. This study also states preprocessing tasks required for document classification with natural language processing. To represent term–document relations, a recommended metric TF-IDF was chosen to construct a weight matrix. Agglutinative nature of Turkish documents is handled by Turkish stemming algorithms. At the end of the article, some experimental results and success metrics are projected with accuracy rates and receiver operating characteristic (ROC) curves.

Morphometric Analysis and Classification of Glaucomatous Optic Neuropathy Using Radial Polynomials

To quantify the morphological features of the optic nerve head using radial polynomials, to use these morphometric models as the basis for classification of glaucomatous optic neuropathy via an automated decision tree induction algorithm, and to compare these classification results with established procedures. A cohort of patients with high-risk ocular hypertension or early glaucoma (n=179) and a second cohort of normal subjects (n=96) were evaluated for glaucomatous optic neuropathy using stereographic disc photography and confocal scanning laser tomography. Morphological features of the optic nerve head region were modeled from the tomography data using pseudo-Zernike radial polynomials and features derived from these models were used as the basis for classification by a decision tree induction algorithm. Decision tree classification performance was compared with expert classification of stereographic disc photographs and analysis of neural retinal rim thickness by Moorfields Regression Analysis (MRA). Root mean squared error of the morphometric models decreased asymptotically with additional polynomial coefficients, from 62±0.5 (32 coefficients) to 32±5.7 μm (256 coefficients). Optimal morphometric classification was derived from a subset of 64 total features and had low sensitivity (69%), high specificity (88%), very good accuracy (80%), and area under the receiver operating characteristic curve (AUROC) was 88% (95% confidence interval, 78%-98%). In comparison, MRA classification of the same records had a comparatively poorer sensitivity (55%), but had higher specificity (95%), with similar overall accuracy (78%) and AUROC curve, 83% (95% CI, 70%-96%). Pseudo-Zernike radial polynomials provide a mathematically compact and faithful morphological representation of the structural features of the optic nerve head. This morphometric method of glaucomatous optic neuropathy classification has greater sensitivity, and similar overall classification performance (AUROC) when compared with classification by neural retinal rim thickness by MRA in patients with high-risk ocular hypertension and early glaucoma.

Driver’s head pose estimation using a hierarchical classification on an effective feature space

This paper introduces a direct accountability technique for the real-time estimation of the head’s position and orientation. Unlike existing works which rely on feature extraction either in the image domain or in three-dimensional space, our proposed approach based on a hierarchical expert classification will estimate the head pose for real-time applications, specifically for driver assistance systems. We proposed a method for feature extraction which aims at keeping only the angle-related features, independent of identity. We then used a hierarchical algorithm for classification which divides the whole space into smaller spaces. Moreover, we made a performance study aiming at evaluating the accuracy of the proposed approach. Experimental results show that the proposed features are able to describe the pose of the driver face image successfully, with a performance in agreement with the actual poses. This approach is not only useful for automobile safety and security but also can help severely handicapped people and can be used in collision prevention systems and other driving applications.

КОРЕЛЯЦІЯ БЕЗПЕКИ КЕРУВАННЯ ПОВІТРЯНИМ РУХОМ З РІВНЕМ МОВНОЇ ФАХОВОЇ ПІДГОТОВКИ АВІАДИСПЕТЧЕРІВ

Визначено індивідуальні психолінгвістичні можливості диспетчерів для виявленняпрофесійної придатності за допомогою комплексної експертної оцінки з класифікацієючасової затримки при мовній взаємодії в контурі «диспетчер–пілот».The subject of this article is the definition of individual possibilities of dispatchers and level oftheir professional suitability. A complex expert estimation with classification of an hour delay atlanguage interaction in a contour "dispatcher-pilot" is used for this research.Определены индивидуальные психолингвистические возможности диспетчеровдля выявления профессиональной пригодности с помощью комплексной экспертной оценки склассификацией временной задержки при языковом взаимодействии в контуре «диспетчер–пилот».