Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Pattern Recognition, Automated”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 271 records · Page 15Linked to original sources
PubMed2005

Reliability assessment of an automated forced swim test device using two mouse strains.

The Porsolt forced swim test (FST) is one of the most widely used behavioral tests in the evaluation of the antidepressant effects of drugs. It is based on the fact that these drugs reduce the depression-related behaviors of learned helplessness. The model has been modified for use in mice. In contrast to rats, mice are exposed to forced swimming only once and their immobility behavior is measured and considered a "depression-like" phenotype. Like many other behavioral tests, FST can be affected by observer-related artifacts. In recent years, automated testing systems have been developed to decrease artifacts that may greatly influence the interpretation of results. In this work, we used two strains of mice, i.e., C3H/HeJ and C57BL/6J, which differ in their FST immobility times. We employed a new commercially available automated FST device and a blinded observer-based FST, and we examined their ability to measure behavioral differences between these two mouse strains. Our results suggest that the tested automated FST system generates reliable data comparable to results obtained by trained observers.

Algorithms↗
PubMed2006

CT colonography: automated measurement of colonic polyps compared with manual techniques--human in vitro study.

PURPOSE: To prospectively investigate the relative accuracy and reproducibility of manual and automated computer software measurements by using polyps of known size in a human colectomy specimen. MATERIALS AND METHODS: Institutional review board approval was obtained for the study; written consent for use of the surgical specimen was obtained. A colectomy specimen containing 27 polyps from a 16-year-old male patient with familial adenomatous polyposis was insufflated, submerged in a container with solution, and scanned at four-section multi-detector row computed tomography (CT). A histopathologist measured the maximum dimension of all polyps in the opened specimen. Digital photographs and line drawings were produced to aid CT-histologic measurement correlation. A novice (radiographic technician) and an experienced (radiologist) observer independently estimated polyp diameter with three methods: manual two-dimensional (2D) and manual three-dimensional (3D) measurement with software calipers and automated measurement with software (automatic). Data were analyzed with paired t tests and Bland-Altman limits of agreement. RESULTS: Seven polyps (<or=6-mm diameter) could not be extracted by using the software; 20 polyps (5-15-mm diameter) remained for analysis. Automated measurement was not significantly different from histologic size for the experienced reader (mean difference, 0.63 mm; P=.06) or novice reader (mean difference, 0.58 mm; P=.12). With manual 2D measurement and manual 3D measurement, the experienced reader (1.21-mm mean difference, P<.001, and 0.68-mm mean difference, P=.03, respectively) and novice reader (1.54-mm mean difference, P<.001, and 0.84-mm mean difference, P=.002, respectively) significantly underestimated polyp size. Interobserver agreement was good and similar for all three methods (95% limits of agreement span, approximately 2.5 mm). Intraobserver agreement was related to reader experience, with differences of up to 2.5 mm within expected limits of agreement. CONCLUSION: For polyps smaller than 1 cm, measurement differences of up to 2.5 mm are within the expected limits of inter- and intraobserver agreement for all measurement techniques. Automated and manual 3D polyp measurements are more accurate than manual 2D measurements.

Adolescent↗
PubMed2006

Automated vs. manual assessment of left ventricular function in cardiac multidetector row computed tomography: comparison with magnetic resonance imaging.

We compared semiautomatic contour detection and manual contour tracing in cardiac multidetector row computed tomography (MDCT) with magnetic resonance imaging (MRI) for calculation of left-ventricular (LV) volumes. The study included 30 patients who underwent contrast-enhanced cardiac MDCT and cardiac cine-MRI. Were calculated 8 mm short-axis slices from MDCT data using three-dimensional multiphase image reconstruction. LV volumes including peak ejection rate and peak filling rate were calculated from manually and semiautomatically determined contours. Results were compared to those from cine-MRI with manually drawn contours as the standard of reference. We found good agreement for the LV volumes, with an ejection fraction of 47.1+/-9.4% for manually drawn contours, 47.9+/-9.9% for semiautomatically detected contours on MDCT, and 48.0+/-10.2% for MRI. Except for peak-filling rate analysis of variance revealed no difference between any of these techniques. Bland-Altman plots and Lin's concordance correlation coefficient showed best agreement between MRI and manual contour tracing in MDCT. Calculation of LV volumes using either semiautomatic or manual contour tracing in cardiac MDCT is therefore feasible when compared to MRI. Automated contour detection needs to be improved to equal manual contour tracing.

Algorithms↗
PubMed2006

Model-based automated extraction of microtubules from electron tomography volume.

We propose a model-based automated approach to extracting microtubules from noisy electron tomography volume. Our approach consists of volume enhancement, microtubule localization, and boundary segmentation to exploit the unique geometric and photometric properties of microtubules. The enhancement starts with an anisotropic invariant wavelet transform to enhance the microtubules globally, followed by a three-dimensional (3-D) tube-enhancing filter based on Weingarten matrix to further accentuate the tubular structures locally. The enhancement ends with a modified coherence-enhancing diffusion to complete the interruptions along the microtubules. The microtubules are then localized with a centerline extraction algorithm adapted for tubular objects. To perform segmentation, we novelly modify and extend active shape model method. We first use 3-D local surface enhancement to characterize the microtubule boundary and improve shape searching by relating the boundary strength with the weight matrix of the searching error. We then integrate the active shape model with Kalman filtering to utilize the longitudinal smoothness along the microtubules. The segmentation improved in this way is robust against missing boundaries and outliers that are often present in the tomography volume. Experimental results demonstrate that our automated method produces results close to those by manual process and uses only a fraction of the time of the latter.

Algorithms↗
PubMed2002

Automated techniques for visualization and mapping of articular cartilage in MR images of the osteoarthritic knee: a base technique for the assessment of microdamage and submicro damage.

The purpose of this paper is to describe automated techniques for the visualization and mapping of articular cartilage in magnetic resonance (MR) images of the osteoarthritic knee. The MR sequences and analysis software which will be described allow the assessment of cartilage damage using a range of standard scanners. With high field strength systems it would be possible, using these techniques, to assess micro-damage. The specific aim of the paper is to develop and validate software for automated segmentation and thickness mapping of articular cartilage from three-dimensional (3-D) gradient-echo MR images of the knee. The method can also be used for MR-based assessment of tissue engineered grafts. Typical values of cartilage thickness over seven defined regions can be obtained in patients with osteoarthritis (OA) and control subjects without OA. Three groups of patients were studied. The first group comprised patients with moderate OA in the age range 45-73 years. The second group comprised asymptomatic volunteers of 50-65 years; the third group, younger volunteers selected by clinical interview, history and X-ray. In this paper, sagittal 3-D spoiled-gradient steady-state acquisition images were obtained using a 1.5-T GE whole-body scanner with a specialist knee coil. For validation bovine and porcine cadaveric knees were given artificial cartilage lesions and then imaged. The animal validations showed close agreement between direct lesion measurements and those obtained from the MR images. The feasibility of semi-automated segmentation is demonstrated. Regional cartilage thickness values are seen as having practical application for fully automated detection of OA lesions even down to the submicrometer level.

Aged↗
PubMed2003

Computerized scheme for determination of the likelihood measure of malignancy for pulmonary nodules on low-dose CT images.

An automated computerized scheme has been developed for determination of the likelihood measure of malignancy of pulmonary nodules on low-dose helical CT (LDCT) images. Our database consisted of 76 primary lung cancers (147 slices) and 413 benign nodules (576 slices). With this automated computerized scheme, the location of a nodule was first indicated by a radiologist. The outline of the nodule was segmented automatically by use of a dynamic programming technique. Various objective features on the nodules were determined by use of outline analysis and image analysis, and the likelihood measure of malignancy was determined by use of linear discriminant analysis (LDA). The effect of many different combinations of features and the performance of LDA in distinguishing benign nodules from malignant ones were evaluated by means of receiver operating characteristic (ROC) analysis. The Az value (area under the ROC curve) obtained by the computerized scheme in distinguishing benign nodules from malignant ones was 0.828 when a single slice was employed for each of the nodules. However, the Az value was improved to 0.846 when multiple slices were used for determination of the likelihood measure of malignancy. The Az values obtained by the computerized scheme on LDCT images were significantly greater than the Az value of 0.70, which was obtained from our previous observer studies by radiologists in distinguishing benign nodules from malignant ones on LDCT images. The automated computerized scheme for determination of the likelihood measure of malignancy would be useful in assisting radiologists to distinguish between benign and malignant pulmonary nodules on LDCT images.

Algorithms↗
PubMed1987

Automated interference refractometer: an algorithm for locating an irregular fringe pattern.

The interference refractometer is potentially a valuable instrument for the measurement of gas concentrations, but its usefulness has been limited by the necessity to locate visually a pattern of light and dark bands in order to obtain the reading. The instrument is therefore liable to human error and is unsuitable for continuous monitoring. An improved design has been patented, in which this location process is automated by the use of a microprocessor and an array of light-sensitive diodes. To implement this improvement it was necessary to design an algorithm which would reliably locate the pattern. This paper describes the approaches which were considered and the successful algorithm. Some examples are given to illustrate the power of the final algorithm to locate patterns even after severe distortion.

Algorithms↗
PubMed2006

Vertebral morphometry: semiautomatic determination of detailed shape from dual-energy X-ray absorptiometry images using active appearance models.

OBJECTIVES: Manual point placement for vertebral morphometry is time-consuming and imprecise. We evaluated the accuracy of semiautomatic computer determination of the detailed vertebral shape. MATERIALS AND METHODS: The shape and appearance of vertebrae on 250 lateral dual-energy x-ray absorptiometry (DXA) scans were statistically modeled using a sequence of active appearance models of vertebral triplets. The models were matched to unseen scans given an approximate initial location of the center of each vertebra. The segmentation accuracy was analyzed by fracture grade. RESULTS: Segmentation accuracy comparable to manual precision was obtained in the case of normal vertebrae, but the accuracy decreased with increasing fracture severity. We propose methods for improving the robustness for severe fractures. CONCLUSION: Vertebral morphometry measurements may be substantially automated even on noisy data with multiple fractures present. The shape and appearance parameters of the models could provide more powerful quantitative classifiers of osteoporotic vertebral fracture.

Absorptiometry, Photon↗
PubMed2005

Automatic detection and size quantification of infarcts by myocardial perfusion SPECT: clinical validation by delayed-enhancement MRI.

UNLABELLED: We aimed to validate the accuracy of a new automated myocardial perfusion SPECT quantification based on normal limits for detection and sizing of infarcts, using delayed-enhancement MRI (DE-MRI) as a gold standard. METHODS: Eighty-two immediate (201)Tl rest scans and 26 (201)Tl delayed redistribution scans were compared with resting DE-MRI scans acquired within 24 h of SPECT acquisition. The immediate (201)Tl scans were considered for validation of infarct detection and the delayed (201)Tl scans were considered for infarct sizing. A simplified quantification scheme was used in which defect extent (EXT) and total perfusion deficit (TPD) parameters were derived automatically from SPECT images by comparison with sex-matched normal limits and applying a 3.0 average deviation criterion. The total extent of hyperenhancement expressed as the percentage of the left ventricle was derived from DE-MRI images by visual definition of myocardial contours and defects. DE-MRI and SPECT images were fused in 3 dimensions for visual comparison. Phantom data were also quantified using the same EXT and TPD measures for defects ranging from 5% to 70% of the myocardium. RESULTS: The area under the receiver-operator-characteristic curve for the detection of infarct on immediate rest scans was 0.91 +/- 0.03 for EXT and 0.90 +/- 0.03 for TPD (P = not significant). The sensitivity and specificity for the detection of infarct by EXT on immediate (201)Tl rest scan were 87% and 91%, respectively, with the optimal defect size threshold of 4%. Six of 7 cases with DE-MRI defects < 5% were detected by SPECT. Infarct sizes obtained from DE-MRI correlated well with EXT (slope = 0.94, offset = 3.8%; r = 0.84) and TPD (slope = 0.75, offset = 4.2%; r = 0.85) obtained from delayed SPECT (201)Tl scans. Excellent correlation was observed between the SPECT quantification and the physical defect size for the phantom data. The actual size of the defect was better estimated by EXT (slope = 1.00, offset 1.33%; r = 0.99) than by TPD (slope = 0.79, offset = 1.9%; r = 0.99). CONCLUSION: Automated quantification of the EXT on myocardial perfusion SPECT images can reliably detect infarcts and measure infarct sizes.

Aged↗
PubMed1993

Kinematic analysis of human walking gait using digital image processing.

A system using digital image processing techniques for kinematic analysis of human gait has been developed. The system is cheap, easy to use, automated and provides useful detailed quantitative information to the medical profession. Passive markers comprising black annuli on white card are placed on the anatomical landmarks of the subject. Digital images at the standard television rate of 25 per second are acquired of the subject walking past a white background. The images are obtained, stored and processed using standard commercially available hardware, i.e. video camera, video recorder, digital framestore and an IBM PC. Using a single-threshold grey level, all the images are thresholded to produce binary images. An automatic routine then uses a set of pattern recognition algorithms to locate accurately and consistently the markers in each image. The positions of the markers are analysed to determine to which anatomical landmark they correspond, and thus a stick diagram for each image is obtained. There is also a facility where the positions of the markers may be entered manually and errors corrected. The results may be presented in a variety of ways: stick diagram animation, sagittal displacement graphs, flexion diagrams and gait parameters.

Algorithms↗
PubMed2005

Global left ventricular function in cardiac CT. Evaluation of an automated 3D region-growing segmentation algorithm.

The purpose was to evaluate a new semi-automated 3D region-growing segmentation algorithm for functional analysis of the left ventricle in multislice CT (MSCT) of the heart. Twenty patients underwent contrast-enhanced MSCT of the heart (collimation 16 x 0.75 mm; 120 kV; 550 mAseff). Multiphase image reconstructions with 1-mm axial slices and 8-mm short-axis slices were performed. Left ventricular volume measurements (end-diastolic volume, end-systolic volume, ejection fraction and stroke volume) from manually drawn endocardial contours in the short axis slices were compared to semi-automated region-growing segmentation of the left ventricle from the 1-mm axial slices. The post-processing-time for both methods was recorded. Applying the new region-growing algorithm in 13/20 patients (65%), proper segmentation of the left ventricle was feasible. In these patients, the signal-to-noise ratio was higher than in the remaining patients (3.2+/-1.0 vs. 2.6+/-0.6). Volume measurements of both segmentation algorithms showed an excellent correlation (all P<or=0.0001); the limits of agreement for the ejection fraction were 2.3+/-8.3 ml. In the patients with proper segmentation the mean post-processing time using the region-growing algorithm was diminished by 44.2%. On the basis of a good contrast-enhanced data set, a left ventricular volume analysis using the new semi-automated region-growing segmentation algorithm is technically feasible, accurate and more time-effective.

Aged↗
PubMed2004

Extracting human protein interactions from MEDLINE using a full-sentence parser.

MOTIVATION: The living cell is a complex machine that depends on the proper functioning of its numerous parts, including proteins. Understanding protein functions and how they modify and regulate each other is the next great challenge for life-sciences researchers. The collective knowledge about protein functions and pathways is scattered throughout numerous publications in scientific journals. Bringing the relevant information together becomes a bottleneck in a research and discovery process. The volume of such information grows exponentially, which renders manual curation impractical. As a viable alternative, automated literature processing tools could be employed to extract and organize biological data into a knowledge base, making it amenable to computational analysis and data mining. RESULTS: We present MedScan, a completely automated natural language processing-based information extraction system. We have used MedScan to extract 2976 interactions between human proteins from MEDLINE abstracts dated after 1988. The precision of the extracted information was found to be 91%. Comparison with the existing protein interaction databases BIND and DIP revealed that 96% of extracted information is novel. The recall rate of MedScan was found to be 21%. Additional experiments with MedScan suggest that MEDLINE is a unique source of diverse protein function information, which can be extracted in a completely automated way with a reasonably high precision. Further directions of the MedScan technology improvement are discussed. AVAILABILITY: MedScan is available for commercial licensing from Ariadne Genomics, Inc.

Abstracting and Indexing↗
PubMed2006

A new computer-based decision-support system for the interpretation of bone scans.

OBJECTIVE: To develop a completely automated method, based on image processing techniques and artificial neural networks, for the interpretation of bone scans regarding the presence or absence of metastases. METHODS: A total of 200 patients, all of whom had the diagnosis of breast or prostate cancer and had undergone bone scintigraphy, were studied retrospectively. Whole-body images, anterior and posterior, were obtained after injection of 99mTc-methylene diphosphonate. The study material was randomly divided into a training group and a test group, with 100 patients in each group. The training group was used in the process of developing the image analysis techniques and to train the artificial neural networks. The test group was used to evaluate the automated method. The image processing techniques included algorithms for segmentation of the head, chest, spine, pelvis and bladder, automatic thresholding and detection of hot spots. Fourteen features from each examination were used as input to artificial neural networks trained to classify the images. The interpretations by an experienced physician were used as the 'gold standard'. RESULTS: The automated method correctly identified 28 of the 31 patients with metastases in the test group, i.e., a sensitivity of 90%. A false positive classification of metastases was made in 18 of the 69 patients not classified as having metastases by the experienced physician, resulting in a specificity of 74%. CONCLUSION: A completely automated method can be used to detect metastases in bone scans. Future developments in this field may lead to clinically valuable decision-support tools.

Adult↗
PubMed2006

Automated lung segmentation of diseased and artifact-corrupted magnetic resonance sections.

Segmentation of the lungs within magnetic resonance (MR) scans is a necessary step in the computer-based analysis of thoracic MR images. This process is often confounded by image acquisition artifacts and disease-induced morphological deformation. We have developed an automated method for lung segmentation that is insensitive to these complications. The automated method was applied to 23 thoracic MR scans (413 sections) obtained from 10 patients. Two radiologists manually outlined the lung regions in a random sample of 101 sections (n=202 lungs), and the extent to which disease or artifact confounded lung border visualization was evaluated. Accuracy of lung regions extracted by the automated segmentation method was quantified by comparison with the radiologist-defined lung regions using an area overlap measure (AOM) that ranged from 0 (disjoint lung regions) to 1 (complete overlap). The AOM between each observer and the automated method was 0.82 when averaged over all lungs. The average AOM in the lung bases, where lung segmentation is most difficult, was 0.73.

Aged↗
PubMed1994

Neural network approach in multichannel auditory event-related potential analysis.

Even though there are presently no clearly defined criteria for the assessment of P300 event-related potential (ERP) abnormality, it is strongly indicated through statistical analysis that such criteria exist for classifying control subjects and patients with diseases resulting in neuropsychological impairment such as multiple sclerosis (MS). We have demonstrated the feasibility of artificial neural network (ANN) methods in classifying ERP waveforms measured at a single channel (Cz) from control subjects and MS patients. In this paper, we report the results of multichannel ERP analysis and a modified network analysis methodology to enhance automation of the classification rule extraction process. The proposed methodology significantly reduces the work of statistical analysis. It also helps to standardize the criteria of P300 ERP assessment and facilitate the computer-aided analysis on neuropsychological functions.

Algorithms↗
PubMed2006

A region-of-interest template for three-dimensional stereotactic surface projection images: initial application to the analysis of Alzheimer's disease and mild cognitive impairment.

OBJECTIVE: To construct a region-of-interest (ROI) template for Z-score images of three-dimensional stereotactic surface projections (3-D SSP) and to assess whether the ROI template can be a useful tool for evaluation of brain perfusion abnormalities of neurological disorders. MATERIALS AND METHODS: We constructed the ROI template for Z-score images of 3-D SSP based on the standardized magnetic resonance imaging data of 10 healthy volunteers. We assigned a total of 26 ROIs to Z-score images and superimposed it on Z-score images constructed from the brain perfusion SPECT data of 15 patients with Alzheimer's disease and 10 patients with mild cognitive impairment (MCI) who developed Alzheimer's disease within the following 2 years. We then obtained the mean Z-scores of each ROI and examined them to determine whether the hypoperfusion typical of Alzheimer's disease had been demonstrated quantitatively. We also visually inspected the Z-score image of each patient in both groups to determine whether the areas with the highest Z-scores were demonstrated within the ROIs of regions typical of Alzheimer's disease. RESULTS: In the patients with Alzheimer's disease, our ROI template quantitatively demonstrated hypoperfusion in regions typical of the disease and the Z-scores were very high. In the MCI patients, the mean Z-scores of the ROI in the posterior cingulated gyrus were the highest among all regions. Visual inspection of the Z-score images of each patient in both groups confirmed that the areas with the highest Z-scores were demonstrated within the ROIs in regions typical of Alzheimer's disease in all cases. CONCLUSION: Use of 3-D SSP methods and our ROI template enables automated quantitative evaluation of brain function images over the entire brain surface. In addition, the ROI template may facilitate visual interpretation of functional images of individual patients with neurological disorders.

Adult↗
PubMed2005

The use of multivariate MR imaging intensities versus metabolic data from MR spectroscopic imaging for brain tumour classification.

This study investigated the value of information from both magnetic resonance imaging and magnetic resonance spectroscopic imaging (MRSI) to automated discrimination of brain tumours. The influence of imaging intensities and metabolic data was tested by comparing the use of MR spectra from MRSI, MR imaging intensities, peak integration values obtained from the MR spectra and a combination of the latter two. Three classification techniques were objectively compared: linear discriminant analysis, least squares support vector machines (LS-SVM) with a linear kernel as linear techniques and LS-SVM with radial basis function kernel as a nonlinear technique. Classifiers were evaluated over 100 stratified random splittings of the dataset into training and test sets. The area under the receiver operating characteristic (ROC) curve (AUC) was used as a global performance measure on test data. In general, all techniques obtained a high performance when using peak integration values with or without MR imaging intensities. For example for low- versus high-grade tumours, low- versus high-grade gliomas and gliomas versus meningiomas, the mean test AUC was higher than 0.91, 0.94, and 0.99, respectively, when both MR imaging intensities and peak integration values were used. The use of metabolic data from MRSI significantly improved automated classification of brain tumour types compared to the use of MR imaging intensities solely.

Brain Chemistry↗