SEARCH · PubMed Health

Results for “Automated image segmentation”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

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

At least 379 records · Page 21Linked to original sources

PubMedDate not supplied

Automated brain segmentation from single slice, multislice, or whole-volume MR scans using prior knowledge.

PURPOSE: An automated procedure has been developed to isolate the brain in single/multislice or whole-volume MR images obtained from various sequences. METHOD: T1-weighted, T2-weighted, and inversion recovery images were acquired. The brain segmentation procedure employed (A) a knowledge base that held generic information about the brain in the three orthogonal views and (B) a texture definition and intensity characteristics of features within the head. The brain was segmented by selectively blurring scans using components of B; contour following with region growing was initiated until the isolated feature satisfied the measurements in A. RESULTS: The brain was segmented automatically from 210 subjects (whole volume) and 52 subjects (multi/single slice). Detailed analysis of seven segmented brains showed that < 0.8% of the contour pixels were erroneously identified. Whole-volume head scans consisting of 140 x 256 x 256 pixels were segmented in < 30 min. CONCLUSION: A robust, fast, and efficient procedure has been developed to segment the brain from MR images.

Adult↗

PubMed2001

Automated nasopharyngeal carcinoma detection with dynamic gadolinium-enhanced MR imaging.

OBJECTIVES: The purpose of this research is to develop an automatic medical diagnosis for segmenting nasopharyngeal carcinoma (NPC) with dynamic gadolinium-enhanced MR imaging. METHODS: This system is a multistage process, involving motion correction, head mask generation, dynamic MR data quantitative evaluation, rough segmentation, and rough segmentation refinement. Two approaches, a relative signal increase method and a slope method, are proposed for the quantitative evaluation of dynamic MR data. RESULTS: The NPC detection results obtained using the proposed methods had a rating of 85% in match percent compared with these lesions identified by an experienced radiologist. The match percent for the two proposed methods did not have significant differences. However, the computation cost for the slope method was about twelve times faster than the relative signal increase method. CONCLUSIONS: The proposed methods can identify the NPC regions quickly and effectively. This system can enhance the performance of clinical diagnosis.

Analysis of Variance↗

PubMed2006

Autofluorescence characteristics of early, atrophic, and high-risk fellow eyes in age-related macular degeneration.

PURPOSE: To assess the relationships of drusen, pigment, and focally increased autofluorescence (FIAF) and the reticular pattern of hypoautofluorescence, to distinguish the combined photographic and AF characteristics of early, atrophic, and high-risk fellow eyes in AMD. METHODS: In a retrospective interinstitutional clinical study, AF and color photograph pairs of 221 eyes were examined: 166 eyes of 83 patients with bilateral large, soft drusen, with and without geographic atrophy (GA), and 55 fellow eyes of 55 patients with unilateral choroidal neovascularization (CNV). Forty-two eyes (one eye from each of 42 patients with early or atrophic AMD) were divided into four groups: 14 with drusen only, 9 with drusen and pigment abnormalities, 11 fellow eyes of patients with unilateral GA, and 8 eyes of patients with bilateral GA (acronyms for the groups: D-D, D-Pig, D-GA and GA-GA, respectively). The 55 fellow eyes of patients with CNV were divided into three groups: 19 eyes with no FIAF (CNV-0), 16 with FIAF without reticular AF (CNV-1), and 20 eyes with reticular AF and/or pseudodrusen (CNV-R). Image pairs of eyes with FIAF were registered, and drusen, pigment, and FIAF were segmented using automated background leveling and thresholding. All 221 eyes were surveyed for reticular AF and reticular pseudodrusen. The main outcome measures were (1) the fraction and relative probability of FIAF colocalizing with drusen and pigment and (2) the presence or absence of reticular AF and reticular pseudodrusen. RESULTS: The mean fractions of FIAF that colocalized with large drusen were: D-D group, 0.46 +/- 0.21; D-Pig group, 0.42 +/- 0.29; D-GA group, 0.13 +/- 0.09; and GA-GA group, 0.11 +/- 0.12. Comparisons between groups showed significant differences when comparing either the D-D group or the D-Pig group with either the D-GA group or the GA-GA group (P between 0.0001 and 0.015), whereas other comparisons were nonsignificant (Mann-Whitney rank sum test). The mean probabilities of FIAF colocalizing with large drusen relative to chance (1.0) were: D-D group, 4.7 +/- 2.5; D-Pig group, 4.3 +/- 2.3; D-GA group, 1.4 +/- 0.8; and GA-GA group, 1.8 +/- 1.3, with similar significant differences as for the colocalization fractions. The mean probability of FIAF colocalizing with small to intermediate drusen in the D-D group was 1.5 +/- 1.3, which was not significantly different from chance. In the D-Pig group, the median probability of FIAF colocalizing with pigment abnormalities was 10.0 (range, 1.1-51.0). The AF patterns in 15 of 19 eyes in the CNV-0 group were normal; the remainder had nonreticular hypoautofluorescence only. In the CNV-1 group, the relations of FIAF with drusen and pigment were similar to those in the early AMD groups. CNV-R comprised 20 of 55 eyes in the CNV group, but reticular autofluorescence and/or pseudodrusen were found in only 14 of 166 eyes of the early and atrophic groups. Of the 34 total eyes with reticular AF or pseudodrusen, 28 had both, 4 had reticular AF only, and 2 had reticular pseudodrusen only. CONCLUSIONS: There are clear relationships between AF patterns and clinical AMD status. In early AMD, FIAF's colocalization with large, soft drusen and hyperpigmentation is several times greater than chance, suggesting linked disease processes. In advanced atrophic AMD, FIAF is found mostly adjacent to drusen and GA, suggesting that dispersal of drusen-associated lipofuscin is a marker of atrophic disease progression. In the neovascular case, a large group of fellow eyes have no FIAF abnormalities, suggesting that lipofuscin is not a major determinant of CNV. However, reticular hypoautofluorescence, consistent with widespread inflammatory damage to the RPE, appears to be a highly sensitive imaging marker for the disease that determines reticular pseudodrusen and is strongly associated with CNV.

Atrophy↗

PubMed1995

Automatic vessel segmentation and quantification of the rat aortic ring assay of angiogenesis.

BACKGROUND: Pharmacologic control of angiogenesis is a promising new approach to the treatment of a variety of pathologic conditions including cancer. The recently developed in vitro rat aortic ring model provides a simple, reproducible assay for discovering angiogenic agonists and antagonists. However, quantification of results in this assay is time consuming, tedious, and subjective, because it involves visual inspection of images and manually counting the newly formed microvessels extending from the cultured aortic ring. This report describes an automated image analysis-based procedure for quantification of this assay that overcomes these difficulties. EXPERIMENTAL DESIGN: The designed image processing algorithm segments the vessels from gray scale images. A high-pass filter is used, and the results are separated into nonvascular and vessel compartments based on object size and shape. Quantification relies on identification of vessels intersecting a closed transect set a fixed distance from the aortic ring. The number and the total area of these vessels are determined. The entire operation has been automated and packaged in an application called Vessels. RESULTS: The correlation between computer-determined vessel area/vessel number and visual microvessel count is high (r2 = 0.91 and r2 = 0.86, respectively). CONCLUSION: Vessels offers high-speed, fully automatic batch processing including production of a hard copy for documentation. The application runs on the Apple family of computers. On a Quadra 800, the application can process approximately 30 images/hour, which is approximately 2.5 times faster than manual quantification of this assay.

Animals↗

PubMed2003

Automated three-dimensional assessment of coronary artery anatomy with intravascular ultrasound scanning.

BACKGROUND: Angiography allows the definition of advanced, severe stages of coronary artery disease, but early atherosclerotic lesions, which do not lead to luminal stenosis, are not identified reliably. In contrast, intravascular ultrasound scanning allows the precise characterization and quantification of a wide range of atherosclerotic lesions, independent of the severity of luminal stenosis. METHODS: Three-dimensional (3-D) reconstruction of entire coronary segments is possible with the integration of sequential 2-dimensional tomographic images and allows volumetric analysis of coronary arteries. RESULTS: Automated systems able to recognize lumen and vessel borders and to display 3-D images are becoming available. CONCLUSION: These systems have the potential for on-line 3-D image reconstruction for clinical decision-making and fast routine volumetric analysis in research studies. This review describes 3-D intravascular ultrasound scanning acquisition, analysis, and processing, and the associated technical challenges.

Coronary Artery Disease↗

PubMed1996

Application of textural features to the detection of ocular melanomas in scintigraphy.

A method for automated detection of ocular melanoma in scintigraphic images is described. The algorithm first performs an automatic segmentation of the eyes on specific reference images. The images of the eyes are then analyzed using textural parameters computed in several directions and averaged to damp directional information. Assuming that only one eye per patient will be pathological, the ratio of the textural parameters of the two eyes is computed. A statistical analysis is performed over these ratios to select the most highly discriminating textural parameters and detect the pathological patients. The method has been tested successfully on a population of 23 individuals and we found significant differences between pathological and normal patients.

Algorithms↗

PubMed1989

Evaluation of automated estimation of epithelial volume and its prognostic value in ovarian tumors.

The paper describes an improved segmentation method to measure the percentages of epithelium and stroma in ovarian (tumor) tissue with automated image analysis and evaluates its prognostic value. In the image processing method, a blue-yellow image pair is recorded from standard paraffin sections and stained with pararosanilin Feulgen and naphthol yellow. The blue image is used for automated determination of the total tissue area and the yellow image for the epithelial area. Results are obtained with 114 ovarian tumors of the common epithelial types (14 borderline tumors and 100 invasive carcinomas with varying degrees of differentiation). The fraction of epithelium in the total tissue shows a strong correlation with the epithelial percentage resulting from interactive morphometry (r = 0.991) for 15 tumors of varying histological grades. The prognostic value is evaluated on the 100 invasive carcinomas. Survival analysis implies that the epithelial percentage is of prognostic importance (Mantel-Cox 7.4, p = 0.0064). Multivariate analysis shows that the estimated fraction of epithelium is the strongest factor and that the FIGO stage has additional prognostic value (Mantel-Cox 12.5, p = 0.0004). It can be concluded that epithelial volume, as automatically estimated by image processing, seems useful in predicting the prognosis of patients with ovarian cancers.

Coloring Agents↗

PubMed1998

Registration of MR and SPECT without using external fiducial markers.

The aim of our work is to present, test and validate an automated registration method used for matching brain SPECT scans with corresponding MR scans. The method was applied on a data set consisting of ten brain IDEX SPECT scans and ten T1- and T2-weighted MR scans of the same subjects. Of two subjects a CT scan was also made. (Semi-) automated algorithms were used to extract the brain from the MR, CT and SPECT images. Next, a surface registration technique called chamfer matching was used to match the segmented brains. A perturbation study was performed to determine the sensitivity of the matching results to the choice of the starting values. Furthermore, the SPECT segmentation threshold was varied to study its effect on the resulting parameters and a comparison between the use of MR T1- and T2-weighted images was made. Finally, the two sets of CT scans were used to estimate the accuracy by matching MR to CT and comparing the MR-SPECT match to the SPECT-CT match. The perturbation study showed that for initial perturbations up to 6 cm the algorithm fails in less than 4% of the cases. A variation of the SPECT segmentation threshold over a realistic range (25%) caused an average variation in the optimal match of 0.28 cm vector length. When T2 is used instead of T1 the stability of the algorithm is comparable but the results are less realistic due the large deformations. Finally, a comparison of the direct SPECT-MR match and the indirect match with CT as intermediate yields a discrepancy of 0.4 cm vector length. We conclude that the accuracy of our automatic matching algorithm for SPECT and MR, in which no external markers were used, is comparable to the accuracies reported in the literature for non-automatic methods or methods based on external markers. The proposed method is efficient and insensitive to small variations in SPECT segmentation.

Automation↗

PubMed2004

3D-catFISH: a system for automated quantitative three-dimensional compartmental analysis of temporal gene transcription activity imaged by fluorescence in situ hybridization.

Fluorescence in situ hybridization (FISH) of neural activity-regulated, immediate-early gene (IEG) expression provides a method of functional brain imaging with cellular resolution. This enables the identification, in one brain, of which specific principal neurons were active during each of two distinct behavioral epochs. The unprecedented potential of this differential method for large-scale analysis of functional neural circuits is limited, however, by the time-intensive nature of manual image analysis. A comprehensive software tool for processing three-dimensional, multi-spectral confocal image stacks is described which supports the automation of this analysis. Nuclei counterstained with conventional DNA dyes and FISH signals indicating the sub-cellular distribution of specific, IEG RNA species are imaged using different spectral channels. The DNA channel data are segmented into individual nuclei by a three-dimensional multi-step algorithm that corrects for depth-dependent attenuation, non-isotropic voxels, and imaging noise. Intra-nuclear and cytoplasmic FISH signals are associated spatially with the nuclear segmentation results to generate a detailed tabular/database and graphic representation. Here we present a comprehensive validation of data generated by the automated software against manual quantification by human experts on hippocampal and parietal cortical regions (96.5% concordance with multi-expert consensus). The high degree of reliability and accuracy suggests that the software will generalize well to multiple brain areas and eventually to large-scale brain analysis.

Algorithms↗

PubMed1994

Automated interpretation of planar thallium-201-dipyridamole stress-redistribution scintigrams using artificial neural networks.

UNLABELLED: To develop an automated image interpretation system of planar cardiac 201Tl dipyridamole stress/redistribution scintigrams, the authors used artificial neural networks that associate patterns of segmental myocardial thallium uptake with a diagnostic assessment about the presence, severity and localization of significant coronary artery disease. METHODS: Artificial neural networks were trained and evaluated using the results from segmental thallium analysis and either expert readings in 159 cases or coronary angiography in a subgroup of 81 patients. RESULTS: Based on receiver operating characteristics analysis, the sensitivity for the detection of significant coronary artery disease at a specificity of 90% was 51% compared with angiography and 72% compared with the human expert. For severity and localization of disease, two vascular territories assigned to the vascular bed of the left anterior descending (LAD) artery and to the territory subtended by the left circumflex artery and the right coronary artery together (CX/RCA) were included in the analysis. CONCLUSION: Artificial neural networks may be useful to develop automated computer-based image interpretation systems of 201Tl perfusion scintigrams. However, utilization of large training datasets appears to be a prerequisite to achieve adequate diagnostic performance.

Adult↗

PubMed2004

Intra- and interstudy reproducibility of coronary artery diameter measurements in magnetic resonance coronary angiography.

PURPOSE: To determine the intra- and interstudy reproducibility of right coronary artery diameter assessment using serial magnetic resonance (MR) coronary angiography. MATERIALS AND METHODS: Two-dimensional (2D) navigator-gated segmented fast low angle shot (FLASH) images of the proximal right coronary artery were acquired three times in 11 healthy volunteers, the first two times in the same study session and the third time after repositioning the subject in the scanner. Coronary artery diameters were determined using automated segmentation software and intra- and interstudy reproducibility calculated as the standard deviation (SD) of the signed differences between measurements within and between study sessions, respectively. The reproducibility of the segmentation software was determined by repeated analysis of each individual scan. RESULTS: One subject was excluded from the study due to poor-quality images. In the remaining 10 subjects, the mean (+/- SD) intrastudy difference in coronary artery diameters was -0.05 +/- 0.12 mm, a value that is very similar to between-frame (same-film) differences reported in quantitative coronary angiography (QCA). The mean (+/- SD) interstudy difference in coronary artery diameters was 0.16 +/- 0.43 mm, although this was greatly skewed by one subject with poor image plane repositioning. Excluding that subject resulted in a mean (+/- SD) interstudy difference of 0.04 +/- 0.20 mm. The reproducibility of the segmentation software was excellent, with the mean difference between repeat analyses of the images being 0.00 +/- 0.03 mm. CONCLUSION: The intrastudy variability of coronary artery diameter measurements is low, potentially allowing MR coronary angiography to be used as a tool for the noninvasive assessment of serial changes following pharmacological intervention. A major contributing factor to this is the high reproducibility of the segmentation software. Interstudy variability is approximately three times the intrastudy variability.

Adult↗

PubMed1999

Quantitative measurements in IVUS images.

IntraVascular UltraSound (IVUS) is a catheter-based technique which provides real-time high resolution tomographic images of both the lumen and arterial wall of a coronary segment, this in contrast to X-ray arteriography that provides a shadow image (luminogram) of the entire lumen. Nowadays the lumen and vessel parameters are measured manually, which is very time consuming and suffers from high inter- and intra-obser variability. With the continuing improvement in IVUS imaging, it is now feasible to develop and clinically apply automated methods of three-dimensional quantitative analysis of the coronary vessel morphology in an objective and reproducible way with automated contour detection techniques (QCU). Quantification, in 2D and 3D, as well as volume rendering for visualization of the IVUS images requires segmentation of the images (contour detection). The 3D contour detection system described in this article is based on the combination of contour detection in the transversal and sagital view. This article provides some of the basic principles of IVUS, the IVUS image quantification, the three-dimensional reconstruction and the contour detection and quantification in three-dimensional IVUS images.

Coronary Disease↗

PubMed2006

Semi-automated segmentation and quantification of adipose tissue in calf and thigh by MRI: a preliminary study in patients with monogenic metabolic syndrome.

BACKGROUND: With the growing prevalence of obesity and metabolic syndrome, reliable quantitative imaging methods for adipose tissue are required. Monogenic forms of the metabolic syndrome include Dunnigan-variety familial partial lipodystrophy subtypes 2 and 3 (FPLD2 and FPLD3), which are characterized by the loss of subcutaneous fat in the extremities. Through magnetic resonance imaging (MRI) of FPLD patients, we have developed a method of quantifying the core FPLD anthropometric phenotype, namely adipose tissue in the mid-calf and mid-thigh regions. METHODS: Four female subjects, including an FPLD2 subject (LMNA R482Q), an FPLD3 subject (PPARG F388L), and two control subjects were selected for MRI and analysis. MRI scans of subjects were performed on a 1.5T GE MR Medical system, with 17 transaxial slices comprising a 51 mm section obtained in both the mid-calf and mid-thigh regions. Using ImageJ 1.34 n software, analysis of raw MR images involved the creation of a connectedness map of the subcutaneous adipose tissue contours within the lower limb segment from a user-defined seed point. Quantification of the adipose tissue was then obtained after thresholding the connected map and counting the voxels (volumetric pixels) present within the specified region. RESULTS: MR images revealed significant differences in the amounts of subcutaneous adipose tissue in lower limb segments of FPLD3 and FPLD2 subjects: respectively, mid-calf, 15.5% and 0%, and mid-thigh, 25.0% and 13.3%. In comparison, old and young healthy controls had values, respectively, of mid-calf, 32.5% and 26.2%, and mid-thigh, 52.2% and 36.1%. The FPLD2 patient had significantly reduced subcutaneous adipose tissue compared to FPLD3 patient. CONCLUSION: Thus, semi-automated quantification of adipose tissue of the lower extremity can detect differences between individuals of various lipodystrophy genotypes and represents a potentially useful tool for extended quantitative phenotypic analysis of other genetic metabolic disorders.

Journal Article↗

PubMed1999

Quantification of eosinophilic granule protein deposition in biopsies of inflammatory skin diseases by automated image analysis of highly sensitive immunostaining.

Eosinophilic granulocytes are major effector cells in inflammation. Extracellular deposition of toxic eosinophilic granule proteins (EGPs), but not the presence of intact eosinophils, is crucial for their functional effect in situ. As even recent morphometric approaches to quantify the involvement of eosinophils in inflammation have been only based on cell counting, we developed a new method for the cell-independent quantification of EGPs by image analysis of immunostaining. Highly sensitive, automated immunohistochemistry was done on paraffin sections of inflammatory skin diseases with 4 different primary antibodies against EGPs. Image analysis of immunostaining was performed by colour translation, linear combination and automated thresholding. Using strictly standardized protocols, the assay was proven to be specific and accurate concerning segmentation in 8916 fields of 520 sections, well reproducible in repeated measurements and reliable over 16 weeks observation time. The method may be valuable for the cell-independent segmentation of immunostaining in other applications as well.

Automation↗

PubMed1993

Accuracy and precision of quantitative digital coronary arteriography: observer-, short-, and medium-term variabilities.

Coronary arteriograms are increasingly acquired and stored in digital format, which allows instantaneous review of the pictorial data during the cardiac catheterization procedure. To support the angiographer in choosing the optimal sizes of the recanalization devices and studying the efficacy of the recanalization procedures, we have developed a new analytical software package (Automated Coronary Analysis = ACA) on the Philips DCI (-SX) digital cardiac imaging system. The ACA-package allows the objective and reproducible assessment of the morphologic and functional severity of coronary obstructions. Required user interaction is limited to the definition of the start and end points of the coronary segment to be analyzed. Automated contour detection is based on the use of first and second derivative functions along scanlines perpendicular to the automatically computed vessel pathline in the first iteration and perpendicular to the initial contours in the second iteration. These derivative functions have been modified based on the line spread function of the X-ray imaging chain, which is of particular importance for the accurate measurement of small vessel sizes. Phantom studies have indeed demonstrated that vessel sizes down to 0.66 mm can be measured accurately and reproducibly. Inter- and intraobserver variability studies have demonstrated a variability in the obstruction diameter of 0.11 mm and 0.10 mm, respectively, and in the percent diameter stenosis of 5.64% and 3.18%, respectively. These variability studies have been extended to short-term studies with repeated acquisition in the same angiographic views after 5 min and to medium-term studies with repeated acquisition in the initial angiographic views at the end of the catheterization procedures. With these standardized repeated acquisition and analysis procedures, the variabilities in the obstruction diameters increased to 0.19 and 0.18 mm, respectively, and remained below 6% in the percent diameter stenosis (5.61% and 5.28%, respectively). With an analysis time of approximately 15 sec on the DCI-SX, an efficient tool is now available in the catheterization laboratory for the objective and reproducible assessment of vessel dimensions and changes therein as a result of recanalization procedures.

Coronary Angiography↗

PubMed2006

Neural consequences of sleep disordered breathing: the role of intermittent hypoxia.

Sleep disordered breathing is characterised by periodic breathing, episodes of hypoxia and repeated arousals from sleep; symptoms include excessive daytime sleepiness, impairment of memory, learning and attention. Recent evidence from animal studies suggests that both intermittent hypoxia and sleep fragmentation can independently lead to neuronal defects in the hippocampus and pre frontal cortex; areas known to be closely associated with neural processing of memory and executive function. We have previously shown that sleep disordered breathing is associated with loss of gray matter concentration within the left hippocampus (47). We have now confirmed and extended this finding in 22 right handed, newly diagnosed male patients (mean (sd): age 51.8 (15.4) yrs, apnea/hypopnea index 53.1 (14.0) events/hr, minimum nocturnal oxygen saturation 75 (8.4) %) and 17 controls matched for age and handedness. Voxel-based morphometry, an automated unbiased technique, was used to characterise changes in gray matter concentration. The magnetic resonance images were segmented and grey matter concentration determined voxel by voxel. Analysis of variance was then preformed, adjusted for overall image intensity, with age as a covariant. Additional to the deficit in the left hippocampus, we found more extensive loss of gray matter bilaterally in the parahippocampus. No additional focal lesions were seen in other brain regions. Based on our findings and data from other human and animal studies, we speculate that in patients with sleep disordered breathing intermittent hypoxia is associated with neural deficit, and further that such lesions may lead to cognitive dysfunction.

Adult↗

PubMedDate not supplied

Normalized accurate measurement of longitudinal brain change.

PURPOSE: Quantitative measurement of change in brain size and shape (e.g., to estimate atrophy) is an important current area of research. New methods of change analysis attempt to improve robustness, accuracy, and extent of automation. A fully automated method has been developed that achieves high estimation accuracy. METHOD: A fully automated method of longitudinal change analysis is presented here, which automatically segments brain from nonbrain in each image, registers the two brain images while using estimated skull images to constrain scaling and skew, and finally estimates brain surface motion by tracking surface points to subvoxel accuracy. RESULTS AND CONCLUSION: The method described has been shown to be accurate ( approximately 0.2% brain volume change error) and to achieve high robustness (no failures in several hundred analyses over a range of different data sets).

Adult↗

PubMed1992

A processing strategy for automated Papanicolaou smear screening.

A multilayer processing strategy was developed for the automatic screening of conventionally prepared Papanicolaou smears. The processing stages include image segmentation, feature extraction, object classification and slide classification. Mathematical morphology functions were implemented in hardware with custom-built gate array processors for image segmentation. There were 68 features used for classifier training. In object classification we combined the evidential supports of a binary decision tree classifier and a multilayer perceptron classifier to achieve an integrated decision. In this feasibility study, 449 conventionally prepared cervical Papanicolaou smears were tested in a prototype research system between January and May 1991. The 95% confidence interval for the slide false-negative rate was 1-9%, and the 95% confidence interval for the slide sort rate was 45-55%. The estimated sort rate for clearly normal slides is within the range required for a cost-efficient screening system, and the estimated false-negative rate for premalignant and malignant smears is an improvement over published false-negative rates for human performance. Several performance improvement efforts are still under way. We expect that they will result in a vastly reduced slide false-negative rate.

Automation↗