Localization of cutaneous lesions in digital images

Digital imaging could potentially provide a rapid, objective, and quantitative means of detecting changes in important skin conditions, especially the dysplasic nevus syndrome. Image analysis techniques can be applied to digital images to automate the search for changes in moles or other features. Consistent determination of lesion boundaries, perimeter, and area in digital images is a vital first step in this process. In this paper, we show how bilaterally symmetric Laplacian-of-a-Gaussian filters can be used to recover the borders of selected lesions while remaining robust with respect to factors such as the camera point spread function and additive noise. Tests on real and synthetic images demonstrate that lesion borders, area, and perimeter can be obtained with a high degree of reliability. Boundaries are routinely found to within +/- 0.2 pixels, and area and perimeter measurements vary by less than 10% when imaging spot targets and actual cutaneous lesions under a realistic range of experimental conditions.     

True 3‐D scanned voxel displays using single or multiple light sources

Abstract— Conventional stereoscopic displays require viewers to unnaturally keep eye accommodation fixed at one focal distance while they dynamically change vergence to view objects at different distances. This forced decoupling of reflexively linked processes fatigues eyes, causes discomfort, compromises image quality, and may lead to pathologies in developing visual systems. Volumetric displays can overcome this conflict, but only for small objects placed within a limited range of viewing distances and accommodation levels, and cannot render occlusion cues correctly. Our multi‐planar True 3‐D displays generate accommodation cues that match vergence and stereoscopic retinal disparity demands and can display images and objects at viewing distances throughout the full range of human accommodation (from 6.25 cm to infinity), better mimicking natural vision and minimizing eye fatigue.     

Preparation of calcium aluminate matrix composites by combustion synthesis

CaO–Al₂O₃–TiB₂ composites have been produced by the Combustion Synthesis technique. These materials have matrices based on binary calcium-aluminate compounds, i.e., Ca₃Al₂O₆ (C₃A), Ca₁₂Al₁₄O₃₃ (C₁₂A₇), CaAl₂O₄ (CA), CaAl₄O₇ (CA₂) and CaAl₁₂O₁₉ (CA₆). Except for samples with the matrix composition of C₃A, the combustion synthesis reactions can be characterized as stable self-propagating waves with combustion temperatures ranging from 2125 K to 2717 K and combustion wave velocity from 4.0 mm/s to 10.6 mm/s. For samples with a matrix composition of C₁₂A₇, CA, and CA₂, predominantly equilibrium compound phase was formed, while for samples with a matrix composition of C₃A, non-equilibrium phases were also present. There was no evidence of CA₆ formation for samples with a matrix composition corresponding to CA₆.     

Improved estimation of fraction images using partial imagerestoration

It is noted that mixed pixels are unavoidable in remote sensing imagery, but signature mixing models can be effective in improving the usefulness of the image data. Unmixing error caused by sensor system point spread function (PSF) degradation is studied, and the application of image restoration before spectral unmixing to improve the unmixing results is investigated. It is shown that the characteristics of the problem makes the design of a restoration filter differ from that for visual enhancement. Simulation results show that, depending on the scene complexity, unmixing error can be reduced by up to 72% using the proposed restoration scheme. It is concluded that image restoration is useful in increasing the accuracy of spectral unmixing     

Association of parvovirus B19 genome in children with myocarditis and cardiac allograft rejection: diagnosis using the polymerase chain reaction

BACKGROUND: Inflammatory diseases of the heart, including myocarditis and cardiac transplant rejection, are important causes of morbidity and mortality in children. Although viral infection may be suspected in either of these clinical conditions, the definitive etiology is often difficult to ascertain. Furthermore, the histology is identical for both disorders. Coxsackievirus has long been considered the most common cause of viral myocarditis; however, we previously demonstrated by polymerase chain reaction (PCR) analysis that many different, and sometimes unexpected, viruses may be responsible for myocarditis and cardiac rejection. In this study, we describe the association of parvovirus genome identified through PCR analysis of cardiac tissue in the clinical setting of myocarditis and cardiac allograft rejection. METHODS AND RESULTS: Myocardial tissue from endomyocardial biopsy, explant, or autopsy was analyzed for parvovirus B19 using primers designed to amplify a 699-base pair PCR product from the VP1 gene region. Samples tested included those obtained from patients with suspected myocarditis (n=360) or transplant rejection (n=200) or control subjects (n=250). Parvoviral genome was identified through PCR in 9 patients (3 myocarditis; 6 transplant) and no control patients. Of the 3 patients with myocarditis, 1 presented with cardiac arrest leading to death, 1 developed dilated cardiomyopathy, and the other gradually improved. Four of the 6 transplant patients had evidence of significant rejection on the basis of endomyocardial biopsy histology. All transplant patients survived the infection. CONCLUSIONS: Parvovirus is associated with myocarditis in a small percentage of children and may be a potential contributor to cardiac transplant rejection. PCR may provide a rapid and sensitive method of diagnosis.     

Infection after pediatric heart transplantation: results of a multiinstitutional study. The Pediatric Heart Transplant Study Group

BACKGROUND: Detailed information regarding the spectrum and predictors of infection after heart transplantation in children is limited because of relatively small numbers of patients at any single institution. We therefore used combined data obtained from the Pediatric Heart Transplant Study Group to gain additional information regarding infectious complications in the pediatric population. METHODS: To determine the time-related risk of infection and death related to infection in a large pediatric patient population, we analyzed data related to 332 pediatric patients (undergoing heart transplantation between January 1, 1993, and December 31, 1994) from 22 institutions in the Pediatric Heart Transplant Study Group. RESULTS: Among the 332 total patients, 276 infections were identified in 136 patients. Of those patients with development of infection, a single infection episode was reported in 54% of patients, 21% had two infections, and 25% had three or more infections. Of the 276 infections, 164 (60%) were bacterial, 51 (18%) were due to cytomegalovirus, 35 (13%) were other viral (noncytomegalovirus) infections, 19 (7%) were fungal, and 7 (2%) were protozoal. Bacterial infections were more common in infants younger than 6 months of age at time of transplantation, comprising 73% of all infections as compared with 49% in patients older than 6 months of age. The incidence of bacterial infection peaked during the first month after transplantation, with the actuarial likelihood of a bacterial infection among all patients reaching 25% at 2 months. The most common sites of bacterial infection were blood and lung (74% of bacterial infections). Cytomegalovirus accounted for 59% of viral infections, with a peak hazard occurring at 2 months after transplantation. Among all infections, cytomegalovirus was less common in infants younger than 6 months of age (8% of all infections) than in older patients (25%). By multivariate analysis, risk factors for early infection included younger recipient age (p = 0.05), mechanical ventilation at time of transplantation (p = 0.0002), positive donor cytomegalovirus serologic study result with negative recipient result (p = 0.004), and longer donor ischemic time (p = 0.04). The overall mortality rate from infection was 5%, with an actuarial freedom from death related to infection of 92% at 1 year after transplantation. The mortality rate was high in patients with fungal infections (52%), yet was low for those with cytomegalovirus infection (6%). Infections accounted for 27% of the overall mortality rate in infants younger than 6 months of age, compared with 16% for older patients. CONCLUSIONS: Although most infections in pediatric heart transplant recipients are successfully treated, infection remains an important cause of posttransplantation morbidity and death, especially in infants. Bacterial infections predominate within the first month after transplantation, whereas the peak hazard for viral infections occurs approximately 2 months after transplantation. Cytomegalovirus infections are common in the pediatric transplant population, but death related to cytomegalovirus is low.     

A detailed comparison of backpropagation neural network andmaximum-likelihood classifiers for urban land use classification

A detailed comparison of the backpropagation neural network and maximum-likelihood classifiers for urban land use classification is presented. Landsat Thematic Mapper images of Tucson, Arizona, and Oakland, California, were used for this comparison. For the Tucson image, the percentage of matching pixels in the two classification maps was only 64.5%, while for the Oakland image it was 83.3%. Although the test site accuracies of the two Tucson maps were similar, the map produced by the neural network was visually more accurate; this difference is explained by examining class regions and density plots in the decision space and the continuous likelihood values produced by both classifiers. For the Oakland scene, the two maps were visually and numerically similar, although the neural network was superior in suppression of mixed pixel classification errors. From this analysis, the authors conclude that the neural network is more robust to training site heterogeneity and the use of class labels for land use that are mixtures of land cover spectral signatures. The differences between the two algorithms may be viewed, in part, as the differences between nonparametric (neural network) and parametric (maximum-likelihood) classifiers. Computationally, the backpropagation neural network is at a serious disadvantage to maximum-likelihood, taking nearly an order of magnitude more computing time when implemented on a serial workstation     

扫描体素显示器

传统的立体显示器（stereoscopic display）需要观众人为地将眼睛的焦距固定在某一个距离,不断改变左、右眼收敛点来观察不同距离的物体,因此会造成眼疲劳和不适感,也降低了图像质量。立体显示器（Volumetric display）可以克服这个问题,但仅限于一定观看距离内和一定调节能力的小物体;同时,它们也不能正确地提供闭塞信号。本文描述了一种可能的解决办法——多平面扫描体素显示器（multi—planar scanned voxel display）。     

Thematic Mapper Image Quality: Registration, Noise, And Resolution

This paper provides an assessment of Thematic Mapper data quality in terms of band-to-band registration, periodic noise, and spatial resolution. Based on the Thematic Mapper images analyzed so far, the band-to-band registration accuracy is very good. For bands within the same focal plane, the mean misregistrations are well within the specification, 0.2 pixel, except for the thermal band. The thermal band was misregistered by three pixels in each direction in early data products. The error in the across-scan direction was close to zero in later data products. For bands between the cooled and uncooled focal planes, there was a consistent mean misregistration of 0.5 pixels along-scan and 0.2-0.3 pixels across-scan, larger than the specified 0.3 pixel error for bands between focal planes. An analysis of the standard deviation of the misregistration indicated all band combinations would meet the registration specifications if the mean misregistrations were removed by the data processing software. Analysis of the periodic noise in one image indicated a noise component in bands 1-4 with a spatial frequency of 0.31 cycles/pixel. Other lower amplitude periodic components were also present. The periodic noise components obscured detail in areas of low contrast. Modulation transfer function (MTF) analysis in a comparative mode showed no difference in MTF between the forward and backward scans. The difference in MTF between radiometrically corrected data and geometrically corrrected data appeared to be attributable largely to the cubic convolution resampling used to derive the geometrically corrected data.