RESUMEN
The Laplacian of Gaussian (LoG) filter is widely used in interest point detection. However, low-contrast image structures, though stable and significant, are often submerged by the high-contrast ones in the response image of the LoG filter, and hence are difficult to be detected. To solve this problem, we derive a generalized LoG filter, and propose a zero-norm LoG filter. The response of the zero-norm LoG filter is proportional to the weighted number of bright/dark pixels in a local region, which makes this filter be invariant to the image contrast. Based on the zero-norm LoG filter, we develop an interest point detector to extract local structures from images. Compared with the contrast dependent detectors, such as the popular scale invariant feature transform detector, the proposed detector is robust to illumination changes and abrupt variations of images. Experiments on benchmark databases demonstrate the superior performance of the proposed zero-norm LoG detector in terms of the repeatability and matching score of the detected points as well as the image recognition rate under different conditions.
RESUMEN
Objective To evaluate the diagnosis accuracy for coronary stenosis from diameter measurements of differ?ent severities of stenosis by using gemstone spectral imaging (GSI). Methods Totally 3 different coronary artery models(in?ternal diameter:3.00 mm) that with different severities of stenosis (residual lumen diameter ratio:75%, 50%and 25%) were placed in a pulsating cardiac phantom (ALPHA 1-VT PC, Fuyo Corporation, Japan). The coronary phantom was scanned in two models:gemstone spectral imaging (GSI) and axial scan model(AXIAL). All the spectral imaging data were analyzed by using GSI viewer to reconstruct the VMS images (40-140 keV by 10 keV interval) and AXIAL. CT values (HU) of the re?mained lumen from all different stenosis lumens were measured. The residual rate (%) was calculated by dividing the diame?ters of the remained lumen to normal lumen on the reformatted short axial images. One-way ANOVA was used to compare the measurement difference of residual rates between VMS images. Student t-test was used to compare the measurement dif?ference of residual rate between VMS and AXIAL. Bland-Altman test was used to compare the measured residual rates and the gold standard. Results There were no significant differences in measured residual rate betweeb VMS images (P>0.05). From student t-test, only significant difference was found on 50%stenosis between 70 keV VMS and AXIAL(t=4.617,P<0.042). From the Bland-Altman test, measurement of residual rate was more accurate from 70 keV VMS than that of AXIAL by taking the stenosis rate of real model as gold standard(t=14.560,P<0.001). Conclusion VMS image (70 keV) shows more accurate rate than AXIAL images on both diameter measurement and evaluation of the coronary stenosis.