| 摘要 |
A system, method and computer program product is provided for screening and analyzing spectral images of a microcirculatory system to measure blood characteristics, such as hemoglobin concentration. In an embodiment, multi-stage screening is performed on each input image to determine whether the image is suitable for hemoglobin measurement and prediction. During the screening process the following quantities are measured: image intensity, image background intensity variation, image focus condition, image motion blur condition, number of vessels in the image with certain range of diameters, average vessel edge contrast, number of bubble segments in the image and the similarity between two images. The multi-stage screening cycle includes intensity screening, DWT decomposition, focus measurement, motion blur measurement, vessel and bubble detection, and repetition detection. During the screening stages, the image is classified into one of three categories: excellent, mediocre and poor. Only the images with an excellent rating is used for further Hb estimation. But both excellent and mediocre images would launch an autofocusing process provided by focus measurement. In another embodiment, a Fast Fourier transform is used to develop a spectral ratio to detect excellent images. The present invention provides real time hemoglobin prediction on the subject under investigation. |
