abstract |
(57) An object of the present invention is to provide a new and improved method for automatic detection of an accurate boundary of the rib cage in a digitized chest X-ray image. A computerized method for disease detection and characterization in images obtained from chest x-ray images, comprising upper lung edges, right and left ribcage edges, and right and left hemidiaphragms. The images in the chest x-ray image are processed to determine the boundaries of the rib cage including the edges of the. The tissue measure, including the RMS change in pixel values of the region of interest, is converted to relative exposure to correct for system noise components present in the system used to generate the image. A tissue index and a geometric pattern index are generated. Also, a histogram of that index is generated, It is input and applied to a trained artificial neural network that classifies whether the image is normal or abnormal. |