Abstract
A priori knowledge of kidney stone composition is an important factor in determining the optimal method of therapy. In this study, we have shown that stone composition can be determined from x-ray images. This was accomplished by using 25 mathematical constructs to describe the stone image and by using these as inputs to a neural network for pattern recognition. A data base of stone images spanning the compositions of commonly occurring kidney stones of calcium oxalate, uric acid, struvite, and brushite and cystine stones was created by digitizing of the x-ray images. For each image, a region-of-interest inscribed by the image area was defined, from which descriptors of the image were calculated. A subset of the images was used to train the neural network, and the remaining cases were used to test the trained network. Based on the small data base available, the overall accuracy was found to be greater than 70%. For this study, specimen radiographs, i.e., radiographs of stones taken outside the body, were used. We are extending the identification scheme to use radio graphs of stones in vivo.
© 1990 Optical Society of America
PDF ArticleMore Like This
JOAN K. FRISOLI, KEVIN T. SCHOMACKER, KRISHNA M. BHATTA, NORMAN S. NISHIOKA, and THOMAS F. DEUTSCH
CTUH96 Conference on Lasers and Electro-Optics (CLEO:S&I) 1990
Johannes Melcher, Benedikt Günther, Martin Dierolf, Klaus Achterhold, and Franz Pfeiffer
EW4A.5 Compact EUV & X-ray Light Sources (EUVXRAY) 2022
John X. Li and Francis T. S. Yu
MN4 OSA Annual Meeting (FIO) 1990