Premium
Localization of inter‐rib spaces for lung texture analysis and computer‐aided diagnosis in digital chest images
Author(s) -
Powell Gregory F.,
Doi Kunio,
Katsuragawa Shigehiko
Publication year - 1988
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.596209
Subject(s) - region of interest , rib cage , sampling (signal processing) , computer science , computer aided diagnosis , artificial intelligence , texture (cosmology) , computer vision , computer aided , pattern recognition (psychology) , radiology , image (mathematics) , medicine , anatomy , filter (signal processing) , programming language
An automated method for sampling lung textures in digital posterior/anterior chest images is being developed for use in computer‐aided diagnosis of interstitial pulmonary diseases. In our present approach, two vertical profiles in the periphery of both lungs are fitted with a shift‐variant sinusoidal function from which we estimate locations of posterior ribs and inter‐rib spaces. Regions of interest (ROI's) for sampling lung textures are then automatically centered on the calculated locations of inter‐rib spaces. In tests with 66 chest images, the overall success rate in placing 6.4 mm×6.4 mm ROI's within inter‐rib spaces with this method was 71%, with an average of 18 ROI's selected in 4–5 s/image by a VAX11/750 computer. When four additional alternative ROI's were selected on the sides of each original ROI, the success rate in having at least one ROI correctly located in an inter‐rib space increased to 94%. Since we are still developing a fully automated sampling method, the present approach has been incorporated into a semiautomated method that is currently being used to sample lung textures from a large number of clinical cases.