COMPUTER-AIDED DIAGNOSIS FOR INTERSTITIAL INFILTRATES IN CHEST RADIOGRAPHS - OPTICAL-DENSITY DEPENDENCE OF TEXTURE MEASURES

We have been developing a computerized scheme for automated detection and characterization of interstitial infiltrates based on the Fourier transform of lung texture. To improve the performance of the scheme, w hich was developed using digitized screen-film radiographs, optical-de nsity dependence of both the gradient of the film used and the system noise associated with the laser scanner were investigated. Two hundred chest radiographs, including 100 abnormal cases with interstitial inf iltrates, were digitized using a laser scanner. The root-mean-square ( RMS) variations and the first moments of the power spectra, which corr espond to the magnitude and coarseness of lung texture, were determine d by Fourier transform of lung textures in numerous regions of interes t (ROIs). The RMS variation was dependent upon the average optical den sity in the ROI, though no obvious trend existed for the first moment of the power spectrum. Dependence of the RMS variations on optical den sity was corrected for using the gradient curve of the film. Also, sys tem noise associated with the laser scanner was corrected. Results ind icated that the specificity was improved from 81% (without correction) to 89% (with corrections), without any loss of sensitivity (90%). Thu s, the correspondence between the computer output and consensus interp retation of radiologists was improved with the new scheme compared to the previous one. This improved computerized scheme may be useful to r adiologists in detecting interstitial infiltrates in chest radiographs .