ELECTRONIC FILE CHARACTERISTICS: 44 files; MS Word (.DOC), Audio Video Interleave (.AVI), JPEG image, and TSCC application. PHYSICAL DESCRIPTION: 1 computer laser optical disc (CD-ROM); 4 3/4 in.; 70.4 MB. SYSTEMS DETAIL NOTE: IBM-clone PC-compatible. ABSTRACT: Assessment of density patterns is critical for quantification of the morphology in healthy and diseased lungs. We propose a fast and accurate method for semiautomatic lung density distribution assessment. The algorithm is based on a customized, commercially available software package that supports various file formats. Analysis of porcine CT scans was done using a software package, Able Software Corp., 3D-Doctor, Lexington, MA. Briefly, axial CT scans were semi automatically analyzed, slice-by-slice, using the interactive segmentation function of the 3D-Doctor software. Pre, and post, contusion image stacks were quantified and compared with respect to density distribution. The pulmonary parenchyma was interactively segmented employing 5 regimens. Each of the regimens represented a window of Hounsfield units (HU) corresponding to level of aeration of the parenchyma as reported by Gattinoni et al. Air (HU value- 1000), hyperinflated areas (HU window from -998 to -902), normally aerated areas (HU window from -900 to -500) poorly aerated areas (HU window from -998 to -902), normally aerated areas (HU window from -900 to -500), poorly aerated areas (HU window from -498 to -100) and non-aerated areas (HU windows from -98 to +100) were defined in each of the slices for each of the lungs. The resulting data were represented with respect to number of pixels, total density, mean density, object histogram as well as surface area and volume data for each of the objects. 3-D reconstruction was performed via simple surface rendering by the program. 3-D reconstructed rotational images were available to display areas and regions of the analyzed lung parenchyma both as a whole and in separate with respect to aeration and 7