Multi-scale analysis of lung computed tomography images

Abstract

A computer-aided detection (CAD) system for the identification of

Comparing and combining algorithms for computer-aided detection of pulmonary nodules in computed tomography scans: The ANODE09 study

“…It appeared that vessel branchings were the most common cause for false positives. This is in agreement with observations in Gori et al (2007), Lee et al (2005), Das et al (2006) regarding the false positives produced by other commercial systems not included in this study. Interestingly, there were quite a few findings where two or more vessels seemed to be in contact with each other, forming a point that looked nodular to anyone not tracing the vessel tree through several sections.…”

“…Both are three-step procedures (Buscema, 2004;Gori and Mattiuzzi, 2008;Gori et al, 2007;Retico et al, 2008;Retico et al, 2009;Gori et al, 2009): 1. Lung segmentation: an approach based on thresholding, region growing and morphological operators is implemented, once the scans have been isotropically resampled.…”

“…3. FP reduction: an original procedure, the Voxel-based neural approach (Gori et al, 2007;Retico et al, 2008;Retico et al, 2009;Gori et al, 2009), was developed to reduce the number of FPs in the lists of internal and juxtapleural candidate nodules. First, a region of interest (ROI) including voxels belonging to the candidate nodule was defined from each location provided by the previous step.…”

Computer‐aided detection of lung nodules via 3D fast radial transform, scale space representation, and Zernike MIP classification

“…Several CAD systems have been developed so far, with fairly different approaches. [39][40][41][42][43][44][45][46][47][48][49][50][51][52][53]38,[54][55][56][57] The first attempts in this field were based on 2D techniques mainly because CT scans with axial spacing comparable with image resolution have become available only in recent years. This improvement has favored the adoption of 3D algorithms in some or all parts of the CAD systems.…”

“…In the last couple of years, though, the number and quality of exams present in the LIDC database have reached levels that can be effectively used for CAD evaluation. [50][51][52][53]38,[54][55][56][57] Table VI shows a comparison of the CAD performance with studies presenting the results obtained using the LIDC data.…”