TY - GEN
T1 - A multiscale Laplacian of Gaussian filtering approach to automated pulmonary nodule detection from whole-lung low-dose CT scans
AU - Fotin, Sergei V.
AU - Reeves, Anthony P.
AU - Biancardi, Alberto M.
AU - Yankelevitz, David F.
AU - Henschke, Claudia I.
PY - 2009
Y1 - 2009
N2 - The primary stage of a pulmonary nodule detection system is typically a candidate generator that efficiently provides the centroid location and size estimate of candidate nodules. A scale-normalized Laplacian of Gaussian (LOG) filtering method presented in this paper has been found to provide high sensitivity along with precise locality and size estimation. This approach involves a computationally efficient algorithm that is designed to identify all solid nodules in a whole lung anisotropic CT scan. This nodule candidate generator has been evaluated in conjunction with a set of discriminative features that target both isolated and attached nodules. The entire detection system was evaluated with respect to a sizeenriched dataset of 656 whole-lung low-dose CT scans containing 459 solid nodules with diameter greater than 4 mm. Using a soft margin SVM classifier, and setting false positive rate of 10 per scan, we obtained a sensitivity of 97% for isolated, 93% for attached, and 89% for both nodule types combined. Furthermore, the LOG filter was shown to have good agreement with the radiologist ground truth for size estimation
AB - The primary stage of a pulmonary nodule detection system is typically a candidate generator that efficiently provides the centroid location and size estimate of candidate nodules. A scale-normalized Laplacian of Gaussian (LOG) filtering method presented in this paper has been found to provide high sensitivity along with precise locality and size estimation. This approach involves a computationally efficient algorithm that is designed to identify all solid nodules in a whole lung anisotropic CT scan. This nodule candidate generator has been evaluated in conjunction with a set of discriminative features that target both isolated and attached nodules. The entire detection system was evaluated with respect to a sizeenriched dataset of 656 whole-lung low-dose CT scans containing 459 solid nodules with diameter greater than 4 mm. Using a soft margin SVM classifier, and setting false positive rate of 10 per scan, we obtained a sensitivity of 97% for isolated, 93% for attached, and 89% for both nodule types combined. Furthermore, the LOG filter was shown to have good agreement with the radiologist ground truth for size estimation
KW - Algorithm evaluation
KW - Automated pulmonary nodule detection
KW - Computed tomography (CT)
KW - Computer-assisted diagnosis (CAD)
KW - Laplacian of Gaussian (LOG)
KW - Validation
UR - https://www.scopus.com/pages/publications/66749097839
U2 - 10.1117/12.811420
DO - 10.1117/12.811420
M3 - Conference contribution
AN - SCOPUS:66749097839
SN - 9780819475114
T3 - Progress in Biomedical Optics and Imaging - Proceedings of SPIE
BT - Medical Imaging 2009
T2 - Medical Imaging 2009: Computer-Aided Diagnosis
Y2 - 10 February 2009 through 12 February 2009
ER -