PURPOSE: Our aim is to develop an automatic method which can detect diverse focal liver lesions (FLLs) in 3D CT volumes. METHOD: A hybrid generative-discriminative framework is proposed. It first uses a generative model to describe non-lesion components and then identifies all candidate FLLs within a 3D liver volume by eliminating non-lesion components. It subsequently uses a discriminative approach to suppress false positives with the advantage of tumoroid, a novel measurement combining three shape features spherical symmetry, compactness and size. RESULTS: This method was tested on 71 abdominal CT datasets (5,854 slices from 61 patients, with 261 FLLs covering six pathological types) and evaluated using the free-response receiver operating characteristic (FROC) curves. Overall, it achieved a true positive rate of 90 % with one false positive per liver. It degenerated gently with the decrease in lesion sizes to 30 ml. It achieved a true-positive rate of 36 % when tested on the lesions less than 4 ml. The average computing time of the lesion detection is 4 min and 28 s per CT volume on a PC with 2.67 GHz CPU and 4.0 GB RAM. CONCLUSIONS: The proposed method is comparable to the radiologists' visual investigation in terms of efficiency. The tool has great potential to reduce radiologists' burden in going through thousands of images routinely.
PURPOSE: Our aim is to develop an automatic method which can detect diverse focal liver lesions (FLLs) in 3D CT volumes. METHOD: A hybrid generative-discriminative framework is proposed. It first uses a generative model to describe non-lesion components and then identifies all candidate FLLs within a 3D liver volume by eliminating non-lesion components. It subsequently uses a discriminative approach to suppress false positives with the advantage of tumoroid, a novel measurement combining three shape features spherical symmetry, compactness and size. RESULTS: This method was tested on 71 abdominal CT datasets (5,854 slices from 61 patients, with 261 FLLs covering six pathological types) and evaluated using the free-response receiver operating characteristic (FROC) curves. Overall, it achieved a true positive rate of 90 % with one false positive per liver. It degenerated gently with the decrease in lesion sizes to 30 ml. It achieved a true-positive rate of 36 % when tested on the lesions less than 4 ml. The average computing time of the lesion detection is 4 min and 28 s per CT volume on a PC with 2.67 GHz CPU and 4.0 GB RAM. CONCLUSIONS: The proposed method is comparable to the radiologists' visual investigation in terms of efficiency. The tool has great potential to reduce radiologists' burden in going through thousands of images routinely.
Authors: K James; E Eisenhauer; M Christian; M Terenziani; D Vena; A Muldal; P Therasse Journal: J Natl Cancer Inst Date: 1999-03-17 Impact factor: 13.506
Authors: Isaac R Francis; Richard H Cohan; Nancy J McNulty; Joel F Platt; Melvyn Korobkin; Achamyeleh Gebremariam; Kartik I Ragupathi Journal: AJR Am J Roentgenol Date: 2003-05 Impact factor: 3.959
Authors: H C Addley; N Griffin; A S Shaw; L Mannelli; R A Parker; S Aitken; H Wood; S Davies; G J Alexander; D J Lomas Journal: Clin Radiol Date: 2011-02-04 Impact factor: 2.350
Authors: Michel Bilello; Salih Burak Gokturk; Terry Desser; Sandy Napel; R Brooke Jeffrey; Christopher F Beaulieu Journal: Med Phys Date: 2004-09 Impact factor: 4.071
Authors: J R Sims; L Rezai Gharai; P W Schaefer; M Vangel; E S Rosenthal; M H Lev; L H Schwamm Journal: Neurology Date: 2009-06-16 Impact factor: 9.910
Authors: Ricardo de Lima Thomaz; Pedro Cunha Carneiro; João Eliton Bonin; Túlio Augusto Alves Macedo; Ana Claudia Patrocinio; Alcimar Barbosa Soares Journal: Med Biol Eng Comput Date: 2017-10-16 Impact factor: 2.602
Authors: Idit Diamant; Assaf Hoogi; Christopher F Beaulieu; Mustafa Safdari; Eyal Klang; Michal Amitai; Hayit Greenspan; Daniel L Rubin Journal: IEEE J Biomed Health Inform Date: 2015-09-11 Impact factor: 5.772