Computers in Biology and Medicine 38 (2008) 525 – 534 www.intl.elsevierhealth.com/journals/cobm Lung nodule detection in low-dose and thin-slice computed tomography A. Retico a , ∗ , P. Delogu a , b , M.E. Fantacci a , b , I. Gori a , c , A. Preite Martinez d a Istituto Nazionale di Fisica Nucleare, Sezione di Pisa, Largo Pontecorvo 3, 56127 Pisa, Italy b Dipartimento di Fisica dell’Università di Pisa, Largo Pontecorvo 3, 56127 Pisa, Italy c Bracco Imaging S.p.A., Via E. Folli 50, 20134 Milano, Italy d Centro Studi e Ricerche Enrico Fermi, Via Panisperna 89/A, 00184 Roma, Italy Received 28 March 2007; accepted 1 February 2008 Abstract A computer-aided detection (CAD) system for the identification of small pulmonary nodules in low-dose and thin-slice CT scans has been developed. The automated procedure for selecting the nodule candidates is mainly based on a filter enhancing spherical-shaped objects. A neural approach based on the classification of each single voxel of a nodule candidate has been purposely developed and implemented to reduce the amount of false-positive findings per scan. The CAD system has been trained to be sensitive to small internal and sub-pleural pulmonary nodules collected in a database of low-dose and thin-slice CT scans. The system performance has been evaluated on a data set of 39 CT containing 75 internal and 27 sub-pleural nodules. The FROC curve obtained on this data set shows high values of sensitivity to lung nodules (80–85% range) at an acceptable level of false positive findings per patient (10–13 FP/scan).  2008 Elsevier Ltd. All rights reserved. Keywords: Computer-aided detection (CAD); Low-dose computed tomography (LDCT); Thin-slice CT; Image processing 1. Introduction Lung cancer is one of the main public health issue in devel- oped countries. The overall 5-year survival rate is only 10–15% [1,2] and no significant improvement has occurred in the last 20 years [3]. Approximately 70% of lung cancers are diagnosed at too advanced stages for the treatments to be effective [4]. Five-year survival rate for early-stage lung cancers (stage I), which can reach 70% [5], is sensibly higher than for cancers diagnosed at more advanced stages. Early diagnoses and more effective treatments have contributed to the reduction of lung cancer deaths in the last decade. The mortality rate for lung cancer among men in the United States has started to decrease by about 1.5% per year since 1991, whereas it has leveled off after increasing for many decades among women [1]. A simi- lar trend in lung cancer mortality rate for the male population has been observed in most European countries; by contrast, in case of the female population it is still rising, despite a decline ∗ Corresponding author. Tel.: +39 502214459; fax: +39 502214460. E-mail address: retico@df.unipi.it (A. Retico). 0010-4825/$ - see front matter  2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.compbiomed.2008.02.001 in the mortality of young women has been observed over the last decade [6,7]. Lung cancer most commonly manifests itself as non-calcified pulmonary nodules. They are rather spherical objects that can be characterized by very low CT values and/or low contrast. Nodules may have CT values in the same range of those of blood vessels and airway walls and may be strongly connected to them. As computed tomography (CT) has been shown as the most sensitive imaging modality for the identification of small pulmonary nodules, particularly since the introduction of the multi-detector-row and helical CT technologies [8], screening programs based on low-dose CT (LDCT) are regarded as a promising technique for detecting small, early-stage lung can- cers [9,10]. As the efficacy of CT-based screening programs in reducing the mortality rate for lung cancer has not been yet demonstrated, different and opposing opinions are being pointed out on this topic by many experts [11,12]. This is the ideal context for conducting a screening trial. Low-dose CT settings are currently recommended by the screening trial protocols. However, it is not trivial in this case to