BACKGROUND: A radiation dose of any magnitude can produce a detrimental effect manifesting as an increased risk of cancer. Cancer development may be delayed for many years following radiation exposure. Minimizing radiation dose in children is particularly important. However, reducing the dose can reduce image quality and may, therefore, hinder lesion detection. OBJECTIVE: We investigated the effects of reducing the image signal-to-noise ratio (SNR) on CT lung nodule detection for a range of nodule sizes. MATERIALS AND METHODS: A simulated nodule was placed at the periphery of the lung on an axial CT slice using image editing software. Multiple copies of the manipulated image were saved with various levels of superimposed noise. The image creation process was repeated for a range of nodule sizes. For a given nodule size, output images were read independently by four Fellows of The Royal College of Radiologists. RESULTS: The overall sensitivities in detecting nodules for the SNR ranges 0.8-0.99, 1-1.49, and 1.5-2.35 were 40.5%, 77.3% and 90.3%, respectively, and the specificities were 47.9%, 73.3% and 75%, respectively. The sensitivity for detecting lung nodules increased with nodule size and increasing SNR. There was 100% sensitivity for the detection of nodules of 4-10 mm in diameter at SNRs greater than 1.5. CONCLUSION: Reducing medical radiation doses in children is of paramount importance. For chest CT examinations this may be counterbalanced by reduced sensitivity and specificity combined with an increased uncertainty of pulmonary nodule detection. This study demonstrates that pulmonary nodules of 4 mm and greater in diameter can be detected with 100% sensitivity provided that the perceived image SNR is greater than 1.5.
BACKGROUND: A radiation dose of any magnitude can produce a detrimental effect manifesting as an increased risk of cancer. Cancer development may be delayed for many years following radiation exposure. Minimizing radiation dose in children is particularly important. However, reducing the dose can reduce image quality and may, therefore, hinder lesion detection. OBJECTIVE: We investigated the effects of reducing the image signal-to-noise ratio (SNR) on CT lung nodule detection for a range of nodule sizes. MATERIALS AND METHODS: A simulated nodule was placed at the periphery of the lung on an axial CT slice using image editing software. Multiple copies of the manipulated image were saved with various levels of superimposed noise. The image creation process was repeated for a range of nodule sizes. For a given nodule size, output images were read independently by four Fellows of The Royal College of Radiologists. RESULTS: The overall sensitivities in detecting nodules for the SNR ranges 0.8-0.99, 1-1.49, and 1.5-2.35 were 40.5%, 77.3% and 90.3%, respectively, and the specificities were 47.9%, 73.3% and 75%, respectively. The sensitivity for detecting lung nodules increased with nodule size and increasing SNR. There was 100% sensitivity for the detection of nodules of 4-10 mm in diameter at SNRs greater than 1.5. CONCLUSION: Reducing medical radiation doses in children is of paramount importance. For chest CT examinations this may be counterbalanced by reduced sensitivity and specificity combined with an increased uncertainty of pulmonary nodule detection. This study demonstrates that pulmonary nodules of 4 mm and greater in diameter can be detected with 100% sensitivity provided that the perceived image SNR is greater than 1.5.
Authors: Magnus Båth; Markus Håkansson; Sara Börjesson; Susanne Kheddache; Anna Grahn; François O Bochud; Francis R Verdun; Lars Gunnar Månsson Journal: Radiat Prot Dosimetry Date: 2005 Impact factor: 0.972
Authors: Nancy R Fefferman; Elan Bomsztyk; Angela M Yim; Rafael Rivera; John B Amodio; Lynne P Pinkney; Naomi A Strubel; Marilyn E Noz; Henry Rusinek Journal: Radiology Date: 2005-09-16 Impact factor: 11.105
Authors: C Kimme-Smith; D R Aberle; J W Sayre; E M Hart; S M Greaves; K Brown; D A Young; M D Deseran; T Johnson; S L Johnson Journal: AJR Am J Roentgenol Date: 1995-08 Impact factor: 3.959