Analysis of the effects of dose reduction on the detection of lung nodules on low- and ultra-low-dose CT scans of the chest chest CT scans for lung cancer screening (a phantom study)

The detection of lung nodules in computed tomography of the chest plays a major role in the early diagnosis of lung cancer, potentially reducing this disease's mortality. However, the use of ionizing radiation inherent to the method can eventually determine side effects, including carcinogenesis. This problem led to the development of protocols that use lower X-ray doses as a way to minimize the risk of adverse outcomes. Though, lower doses can compromise the quality of the exam, even resulting in non-diagnostic images. This pursuit for the perfect balance between dose and image-quality is a challenge of modern radiology. This study aims to analyze the effects of dose reduction in computed tomography of the chest over the detection of simulated lung nodules in a phantom. The phantom was scanned using five different protocols (one ultra-low-dose, three low-dose and a conventional chest computed tomography). The three different low dose protocols consisted of the one recommended by the American Association of Physicists in Medicine (conventional low dose protocol) and two experimental protocols (lower kV and lower mAs). We compared the performance of four radiologists with expertise in chest radiology in detecting the nodules in each protocol and evaluated interobserver agreement. We also measured the amount of radiation in each protocol using thermoluminescent dosimeters, in order to precisely compare their respective doses. The equivalent dose in the ultra-low-dose protocol was less than a third (0.5 mSv) of that of the conventional low-dose protocol (1.8 mSv). The experimental low-dose protocols (lower kV and lower mAs) led to discrete effective-dose reduction (1.34 and 1.32 mSv, respectively). There was no statistically significant difference in lung nodule detection when comparing different low-dose protocols with each other or with ultra-low-dose protocol (p = 0.208 to p=1.000), however, difference was significant when comparing them to conventional CT (p < 0.001). Lower-attenuating nodules (<-600 UH) also had significant difference in detection rates when comparing ultra-low-dose and low-dose CT with conventional CT (p < 0.001 a p = 0.007). The level of agreement between different observers in all protocols ranged from moderate (k = 0.41) to perfect (k = 1.0). These facts show that the performance of radiologists is similar in low and ultra-low-dose protocols, offering alternatives with doses lower than 1 mSv to perform lung cancer screening. (author)