Radiation exposure during CT imaging has become a major concern for imaging device manufacturers, radiologists, patients and radiation safety organizations, such as the French Radioprotection and Nuclear Safety Institute (IRSN). Using the iterative reconstruction (IR) algorithms offered by device manufacturers is one of the approaches implemented to reduce radiation exposure. Numerous investigations have been carried out to assess the impact of IR algorithms, usually by testing their results against filtered back projection (FBP). The majority of these studies have used both algorithms to reconstruct the same raw data and then made a head-to-head comparison. In other studies, successive acquisitions of the same patients were compared, but with different exposure parameters. Finally, IR algorithms have also been evaluated using phantom studies. Even if some variability is observed depending on the methodology used, the region examined, the image quality required and the algorithm itself, the results of these studies are consistent. The latest IR algorithms significantly enhance the signal/noise ratio while maintaining image quality and, more importantly, spatial resolution. These improvements, whatever the anatomical region examined, make it possible to reduce exposure parameters, hence decreasing patient radiation dose. Many authors therefore consider that they are a suitable alternative to FBP [1][2][3][4][5][6][7][8].In this issue of Diagnostic and Interventional Imaging, Pontana et al. reported a novel approach for assessing the impact of IR on CT pulmonary angiography [9]. In their article, the authors evaluated CT angiograms obtained with a dual-source CT unit. They compared the image quality achieved when both X-ray sources are used together and the images reconstructed using FBP with data from a single source at 40% of the total effective dose and IR. This approach allowed the authors to directly compare low-dose acquisition with IR and classic acquisition with FBP in the same patients. Their results demonstrate that the images obtained with IR and a dose reduced by 60% are of better quality than those obtained with FBP and the full dose.This issue of Diagnostic and Interventional Imaging also includes a phantom study (Catphan ® 500 and anthropomorphic phantom) by Greffier et al. [10]. These researchers intended to help radiologists find the optimal combination of parameters that should be used with IR algorithms. They reported 4032 possible combinations of acquisition parameters depending on the kVp, mAs, reconstruction kernel, slice thickness, and reconstruction type. This study also quantified the reduction of the noise power spectrum for each algorithm and revealed the limitations of such algorithms. They found that excessive power levels amplify the image smoothing effect and are reflected by a leftwards shift of the peaks towards lower frequencies.