Anais Viry1, Christoph Aberle2, Damien Racine3, Jean-François Knebel4, Sebastian T Schindera5, Sabine Schmidt6, Fabio Becce6, Francis R Verdun3. 1. Institute of Radiation Physics, Lausanne University Hospital, Rue du Grand-Pré 1, 1007 Lausanne, Switzerland. Electronic address: anais.viry@chuv.ch. 2. Department of Radiology, University Hospital Basel, University of Basel, Basel, Switzerland. 3. Institute of Radiation Physics, Lausanne University Hospital, Rue du Grand-Pré 1, 1007 Lausanne, Switzerland. 4. Laboratory for Investigative Neurophysiology (The LINE), Department of Radiology and Department of Clinical Neurosciences, Lausanne University Hospital, Lausanne, Switzerland; Electroencephalography Brain Mapping Core, Center for Biomedical Imaging (CIBM), Lausanne and Geneva, Switzerland. 5. Department of Radiology, Kantonsspital Aarau, Aarau, Switzerland. 6. Department of Diagnostic and Interventional Radiology, Lausanne University Hospital, Lausanne, Switzerland.
Abstract
PURPOSE: To investigate how various generations of iterative reconstruction (IR) algorithms impact low-contrast detectability (LCD) in abdominal computed tomography (CT) for different patient effective diameters, using a quantitative task-based approach. METHODS: Investigations were performed using an anthropomorphic abdominal phantom with two optional additional rings to simulate varying patient effective diameters (25, 30, and 35 cm), and containing multiple spherical targets (5, 6, and 8 mm in diameter) with a 20-HU contrast difference. The phantom was scanned using routine abdominal protocols (CTDIvol, 5.9-16 mGy) on four CT systems from two manufacturers. Images were reconstructed using both filtered back-projection (FBP) and various IR algorithms: ASiR 50%, SAFIRE 3 (both statistical IRs), ASiR-V 50%, ADMIRE 3 (both partial model-based IRs), or Veo (full model-based IR). Section thickness/interval was 2/1 mm or 2.5/1.25 mm, except 0.625/0.625 mm for Veo. We assessed LCD using a channelized Hotelling observer with 10 dense differences of Gaussian channels, with the area under the receiver operating characteristic curve (AUC) as a figure of merit. RESULTS: For the smallest phantom (25-cm diameter) and smallest lesion size (5-mm diameter), AUC for FBP and the various IR algorithms did not significantly differ for any of the tested CT systems. For the largest phantom (35-cm diameter), Veo yielded the highest AUC improvement (8.5%). Statistical and partial model-based IR algorithms did not significantly improve LCD. CONCLUSION: In abdominal CT, switching from FBP to IR algorithms offers limited possibilities for achieving significant dose reductions while ensuring a constant objective LCD.
PURPOSE: To investigate how various generations of iterative reconstruction (IR) algorithms impact low-contrast detectability (LCD) in abdominal computed tomography (CT) for different patient effective diameters, using a quantitative task-based approach. METHODS: Investigations were performed using an anthropomorphic abdominal phantom with two optional additional rings to simulate varying patient effective diameters (25, 30, and 35 cm), and containing multiple spherical targets (5, 6, and 8 mm in diameter) with a 20-HU contrast difference. The phantom was scanned using routine abdominal protocols (CTDIvol, 5.9-16 mGy) on four CT systems from two manufacturers. Images were reconstructed using both filtered back-projection (FBP) and various IR algorithms: ASiR 50%, SAFIRE 3 (both statistical IRs), ASiR-V 50%, ADMIRE 3 (both partial model-based IRs), or Veo (full model-based IR). Section thickness/interval was 2/1 mm or 2.5/1.25 mm, except 0.625/0.625 mm for Veo. We assessed LCD using a channelized Hotelling observer with 10 dense differences of Gaussian channels, with the area under the receiver operating characteristic curve (AUC) as a figure of merit. RESULTS: For the smallest phantom (25-cm diameter) and smallest lesion size (5-mm diameter), AUC for FBP and the various IR algorithms did not significantly differ for any of the tested CT systems. For the largest phantom (35-cm diameter), Veo yielded the highest AUC improvement (8.5%). Statistical and partial model-based IR algorithms did not significantly improve LCD. CONCLUSION: In abdominal CT, switching from FBP to IR algorithms offers limited possibilities for achieving significant dose reductions while ensuring a constant objective LCD.
Authors: Brian R Herts; Andrew Schreiner; Frank Dong; Andrew Primak; Jennifer Bullen; Wadih Karim; Douglas Nachand; Sara Hunter; Mark E Baker Journal: J Appl Clin Med Phys Date: 2020-12-28 Impact factor: 2.102
Authors: Anaïs Viry; Christoph Aberle; Thiago Lima; Reto Treier; Sebastian T Schindera; Francis R Verdun; Damien Racine Journal: Eur Radiol Date: 2021-07-29 Impact factor: 5.315