PURPOSE: The purpose of the study was to determine whether a model-based iterative reconstruction (MBIR) technique improves diagnostic confidence and detection of pulmonary embolism (PE) compared with hybrid iterative reconstruction (HIR) and filtered back projection (FBP) reconstructions in patients undergoing computed tomography pulmonary angiography. MATERIALS AND METHODS: The study was approved by our institutional review board. Fifty patients underwent computed tomography pulmonary angiography at 100 kV using standard departmental protocols. Twenty-two of 50 patients had studies positive for PE. All 50 studies were reconstructed using FBP, HIR, and MBIR. After image randomization, 5 thoracic radiologists and 2 thoracic radiology fellows graded each study on a scale of 1 (very poor) to 5 (ideal) in 4 subjective categories: diagnostic confidence, noise, pulmonary artery enhancement, and plastic appearance. Readers assessed each study for the presence of PE. Parametric and nonparametric data were analyzed with repeated measures and Friedman analysis of variance, respectively. RESULTS: For the 154 positive studies (7 readers × 22 positive studies), pooled sensitivity for detection of PE was 76% (117/154), 78.6% (121/154), and 82.5% (127/154) using FBP, HIR, and MBIR, respectively. PE detection was significantly higher using MBIR compared with FBP (P = 0.016) and HIR (P = 0.046). Because of nonsignificant increase in FP studies using HIR and MBIR, accuracy with MBIR (88.6%), HIR (87.1%), and FBP (87.7%) was similar. Compared with FBP, MBIR led to a significant subjective increase in diagnostic confidence, noise, and enhancement in 6/7, 6/7, and 7/7 readers, respectively. Compared with HIR, MBIR led to significant subjective increase in diagnostic confidence, noise, and enhancement in 5/7, 5/7, and 7/7 readers, respectively. MBIR led to a subjective increase in plastic appearance in all 7 readers compared with both FBP and HIR. CONCLUSIONS: MBIR led to significant increase in PE detection compared with FBP and HIR. MBIR led to qualitative improvements in diagnostic confidence, perceived noise, and perceived enhancement compared with FBP and HIR.
PURPOSE: The purpose of the study was to determine whether a model-based iterative reconstruction (MBIR) technique improves diagnostic confidence and detection of pulmonary embolism (PE) compared with hybrid iterative reconstruction (HIR) and filtered back projection (FBP) reconstructions in patients undergoing computed tomography pulmonary angiography. MATERIALS AND METHODS: The study was approved by our institutional review board. Fifty patients underwent computed tomography pulmonary angiography at 100 kV using standard departmental protocols. Twenty-two of 50 patients had studies positive for PE. All 50 studies were reconstructed using FBP, HIR, and MBIR. After image randomization, 5 thoracic radiologists and 2 thoracic radiology fellows graded each study on a scale of 1 (very poor) to 5 (ideal) in 4 subjective categories: diagnostic confidence, noise, pulmonary artery enhancement, and plastic appearance. Readers assessed each study for the presence of PE. Parametric and nonparametric data were analyzed with repeated measures and Friedman analysis of variance, respectively. RESULTS: For the 154 positive studies (7 readers × 22 positive studies), pooled sensitivity for detection of PE was 76% (117/154), 78.6% (121/154), and 82.5% (127/154) using FBP, HIR, and MBIR, respectively. PE detection was significantly higher using MBIR compared with FBP (P = 0.016) and HIR (P = 0.046). Because of nonsignificant increase in FP studies using HIR and MBIR, accuracy with MBIR (88.6%), HIR (87.1%), and FBP (87.7%) was similar. Compared with FBP, MBIR led to a significant subjective increase in diagnostic confidence, noise, and enhancement in 6/7, 6/7, and 7/7 readers, respectively. Compared with HIR, MBIR led to significant subjective increase in diagnostic confidence, noise, and enhancement in 5/7, 5/7, and 7/7 readers, respectively. MBIR led to a subjective increase in plastic appearance in all 7 readers compared with both FBP and HIR. CONCLUSIONS: MBIR led to significant increase in PE detection compared with FBP and HIR. MBIR led to qualitative improvements in diagnostic confidence, perceived noise, and perceived enhancement compared with FBP and HIR.