Applications Of Iterative Reconstruction Techniques In Low Dose Body CT Examinations And Image Quality Assessment

Part One Preliminary Application of Iterative Reconstruction in Image Space in Non-enhanced Chest CT ExaminationObjective To evaluate the image quality and dose reduction capability in non-enhanced chest CT examinations using iterative reconstruction in image space (IRIS).Materials and Methods A CT water phantom was scanned at120kV/150mAs and100kV/270mAs as the reference, and mAs was decreased in10%intervals down to40%of the reference value. Image noise was evaluated and compared between filtered back-projection (FBP) and IRIS reconstructed data.In the patient study, non-enhanced chest CT examinations were performed on90patients who were randomly assigned into3groups:Group A (n=30) standard dose protocol,120kV/110mAs; Group B (n=30) low tube voltage protocol,100kV/110mAs; Group C (n=30) low tube current protocol,120kV/67mAs. All images were reconstructed by FBP and IRIS algorithms using matched kernels of B30and130. The objective image noise (OIN), signal to noise ratio (SNR) and contrast to noise ratio (CNR) of the aorta and latissimus dorsi were measured. The subjective image quality and noise were scored using a3-point scale by2experienced radiologists. The results of the subjective and objective image assessment were compared between group B and C (low dose) IRIS and group A (standard dose) FBR.Results The phantom study showed comparable image noise between the scans using60%dose with IRIS and100%dose with FBP for both120and100kV. In the patient study, group A, B and C had effective dose of3.81±0.43,2.40±0.19, and2.41±0.15mSv. IRIS had significantly improved OIN, SNR, and CNR than FBP on the same patient. The OIN, SNR and CNR using IRIS in group B and C were improved or comparable to those in group A using FBP. No significant difference was found in subjective image quality and noise between group B/C using IRIS and group A using FBP.Conclusion Compared with FBP, IRIS can maintain or improve image quality on non-enhanced chest CT scan reconstruction while saving40%radiation dose. Part Two Half-Dose Abdominal CT Image Quality and Small Lesion Conspicuity Using Sinogram Affirmed Iterative Reconstruction TechniquePurpose To evaluate the image quality and small lesion conspicuity in half-dose abdominal CT examination using Sinogram Affirmed Iterative Reconstruction (SAFIRE) technique relative to standard full-dose Filtered Back-projection (FD-FBP).Materials and methods A CT Catphan(?)500phantom was scanned at120kV/250mAs and100kV/410mAs as the reference, and mAs was decreased in10%intervals down to20%of the reference value. Space resolution, dense resolution, and image noise was evaluated and compared between filtered back-projection (FBP) and SAFIRE (strength3, s3) reconstructed data.Sixty patients were given extra scans following the unenhanced phase (30patients) and venous phase (30patients) of the conventional standard abdominal CT with120kV/200mAs and real-time automatic exposure control (CARE Dose4D). The extra scans covered axial10cm range of upper abdomen with half-dose protocol120kV/100mAs and other same scan parameters. Afterwords they were reconstructed using SAFIRE with strength3(HD-SAFIRE_s3) and5(HD-SAFIRE_s5). The same range in the corresponding full-dose unenhanced and venous phase was reconstructed using FBP. Using full-dose FBP (FD-FBP) images as a standard reference, both objective and subjective image quality of two half-dose iterative reconstructed images were measured and compared. Small lesions (≤2cm) were counted and the lesion conspicuity and edge sharpness was assessed. The tube current time product and CT volume dose index were recorded and the effective dose was calculated by multiplying dose-length product by0.015mSv·mGy-1·cm-1, the conversion factor of abdomen.Results The phantom study showed the same space resolution and dense resolution between the FBP and SAFIRE_s3images and the comparable image noise between the scans using45%dose with SAFIRE_s3and100%dose with FBP for both120and100kV.Both in unenhanced and venous images, the objective noises of the HD-SAFIRE_s3images were equivalent to those of FD-FBP images. The objective noises of the HD-S AFIRE_s5images were significantly lower than those of other two sets of images (all p=0.000). Similarly, the subjective noise assessment showed that44/60and47/60of patients were assessed as optimum noise in FD-FBP and HD-SAFIRE_s3images while60/60of patients were assessed as over suppressed in HD-SAFIRE_s5images. In FD-FBP images,86small lesions in33patients were detected. But in two iterative reconstructed images, five lesions in liver and one lesion in right kidney (mean size:5.5±0.5mm, range4.9-5.9mm) of the unenhanced group were missed because of obscured edge. No small lesion was missed in iterative reconstructed images of enhanced group. In overall diagnostic acceptability assessment,47/60and42/60patients were graded as fully acceptable in FD-FBP and HD-SAFIRE_s3images. However, only16/60patients were graded as fully acceptable in HD-SAFIRE_s5images.Conclusions The overall image quality of half-dose images with SAFIRE strength3reconstruction was equivalent to that of full-dose images with FBP reconstruction. But partial small lesions were missed in iterative reconstruction unenhanced images due to insufficient edge sharpness. The SAFIRE provides sufficient lesion conspicuity and image quality in enhanced abdominal CT with half of the radiation dose compared to filtered back-projection.