Dose reduction in abdominal CT by 3D filtering

Through the rapid growth of multislice Computer Tomography (CT) imaging, radiation protection has become a major issue in the radiological community. Optimizing CT scanning is a key task when keeping the radiation doses as low as reasonable achievable (the ALARA principle). Post processing filters can improve and restore grainy and noisy low dose CT images by enhancing structure and reducing image noise. In our study of 10 patients, we perform a preliminary evaluation of a novel post processing filter, which does picture element correlations in all three spatial dimensions. By comparing normal dose pictures with unprocessed low-dose pictures and pictures processed with 2D and 3D filters, we will be able to assess the possible clinical value of the 3D filter. This project is a collaboration between Buskerud Hospital (BU), Buskerud University College, Norwegian Radiation Protection Authority (NRPA) and the Center for Medical Image Science and Visualization (CMIV). The project is part of Lars Borgen's PhD dissertation.

• Staff:

Örjan Smedby , Professor		Project leader		Radiology IMH
Lars Borgen , PhD student		Doctoral student		Univ. of Oslo; Buskerud Hospital
Frode Lærum , Professor		Main supervisor of Lars Borgen		Univ. of Oslo; Akershus University Hospi
Michael Sandborg , Docent		Researcher		Radiation Physics IMH
Martin Hedlund , Technical director				Contextvision AB
Gunnar Farnebäck , Research scientist		Technical support		Contextvision AB
Isabelle Wegmann Hachette , Project leader		Technical evaluation		Contextvision AB

• Former Staff:

• Project Description:

This study will investigate to what extent a 3D post processing filter is able to restore contrast enhanced CT low dose pictures of the abdomen. We hypothesize that the 3D filter is able to restore abdominal CT images acquired with a 30% dose reduced scan to the quality of normal dose images. We also hypothesize that a 3D filter is more effective in restoring low dose pictures than the 2D filter SharpView (SharpViewAB, Linköping, Sweden) [1]. Finally we hypothesize that a 3D filter is more suitable when doing coronal reconstructions of a post-processed volume. The overall aim of the study is to reduce radiation exposure caused by CT to the patient population.

After a pilot study of 2-3 patients, 10-12 patients will be scanned. Age, sex, height, weight abdominal diameter will be recorded. Inclusion criteria: Male or female, Outpatients, Age above 60 years, Referred for abdominal CT. Exclusion criteria: Abdominal diameter less than 80 cm or larger than 100 cm, Gross pancreatic pathology, Gross anatomic anomalies in the investigated region.

After the standard clinical scan of the abdomen, we will perform two additional scans in the pancreatic region in the contrast equilibrium phase: one with normal dose, and one with dose recuced to 70% of normal. The images from the dose reduced scan will be post-processed with SharpView® as well as the novel 3D filter. Post processing is performed by Contextvision in Linköping. Anonymous pictures in randomized order are then scored by 8-10 experienced radiologists at Buskerud Hospital. For each patient, 3mm axial and 3mm coronal images are evaluated. For each patient 4 stacks of axial images (scan II, scan III unprocessed, scan III 2D processed and scan III 3D processed) and 4 stacks of corresponding coronal images, hence 8 image series per patient in total.

Images are read and scored in the program of ViewDEX (Viewer for Digital Evaluation of X-ray images) [2]. ViewDEX displays images in a randomized order and anonymously. The reader is allowed to do the reading in one or several sessions. The images are scored on a 5 point scale, and the image criteria are partially based on the EU Image Quality Criteria. Data analysis is done by with Visual Grading Characteristics (VGC) [3] and possibly other methods appropriate for ordinal visual grading data.

References
1. Martinsen AC, Saether HK, Olsen DR, Skaane P, Olerud HM. Reduction in dose from CT examinations of liver lesions with a new postprocessing filter: a ROC phantom study. Acta Radiol 2008 Apr;49(3):303-9.
2. Börjesson S, Håkansson M, Båth M, Kheddache S, Svensson S, Tingberg A, et al. A software tool for increased efficiency in observer performance studies in radiology. Radiat Prot Dosimetry 2005;114(1-3):45-52.
3. Båth M, Månsson LG. Visual grading characteristics (VGC) analysis: a non-parametric rank-invariant statistical method for image quality evaluation. Br J Radiol 2007 Mar;80(951):169-76.