| Purposes1. To evaluate the effects of different reconstruction settings of knowledge-based iterative reconstruction technique (IMR) on image quality in hepatic contrast-enhanced CT (CECT) at different radiation dose levels2. To determine the most optimized iterative reconstruction technique settings in hepatic contrast-enhanced CT (CECT) at different radiation dose levelsMaterials and MethodsGeneral informationThis prospective study was approved by Guangdong General Hospital (Guangdong Academy of Medical Sciences) institutional review board (Register Number:GDRE2012120H) and all patients who underwent hepatic CECT were enrolled between February and May 2013 written informed consent.According to random number table, patients were randomly assigned to two different protocols in portal-venous phase (Standard dose group, SD-group;Low dose group,LD-group).The inclusion criteria for this study included:(1)referred for a hepatic CECT due to hepatic treatment history or suspicious abdominal symptoms; (2) The body mass index (BMI) is 21～25 kg/m2; (3) Age> 18 years.The exclusion criteria for this study included:(1) pregnant or lactating (female subjects only); (2) had severe cardiac insufficiency; (3) had severe renal failure (serum creatinine clearance rate= 120 umol/L);(4)had severe hepatic insufficiency; (5) some diseases affecting the liver hemodynamic (such as fatty liver); (6) anaphylactic reaction previous history; (7) apparent motion artifact effected image analysis; (8) the anterior abdominal wall subcutaneous fat of the is too thin to fail to be measured.CT Imaging Protocol1. Examining methods and scanning parameters:All patients were examined with a 256-slice CT scanner (Brilliance iCT; Philips Healthcare, Cleveland, Ohio, USA). All patients underwent both non-enhanced CT and triphasic CECT scan including hepatic arterial phase (HAP), portal-venous phase (PVP) and delayed phase. Patients were instructed to hold their breath with tidal inspiration during scans with a supine position and scan area was from the top of the liver to the lower pole of the kidney in a cephalocaudal direction. Scanning parameters:non-enhanced, hepatic arterial phase and delayed phase images were acquired at a fixed dose level with tube voltage of 120 kV and tube current product of 250mAs in standard and low dose groups, while portal-venous phase images were acquired according to standard and low dose groups respectively by adjusting tube voltage and tube current-time products:standard lose group(SD-group),120 kV,50mAs;low dose group(LD-group), 80 kV,500mAs.The other scan parameters were as follows:detector collimation,128 × 0.625 mm; helical pitch,0.899; Reconstruction Matrix:512 × 512;gantry rotation time,0.75 s;FOV,350mm×350;slice thickness,3.0 mm; section interval,3.0 mm. A total of 80 ml of contrast medium (Ultravist 370, Bayer Schering Pharma AG, Berlin, Germany) at a flow rate of 5 ml/s followed by 15 ml of a saline solution was delivered via a 20-gauge catheter inserted into an attained forearm vein with double tube high-pressure syringe. Hepatic arterial phase (HAP), portal-venous phase (PVP) and delayed phase scanning was started 25,60,120 seconds after injection of contrast medium.2. CT Image Reconstruction:PVP images were reconstructed with both filtered back projection (FBP) and IMR techniques. The FBP images were reconstructed by the host CT scanner,while the IMR images were reconstructed by a Philips prototype workstation. In order to provide more diagnostic information and imaging features, two different IMR reconstruction algorithm including body routine (BR) and body soft tissue (BST) algorithm were recommended by the manufacture for abdominal CT, with body routine taking into account of both contrast and sharpness, body soft tissue focusing on low contrast detectability. According to the levels of noise reduction, with three different levels (L1-L3) each provided on the prototype, with L3 providing the maximum noise reduction. with a total of 6 types of IMR reconstruction settings available in clinical practice (L1BR,L2BR,L3BR,L1BST,L2BST and L3BST). According to different CT image reconstruction, the seven image series (L1BR,L2BR,L3BR,L1BST,L2BST,L3BST and FBP) were performed on our study. Image AnalysisAll CT image data sets were reviewed randomly by using on the same workstation (Extended Brilliance Workspace V4.5.2, Philips Healthcare, Cleveland, Ohio, USA) for assessment of Subjective image and Objective image.Two readers were trained to assess image quality so that they would understand the evaluation system, and then to improve interobserver agreement. The each reconstruction images were presented in random order to two readers who were blinded to the reconstruction settings. The patients sex, age and name, as well as the CT scanning parameters,all hospital record numbers, image parameters,the type CT image reconstruction were removed from the images. The imge quantity was evaluated by a unified display screen Philips EBW).1. Subjective Image Quality:For subjective image analysis,two abdomen radiologists (associate chief physician) who were blinded to the reconstruction settings independently used a 4-point subjective scale to assess subjective image quality. Subjective image quality was performed with a window width of 250 HU and a window level of 50HU, respectively, the standard abdominal window setting, but the radiologists had allowed to adjust the window settings during subjective image quality assessments.The subjective image quality assessments included:(1) the low contrast detectability (LCD):1, very poor or suboptimal contrast; 2, acceptable image contrast; 3, above average image contrast; 4, excellent image contrast.(2) Image distortion (ID):1, present and unacceptable; 2, present and interfering with the depiction of major structures; 3, present, but not interfering with depiction of major structures; 4, minimal or absent.(3) Diagnostic confidence (DC):1, cannot provide diagnostic information; 2, poor confidence; 3, probably confident; 4, completely confident.2. Objective Image Quality:The objective image quality indicators include image noise, signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR).An abdomen radiologist’s (associate chief physician) who were blinded to the reconstruction settings independently to analysis objective image quality. Mean CT attenuation and standard deviation (SD) of the hepatic parenchyma were measured. ROIlive was recorded as the mean of three ROIs placed in the left hepatic lobe anterior and the anterior segment of the right hepatic lobe and the posterior segment of the right hepatic lobe on images obtained at the level of the main portal vein. Exclude visible artifacts, bile ducts, and blood vessels from ROI measurements of the hepatic parenchyma. A circular ROI drawn at right spinae muscle and the subcutaneous fat of the anterior abdominal wall on images obtained at the level of the main portal vein. For all measurements, the shape, position, and size of the ROIs were kept constant by applying a copy-and-paste function at the workstation, and recorded Mean CT attenuation and standard deviation (SD). All measurements were performed 3 times and expressed as the mean value. Image noise was defined as the mean standard deviation of the ROI in the hepatic parenchyma. Signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) of hepatic parenchyma were calculated by using the following formula:SNR= CTh/SD n, CNR= (CTh-CTs)/SDn, CTh is the mean attenuation for the hepatic parenchyma; SDn is the mean image noise for the subcutaneous fat. CTs is the mean attenuation for the right spinae muscle. A circular ROI of approximately 100mm2 in size was placed in hepatic parenchyma and right spinae muscle. However, ROI placed in subcutaneous fat was approximately 50 mm2 in size because that lager ROI may approach the edges of muscle in patients whose BMI were less than 25.0kg/m2.3. CT Radiation dose:Dose-length product (DLP) values were recorded for each patient and each protocol. Estimated effective dose (ED) was calculated from DLP as follows:effective dose (ED)=DLP×conversion factor k for the abdomen, which in the case of abdominal CT examinations is 0.015 (mSv·mGy-1·cm-1) all protocols.Statistical AnalysisAll statistical analyses were performed with commercially available software (SPSS version 13.0; SPSS Inc, Chicago,Ill) and MedCalc (MedCalc Mariakerke, Belgium).To compare the sex constituent ratio of SD-group with those of LD-group, we used Χ2 test. To compare the age and effective radiation dose of SD-group with those of LD-group, we used Student t test.The objective image quality indicators:image noise, signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR) among the 7 reconstruction algorithms were tested first with the Kolmogorov-Smirnov test for normality and then with the Levene test for variance homogeneity. Mean attenuation values, image noise values,SNR and CNR were compared using analysis of variance among the 7 reconstruction algorithms.S-N-K test was used when there was a significant difference for each evaluation. If data distribution is inhomogeneous, the Welch and Games-Howell tests were used, Welch test was used when there was a significant difference for each evaluation in performing Welch and Games-Howell tests. Two-sided testing was used. A value of p<0.05 was considered to suggest a statistically significant difference.The subjective image quality indicators:Inter-observer agreement for subjective image scores was measured using Kappa test, and k values were as follows:k value of 0-0.40 indicated poor agreement; k value of 0.41-0.60, moderate agreement; k value of 0.61-0.80, substantial agreement; and k=0.81-1.0, almost agreement. The subjective scores assigned to all reconstruction algorithms were compared by using the Friedman test. If there was significant difference among different groups, pair wise comparisons were performed with the Q test.ResultPopulation demographicsThis study population consisted of fifty-four consecutive patients (30 men and 24 women; age range,20-81 years, mean age 50.4 ±14.4 years).Twenty-nine patients were enrolled in standard dose group (17 men and 12 women; age range, 20～73 years, mean age 51±14 years). Twenty-five patients were enrolled in low dose group (13 men and 12 women; age range,23～81 years, mean age 49±15 years).There was no significant difference between standard and low dose groups with respect to age and gender(gender:x2=0.370, P=0.764; age:t=0.119, P=0.732).Radiation DoseThe effective radiation dose of low dose group was 42% lower for the PVP compared with that of standard dose group(4.2mSv±0.2mSv vs 7.3mSv±0.6mSv, t=15.27, P<0.01).Qualitative image assessmentThe two radiologists showed very good consistency in qualitative assessment (kappa value= 0.63-1). Interobserver agreement between the two radiologists was substantial or almost perfect (kappa value= 0.63-1.00) for image quality assessment。 In standard dose group, kappa value of LCD is 0.64-1.00, kappa value of ID is 0.63-0.84, kappa value of DC is 0.66-0.83. In low dose group, kappa value of LCD is 0.69-0.85, kappa value of ID is 0.71-0.87, kappa value of DC is 0.69-0.80.In standard dose group, L3BR, L1BST, L2BST and L3BST scored highest in LCD.However, L3BR, L2BST and L3BST scored lowest in ID, FBP and L1BR scored highest in ID. L1BR,L2BR and L1BST scored highest in DC L3BST scored lowest. In low dose group, the distribution of all reconstruction algorithms scores in LCD, and ID were similar to standard dose group, L2BST and L3BST scored highest in LCD. L2BST and L3BST scored lowest in ID, FBP and L1BR scored highest in ID.however, there was difference in the distribution of DC scores between the two groups, which was L1BST scored highest in DC while FBP scored lowest in DC.Comparison between groups:(1) LCD scored In standard dose group; No statistical significance was observed between L1BR and L2BR, L3BR and L1BST, L2BST and L3BST (all P>0.05), However, there was significant difference among the other reconstruction algorithms (P<0.05) and FBP were statistical differences compared to the other algorithms. (2) ID scored In standard dose group; No statistical significance was observed between L1BR andFBP,L2BR and L1BST (all P>0.05), However, there was significant difference among the other reconstruction algorithms (P<0.05).There was significant difference L3BR, L2BST, L3BST and the other reconstruction algorithms (P<0.05). (3) DC scored In standard dose group; No statistical significance was observed between L1BR,L2BR, L3BR and L1BST (all P >0.05), FBP were no statistical differences compared to L2BST, L3BST and L2BST were no statistical differences compared to L3BR.However, there was significant difference among the other reconstruction algorithms (P<0.05). (4) LCD scored in low dose group:the distribution of comparison between groups in LCD was similar to standard dose group. (5) ID scored in low dose group:No statistical significance was observed between L2BR, L1BST and L2BST, L1BR and FBP (all P> 0.05), However, there was significant difference among the other reconstruction algorithms (P< 0.05).There was significant difference L3BR, L3BST and the other reconstruction algorithms(P<0.05). (6)DC scored In low dose group:No statistical significance was observed between L1BR and L2BR, L3BR, L2BST and L3BST (all P>0.05), However, there was significant difference among the other reconstruction algorithms (P<0.05).There was significant differenceL1BST, FBP and the other reconstruction algorithms (P<0.05). Objective image assessmentIn standard dose group:The image noise of L1BR image was about 6.5±1.3HU, L2BR was abou 5.5±0.9HU, L3BR was about 4.6±0.6HU, L1BST was about 4.5±0.9HU, L2BST was about 3.8±0.6HU, L3BST was about 3.2±0.2HU,FBP was about 9.8±2.0HU. The SNR of L1BR,L2BR,L3BR,L1BST,L2BST,L3BST,FBP image respectively were about 16.9±6.9,18.4±8.4,20.3±0.3,21.4±1.4,23.0±3.0, 24.6±4.6,13.3±3.3. The CNR of L1BR,L2BR,L3BR,LIBST,L2BST,L3BST, FBP image respectively were about 7.5±2.1,8.4±2.7,9.2±2.2,9.4±2.2, 10.2±2.5,10.9±2.9,6.0±1.9.In low dose group:The image noise of LIBR image was about 8.0±1.2HU, L2BR was about 6.5±0.9HU, L3BR was about 5.3±0.7HU,L1BST was about 5.8±0.2HU, L2BST was about4.8±0.9HU, L3BST was about 3.9±0.6HU,FBP was about 16.2±4.1HU. The SNR of L1BR,L2BR,L3BR,L1BST,L2BST,L3BST,FBP image respectively were about 20.7±3.3,24.0±2.9,27.8±2.9,26.3±4.5,26.9±8.1, 33.1±6.9,12.9±3.3. The CNR of L1BR,L2BR,L3BR,L1BST,L2BST,L3BST, FBP image respectively were about 10.7±3.0,12.9±3.5,14.6±3.9,13.6±4.0,15.1±4.9, 17.4±5.1,6.6±2.3.There was significant difference in the image noise, SNR, and CNR (all, P<0.001) among all reconstruction algorithms for each group (standard dose group:Fnoise =126.47, FSNR=26.11,FCNR=15.50, all P< 0.001; low dose group:Fnoise=131.39, FSNR=35.24,FCNR=17.84, all P< 0.001). All IMR settings reduced image noise and increased SNRs and CNRs compared to FBP algorithm. For the comparison of IMR settings themselves for each group, we found the image noise exhibited a trend to decrease while SNR and CNR trended to increase from LIBR to L3BST in SD-group and SD-group. Meanwhile, image noise, SNR and CNR in LD-group showed the same trends with SD-group. Comparison between groups:(1) Standard dose group:there was no significant differences in image noise between L2BR and L3BR SNR and CNR showed no significant difference between continuous settings. In addition, there was no statistical differences in CNR between L2BR and L1BST, L1BST and L3BST, either (P=0.173,and P=0.261). However, there was significant difference among the other reconstruction algorithms. (2) Low dose group:there was showed no significant differences in the image noise among L2BR and L1BST, L3BR and L2BST, and between L2BST and L3BST. Meanwhile, IMR images showed no significant differences in the SNR between L2BR and L3BR, the CNR showed no significant differences between each two adjacent settings and among L2BR,L3BR,L1BST,and L2BST(P=0.051-0.230). However, there was significant difference among the other reconstruction algorithms.ConclusionsIMR can reduce image noise, increase SNR, CNR and improve image quality in hepatic CECT in both standard and low dose levels. Comparison of low dose protocol in hepatic CECT can satisfy the requirement of clinical diagnosis.L1BR; L2BR and L1BST are the most optimized settings for the standard dose protocol, while L1BST setting performs best for the low dose protocol. |
PDF Full Text Request