OBJECTIVE: To build a helical CT projection data restoration model at random low-dose levels. METHODS: We used a noise estimation module to achieve noise estimation and obtained a low-dose projection noise variance map, which was used to guide projection data recovery by the projection data restoration module. A filtering back-projection algorithm (FBP) was finally used to reconstruct the images. The 3D wavelet group residual dense network (3DWGRDN) was adopted to build the network architecture of the noise estimation and projection data restoration module using asymmetric loss and total variational regularization. For validation of the model, 1/10 and 1/15 of normal dose helical CT images were restored using the proposed model and 3 other restoration models (IRLNet, REDCNN and MWResNet), and the results were visually and quantitatively compared. RESULTS: Quantitative comparisons of the restored images showed that the proposed helical CT projection data restoration model increased the structural similarity index by 5.79% to 17.46% compared with the other restoration algorithms (P < 0.05). The image quality scores of the proposed method rated by clinical radiologists ranged from 7.19% to 17.38%, significantly higher than the other restoration algorithms (P < 0.05). CONCLUSION The proposed method can effectively suppress noises and reduce artifacts in the projection data at different low-dose levels while preserving the integrity of the edges and fine details of the reconstructed CT images.
Algorithms ; Artifacts ; Tomography, Spiral Computed ; Tomography, X-Ray Computed/methods*

OBJECTIVE: To quantitatively analyze image quality of two sets of phantom (CatPhan504 and Cheese) Megavoltage computed tomography (MVCT) images acquired by Helical Tomotherapy with three scanning modes (Fine, Normal and Coarse), and to explore and validate a semi-automatic quality assurance procedure for MVCT images of Helical Tomotherapy. METHODS: On Helical Tomotherapy, CatPan504 and Cheese phantoms were scanned with three pitch levels (Fine, Normal, Coarse: 4 mm, 8 mm, 12 mm/circle) respectively. Pylinac, Matlab and Eclipse were used to calculate and compare spatial resolution, noise level and low contrast resolution of images obtained under three scanning modes respectively. The spatial resolution can be evaluated by the blurring of line-pair CT value in the images of CatPhan504's CTP528 module. The noise level can be evaluated by the integral non-uniformity in the images of Cheese's uniformity module. the low contrast resolution can be evaluated by contrast-to-noise ratio of both phantoms' plug-in module, or visibility of the region of interest (Supra-Slice) in the images of CatPhan504's CTP515 module. RESULTS: Analyses on CatPhan504's line pair module(CTP528 module) showed that the first three line pairs(the gap size are 0.500 cm, 0.250 cm and 0.167 cm respectively) could be clearly observed but blurring began to occur from the fourth line pair(the gap size is 0.125 cm) under Coarse mode. Meanwhile, the first four line pairs were all observable under the Normal and Fine modes. Integral non-integrity index(the value negatively correlated with the noise level) were 0.155 7, 0.136 8 and 0.122 9 for Coarse, Normal and Fine modes respectively. None of the Supra-Slice in CatPhan504's CTP515 module could be observed under three imaging modes. Low contrast contrast-to-noise ratio of Cheese phantom was similar under three modes and the insert visibility exhibited nearly linear growth with the increasing difference between CT average value of the insert material and background. CONCLUSION Superiority and inferiority of three image modes in terms of the three image quality index was not consistent. Evaluation results above could provide reference for more rational decision on scanning modes selection of helical tomotherapy, which was based on image visualization demands in clinical practice. The proposed method could also provide guidance for similar image quality assessment and periodic quality assurance.
Cone-Beam Computed Tomography ; Phantoms, Imaging ; Radiotherapy, Intensity-Modulated ; Tomography, Spiral Computed ; Tomography, X-Ray Computed

PURPOSE: To correlate the enhancement pattern of hepatocellular carcinoma during the arterial phase of spiral CT with vascularity on angiography. MATERIALS AND METHODS: Forty-Two patients with hepatocellular carcinoma underwent spiral CT and angioigraphy. spiral CT was performed with a section thickness of 10mm and a table speedof 10 or 13 mm/sec. 120mL of contrast medium was injected at 3 mL/sec. Spiral CT scans during the arterial phase were obtained 35 seconds after the injection of contrast medium. CT findings of 78 lesions less than 4cm indiameter were correlated with angiographic findings. RESULTS: The attenuation of lesions was high(n=69),iso(n=5), and low(n=4) compared with liver parenchyma during the arterial phase of spiral CT. in lesions with high-, iso-, and low-attenuation during the arterial phase of spiral CT, hypervascularity on angiograms was foundin 63 of 69(91.3%), three of five(60%), and three of four lesions(75%), respectively. Six lesions with high-attenuation on the arterial phase of spiral CT were not seen on angiography. Two iso-attenuated and onelow-attenuated lesion were hypovascular on angiograms. CONCLUSION: The results of this study suggest that with some exceptions there was good correlation between the arterial phase of spiral CT and angiography.
Angiography* ; Carcinoma, Hepatocellular* ; Humans ; Liver ; Tomography, Spiral Computed* ; Tomography, X-Ray Computed*

PURPOSE: To compare the images of arterial phase (AP) and portal venous phase (PVP) in the evaluation of conspicuousness of lesion and peripancreatic arterial and venous opacification in pancreatic diseases. MATERIALS AND METHODS: Dual-phase spiral CT was performed in 37 patients with pancreatic adenocarcinoma and 21 patients with pancreatitis. CT scans were performed with 5mm collimation at 1 : 1 pitch table feed. Images of AP and PVP were obtained at 30 and 65 seconds after administration of contrast material, was initiated. Using a gradingsystem, images were analalysed for conspicuousness of lesion and vascular opacification(grade 1=good, grade2=fair, grade 3=poor). RESULTS: In pancreatic adenocarcinoma, 35 and 36 of 37 cases showed low attenuation on APand PVP, respectively. With regard to conspicuousness of tumour, PVP(mean grade : 1.24) was superior AP(mean grade: 1.43), but not significantly(p=0.0745). In arterial opacification, AP(mean grade : 1.03) was significantly superior to PVP(mean grade : 1.30, ; p=0.0051). In venous opacification, PVP(mean grade : 1.19) was significautlyto AP(mean grade : 2.41 ; p<0.0001). In pancreatitis, 14 and 15 of 21 cases showed localized hypo-attenuating lesion indicating necrosis or fluid collection, on AP and PVP, respectively. With regard to conspicuousness oflesion, PVP(mean grade : 1.61) was superior to AP(mean grade : 1.81), but not significantly(p=0.1088). In arterial opacification, AP(mean grade : 1.05) was significantly superior to PVP(mean grade: 1.38 ; p=0.0180). In venous opacification, PVP(mean grade : 1.10) was significantly superior to AP(mean grade : 2.33 ; p=0.0005). CONCLUSION: For the diagnosis and staging of pancreatic disease, dual-phase spiral CT in arterial and portal venous phase maybe recommendable. The portal venous phase of spiral CT seems, however, to be superior to the arterial phase because the lesion is more conspicuous and there is venous opacification.
Adenocarcinoma ; Diagnosis ; Humans ; Necrosis ; Pancreas* ; Pancreatic Diseases ; Pancreatitis ; Tomography, Spiral Computed* ; Tomography, X-Ray Computed

PURPOSE: To evaluate the effect of variable scanning protocols of computed tomography for evaluation of pre-implant site of the mandible through the comparison of the reformatted cross-sectional images of helical CT scans obtained with various imaging parameters versus those of conventional CT scans. MATERIALS AND METHODS: A dry mandible was imaged using conventional nonoverlapped CT scans with 1 mm slice thickness and helical CT scans with 1 mm slice thickness and pitches of 1.0, 1.5, 2.0, 2.5 and 3.0. All helical images were reconstructed at reconstruction interval of 1 mm. DentaScan reformatted images were obtained to allow standardized visualization of cross-sectional images of the mandible. The reformatted images were reviewed and measured separately by 4 dental radiologists. The image qualities of continuity of cortical outline, trabecular bone structure and visibility of the mandibular canal were evaluated and the distance between anatomic structures were measured by 4 dental radiologists. RESULTS: On image qualities of continuity of cortical outline, trabecular bone structure and visibility of the mandibular canal and in horizontal measurement, there was no statistically significant difference among conventional and helical scans with pitches of 1.0, 1.5 and 2.0. In vertical measurement, there was no statistically significant difference among the conventional and all imaging parameters of helical CT scans with pitches of 1.0, 1.5, 2.0, 2.5 and 3.0. CONCLUSION: The images of helical CT scans with 1 mm slice thickness and pitches of 1.0, 1.5 and 2.0 are as good as those of conventional CT scans with 1 mm slice thickness for evaluation of pre-dental implant site of the mandible. Considering the radiation dose and patient comfort, helical CT scans with 1 mm slice thickness and pitch of 2.0 is recommended for evaluation of pre-implant site of the mandible.
Dental Implants ; Humans ; Mandible ; Tomography, Spiral Computed ; Tomography, X-Ray Computed

PURPOSE: To correlate the preoperative staging of gastric adenocarcinoma using two-phase spiral CT withpathologic staging. MATERIALS AND METHODS: One hundred and eighty patients with gastric cancers confirmed duringsurgery underwent two-phase spiral CT, and were evaluated retrospectively. CT scans were obtained in the proneposition after ingestion of water. Scans were performed 35 and 80 seconds after the start of infusion of 120mL ofnon-ionic contrast material with the speed of 3mL/sec. Five mm collimation, 7mm/sec table feed and 5mmreconstruction interval were used. T- and N-stage were determined using spiral CT images, without knowledge of thepathologic results. Pathologic staging was later compared with CT staging. RESULTS: Pathologic T-stage was T1 in70 cases(38.9%), T2 in 33 (18.3%), T3 in 73 (40.6%), and T4 in 4 (2.2%). Type-I or IIa elevated lesions accountedfor 10 of 70 T1 cases(14.3%) and flat or depressed lesions(type IIb, IIc, or III) for 60 (85.7%). PathologicN-stage was N0 in 85 cases(47.2%), N1 in 42 (23.3%), N2 in 31 (17.2%), and N3 in 22 (12.2%). The detection rate ofearly gastric cancer using two-phase spiral CT was 100.0%(10 of 10 cases) among elevated lesions and 78.3%(47 of60 cases) among flat or depressed lesions. With regard to T-stage, there was good correlation between CT image andpathology in 86 of 180 cases (47.8%). Overstaging occurred in 23.3%(42 of 180 cases) and understaging in 28.9%(52of 180 cases). With regard to N-stage, good correlation between CT image and pathology was noted in 94 of 180cases (52.2%). The rate of understaging(31.7%, 57 of 180 cases) was higher than that of overstaging(16.1%, 29 of180 cases)(p<0.001). CONCLUSION: The detection rate of early gastric cancer using two-phase spiral CT was 81.4%,and there was no significant difference in detectability between elevated and depressed lesions. Two-phase spiralCT for determing the T- and N-stage of gastric cancer was not effective ; it was accurate in abont 50% of casesunderstaging tended to occur.
Adenocarcinoma* ; Eating ; Humans ; Pathology ; Retrospective Studies ; Stomach Neoplasms ; Tomography, Spiral Computed* ; Tomography, X-Ray Computed ; Water

PURPOSE: The aim of our study was to evaluate the usefulness of high-resolution CT scans in addition tohelical CT for characterizing a solitary pulmonary nodule. MATERIALS AND METHODS: Our study included 49 patientswith a solitary pulmonary nodule ; in each patient this was evaluated by both additional high-resolution CT andhelical scanning. Images were evaluated by three independent observers, each of whom read them twice : initiallywith helical CT images only and then with helical images plus high-resolution CT images. After analysis, theobservers recorded the following : histologic diagnosis, benignancy or malignancy of a nodule, and confidence intheir diagnosis (three scales). RESULTS: In differentiating benign and malignant nodules, the accuracy of helicalscans only was 75% (110/147 readings) whereas that of helical plus high-resolution CT scans was 82% (121/147readings) (p=0.001). Correct histologic diagnosis was made in 47% of cases (69/147 readings) when helical scansonly had been evaluated and in 48% of cases (71/147 readings) for which both helical and high-resolotion CT scanswere available (p=0.815). Diagnosis was more often highly confident on the basis of additional high-resolution CTscans (25%) than helical scans only (5%) (p=0.001) . CONCLUSION: By enhancing differential diagnosticaccuracy between benign and malignant nodules and by increasing confidence in the histologic diagnosis of apulmonary nodule, additional high-resolution CT scans are valuable for the evaluation of a solitary pulmonarynodule.
Diagnosis ; Humans ; Solitary Pulmonary Nodule* ; Tomography, Spiral Computed ; Tomography, X-Ray Computed

Congenital bronchial atresia is a rare anomaly, which results from a congenital focal obliteration of a proximal segmental or subsegmental bronchus, with normal development of the distal structures. The short atretic segment leads to the accumulation of mucus within the distal bronchi, forming a bronchocele and air trapping of the alveoli supplied by these bronchi. The diagnostic CT features include the presence of a branching opacity and the bronchocele, which radiate from the hilum and are surrounded by an area of hyperlucency.
Bronchi ; Bronchial Diseases ; Mucus ; Tomography, Spiral Computed ; Tomography, X-Ray Computed

PURPOSE: To evaluate the quantitative accuracy of three-dimensional (3D) images by means of comparing distance measurements on the 3D images with direct measurements of dry human skull according to slice thickness and scanning modes. MATERIALS AND METHODS: An observer directly measured the distance of 21 line items between 12 orthodontic landmarks on the skull surface using a digital vernier caliper and each was repeated five times. The dry human skull was scanned with a Helical CT with various slice thickness (3, 5, 7 mm) and acquisition modes (Conventional and Helical). The same observer measured corresponding distance of the same items on reconstructed 3D images with the internal program of V-works 4.0 TM (Cybermed Inc., Seoul, Korea). The quantitative accuracy of distance measurements were statistically evaluated with Wilcoxons' two-sample test. RESULTS: 11 line items in Conventional 3 mm, 8 in Helical 3 mm, 11 in Conventional 5 mm, 10 in Helical 5 mm, 5 in Conventional 7 mm and 9 in Helical 7 mm showed no statistically significant difference. Average difference between direct measurements and measurements on 3D CT images was within 2 mm in 19 line items of Conventional 3 mm, 20 of Helical 3 mm, 15 of Conventional 5 mm, 18 of Helical 5 mm, 11 of Conventional 7 mm and 16 of Helical 7 mm. CONCLUSION: Considering image quality and patient's exposure time, scanning protocol of Helical 5 mm is recommended for 3D image analysis of the skull in CT.
Cephalometry ; Humans ; Imaging, Three-Dimensional* ; Seoul ; Skull* ; Tomography, Spiral Computed ; Tomography, X-Ray Computed*

PURPOSE: Most hepatic metastases are hypovascular, and CT scanning during peak hepatic enhancement is thusimportant for the detection of hepatic lesions. The purpose of this study was to determine whether images obtainedby bolus tracking show greater hepatic enhancement. MATERIALS AND METHODS: We prospectively evaluated 101 patientswho underwent helical CT of the abdomen, using either a fixed 60-sec delay(n=50) or bolus tracking (n=51). For thelatter, we used a hepatic enhancement threshold of 50HU over baseline on monitor phase to determine a 6-sec delaybetween the monitor and diagnostic scanning phase. For all patients, three region-of-interest measurements wererecorded, one at each of the upper, middle, and lower levels of the liver; the measurements were averaged andsubtracted from the baseline density that measured CT values in three different hepatic segments at the mid-levelof the liver. We compared mean enhancement above the baseline of the liver between fixed 60-sec delay and bolustracking. RESULTS: A statistically significant difference in the enhancement level of the liver (upper, p=.001;middle, p=.001; lower, p=.003) was noted between fixed 60-sec delay (upper, 65.3+/-16HU; middle, 67.4+/-16.5HU;lower, 68.5(19.4HU) and bolus tracking (upper, 75.2+/-15.5HU; middle, 74.4+/-13.7HU;lower, 75.6+/-13.6HU). With fixeddelay, 86% of patients reached 50HU of enhancement, but with bolus tracking, 98% reached this level. For onenhancement value of 60HU, the corresponding figures were 64% and 86%. Mean delay for the transition betweenmonitoring scans and diagnostic scan initiation was 63.8+/-7.9(range, 48-79)secs. For two patients in whom theenhancement curve did not reach threshold (50HU), the default time was 70 sec; one subsequently failed to reachthis same threshold. CONCLUSION: Using the same amounts of contrast material, bolus tracking provides a greaterlevel of hepatic enhancement than a fixed 60-ses delay, and is thus helpful for the detection of hypovascularmasses such as metastatic lesions.
Abdomen ; Humans ; Liver ; Neoplasm Metastasis ; Prospective Studies ; Tomography, Spiral Computed ; Tomography, X-Ray Computed