• Journal of radiological science and technology
• /
• v.47 no.3
• /
• pp.175-182
• /
• 2024

Advanced modeled iterative reconstruction (ADMIRE) represents a repetitive reconstruction method that can adjust strength and kernel, each of which are known to affect computed tomography (CT) image quality. The aim of this study was to quantitatively analyze the noise and spatial resolution of CT images according to ADMIRE control factors. Patient images were obtained by applying ADMIRE strength 2 and 3, and kernel B40 and B59. For quantitative evaluations, the noise level, spatial resolution, and overall image quality were measured using coefficient of variation (COV), edge rise distance (ERD), and natural image quality evaluation (NIQE). The superior values for the average COV, ERD, and NIQE results were obtained for the ADMIRE reconstruction conditions of ADMIRE 2 + B40, ADMIRE 3 + B59, and ADMIRE3 + B59. NIQE, which represents the overall image quality based on no-reference, was about 6.04 when using ADMIRE 3 + B59, showing the best result among the reconstructed image acquisition conditions. The results of this study indicate that the ADMIRE strength and kernel chosen for use in ADMIRE reconstruction have a significant impact on CT image quality. This highlights the importance of adjusting to the control factors in consideration of the clinical environment.

• Korean Journal of Radiology
• /
• v.21 no.6
• /
• pp.660-669
• /
• 2020

Objective: To evaluate the accuracy of a deep learning-based automated segmentation of the left ventricle (LV) myocardium using cardiac CT. Materials and Methods: To develop a fully automated algorithm, 100 subjects with coronary artery disease were randomly selected as a development set (50 training / 20 validation / 30 internal test). An experienced cardiac radiologist generated the manual segmentation of the development set. The trained model was evaluated using 1000 validation set generated by an experienced technician. Visual assessment was performed to compare the manual and automatic segmentations. In a quantitative analysis, sensitivity and specificity were calculated according to the number of pixels where two three-dimensional masks of the manual and deep learning segmentations overlapped. Similarity indices, such as the Dice similarity coefficient (DSC), were used to evaluate the margin of each segmented masks. Results: The sensitivity and specificity of automated segmentation for each segment (1-16 segments) were high (85.5-100.0%). The DSC was 88.3 ± 6.2%. Among randomly selected 100 cases, all manual segmentation and deep learning masks for visual analysis were classified as very accurate to mostly accurate and there were no inaccurate cases (manual vs. deep learning: very accurate, 31 vs. 53; accurate, 64 vs. 39; mostly accurate, 15 vs. 8). The number of very accurate cases for deep learning masks was greater than that for manually segmented masks. Conclusion: We present deep learning-based automatic segmentation of the LV myocardium and the results are comparable to manual segmentation data with high sensitivity, specificity, and high similarity scores.

• Journal of radiological science and technology
• /
• v.40 no.2
• /
• pp.229-235
• /
• 2017

In abdominal Ultrasonography, the fatty liver is diagnosed through hepatic parenchymal echo increased parenchymal density and unclear blood vessel boundary, and according to many studies, abdominal Ultrasonography has 60~90% of sensitivity and 84~95% of specificity in diagnosis of fatty liver, but the result of Ultrasonography is dependent on operators, so there can be difference among operators, and quantitative measurement of fatty infiltration is impossible. Among examinees who same day received abdominal Ultrasonography and chest computed tomography (CT), patients who were diagnosed with a fatty liver in the Ultrasonography were measured with liver Hounsfield Units (HU) of chest CT imaging to analyze the accuracy of the fatty liver diagnosis. Among 720 subject examinees, those who were diagnosed with a fatty liver through abdominal Ultrasonography by family physicians were 448, which is 62.2%. The result of Liver HU measurement in the chest CT imaging of those who were diagnosed with a fatty liver showed that 175 out of 720 had the measured value of less than 40 HU, which is 24.3%, and 173 were included to the 175 among 448 who were diagnosed through Ultrasonography, so 98.9% corresponded. This indicates that the operators' subjective ability has a great impact on diagnosis of lesion in Ultrasonography diagnosis of a fatty liver, and that in check up chest CT, under 40 HU in the measurement of Liver HU can be used for reference materials in diagnosis of a fatty liver.

• Journal of Korean Neurosurgical Society
• /
• v.63 no.3
• /
• pp.386-396
• /
• 2020

Objective : To generate synthetic spine magnetic resonance (MR) images from spine computed tomography (CT) using generative adversarial networks (GANs), as well as to determine the similarities between synthesized and real MR images. Methods : GANs were trained to transform spine CT image slices into spine magnetic resonance T2 weighted (MRT2) axial image slices by combining adversarial loss and voxel-wise loss. Experiments were performed using 280 pairs of lumbar spine CT scans and MRT2 images. The MRT2 images were then synthesized from 15 other spine CT scans. To evaluate whether the synthetic MR images were realistic, two radiologists, two spine surgeons, and two residents blindly classified the real and synthetic MRT2 images. Two experienced radiologists then evaluated the similarities between subdivisions of the real and synthetic MRT2 images. Quantitative analysis of the synthetic MRT2 images was performed using the mean absolute error (MAE) and peak signal-to-noise ratio (PSNR). Results : The mean overall similarity of the synthetic MRT2 images evaluated by radiologists was 80.2%. In the blind classification of the real MRT2 images, the failure rate ranged from 0% to 40%. The MAE value of each image ranged from 13.75 to 34.24 pixels (mean, 21.19 pixels), and the PSNR of each image ranged from 61.96 to 68.16 dB (mean, 64.92 dB). Conclusion : This was the first study to apply GANs to synthesize spine MR images from CT images. Despite the small dataset of 280 pairs, the synthetic MR images were relatively well implemented. Synthesis of medical images using GANs is a new paradigm of artificial intelligence application in medical imaging. We expect that synthesis of MR images from spine CT images using GANs will improve the diagnostic usefulness of CT. To better inform the clinical applications of this technique, further studies are needed involving a large dataset, a variety of pathologies, and other MR sequence of the lumbar spine.

• Journal of Digital Convergence
• /
• v.13 no.9
• /
• pp.431-437
• /
• 2015

This convergence study analyzed the effectiveness of contrast agent reduction by normal saline solution dilution in the computed tomography of arteries of lower limb. 48 patients of 125 cc contrast agent and 30 patients of the same amount divided at a ratio of 7:3 for the contrast agent and normal saline solution were studied. The average attenuation coefficient(HU) and signal to noise ratio(SNR) of abdominal aorta, femoral artery, popliteal artery and posterior tibial artery at each image were evaluated quantitatively and the four criteria in the five point scale was conducted qualitatively by two radiologists and four radiological technologists. In the quantitative evaluation, both HU and SNR had high average score before dilation but there were no statistical significance by independent t-test(p>0.05). In the qualitative evaluation, there were a little differences in the average scores between 4.86~4.77 of original contrast agent and 4.83~4.67 of dilated contrast agent but there were no statistical significance(p>0.05). In the computed tomography of arteries of lower limb, the dilated contrast agent doesn't influence image quality and reduces overall contrast agent and lowers iodine content per unit of molecular therefore will contribute to decrease side effect of contrast agent.

• Journal of Korean Neurosurgical Society
• /
• v.64 no.6
• /
• pp.950-956
• /
• 2021

Objective : Psoas and masseter muscles are known markers of sarcopenia. However, the relative superiority of either muscle as a marker is unclear. Therefore, this study analyzed the two muscles in patients with a prognosis of traumatic brain injury (TBI). Methods : Patients with TBI visiting a regional trauma center between January 2017 and December 2018 were selected, and their medical records were reviewed. TBI patients with an abbreviated injury score (AIS) of 4 or 5 were selected. Patients with an AIS of 4 or 5 at the chest, abdomen, and extremity were excluded. Patients with a hospital stay of 1 to 2 days were excluded. Both muscle areas were measured based on the initial computed tomography. The psoas muscle index (PMI) and the masseter muscle index (MMI) were calculated by dividing both muscle areas by height in meters squared (cm²/m²). These muscle parameters along with other medical information were used to analyze mortality and the Glasgow outcome scale (GOS). Results : A total of 179 patients, including 147 males (82.1%), were analyzed statistically. The mean patient age was 58.0 years. The mortality rate was 16.8% (30 patients). The mean GOS score was 3.7. Analysis was performed to identify the parameters associated with mortality, which was a qualitative study outcome. The psoas muscle area (16.9 vs. 14.4 cm², p=0.028) and PMI (5.9 vs. 5.1 cm²/m², p=0.004) showed statistical differences between the groups. The PMI was also statistically significant as a risk factor for mortality in logistic regression analysis (p=0.023; odds ratio, 0.715; 95% confidence interval, 0.535-0.954). Quantitative analyses were performed with the GOS scores. Bivariate correlation analysis showed a statistically significant correlation between PMI and GOS scores (correlation coefficient, 0.168; p=0.003). PMI (p=0.004, variation inflation factor 1.001) was significant in multiple regression analysis. The masseter muscle area and MMI did not show significance in the study. Conclusion : Larger PMI was associated with statistically significant improved survival and GOS scores, indicating its performance as a superior prognostic marker. Further analyses involving a larger number of patients, additional parameters, and more precise settings would yield a better understanding of sarcopenia and TBI.

• Korean Journal of Radiology
• /
• v.19 no.6
• /
• pp.1187-1195
• /
• 2018

Objective: To compare correlations between pulmonary function test (PFT) results and different reconstruction algorithms and to suggest the optimal reconstruction protocol for computed tomography (CT) quantification of low lung attenuation areas and airways in healthy individuals. Materials and Methods: A total of 259 subjects with normal PFT and chest CT results were included. CT scans were reconstructed using filtered back projection, hybrid-iterative reconstruction, and model-based IR (MIR). For quantitative analysis, the emphysema index (EI) and wall area percentage (WA%) were determined. Subgroup analysis according to smoking history was also performed. Results: The EIs of all the reconstruction algorithms correlated significantly with the forced expiratory volume in one second (FEV1)/forced vital capacity (FVC) (all p < 0.001). The EI of MIR showed the strongest correlation with FEV1/FVC (r = -0.437). WA% showed a significant correlation with FEV1 in all the reconstruction algorithms (all p < 0.05) correlated significantly with FEV1/FVC for MIR only (p < 0.001). The WA% of MIR showed the strongest correlations with FEV1 (r = -0.205) and FEV1/FVC (r = -0.250). In subgroup analysis, the EI of MIR had the strongest correlation with PFT in both eversmoker and never-smoker subgroups, although there was no significant difference in the EI between the reconstruction algorithms. WA% of MIR showed a significantly thinner airway thickness than the other algorithms ($49.7{\pm}7.6$ in ever-smokers and $49.5{\pm}7.5$ in never-smokers, all p < 0.001), and also showed the strongest correlation with PFT in both ever-smoker and never-smoker subgroups. Conclusion: CT quantification of low lung attenuation areas and airways by means of MIR showed the strongest correlation with PFT results among the algorithms used, in normal subjects.

• Journal of the Korean Society of Radiology
• /
• v.14 no.7
• /
• pp.975-980
• /
• 2020

Cervical neural foraminal stenosis is a very common spinal disease that affects a relatively large number of people of all ages. However, since imaging methods that quantitatively provide neural foraminal stenosis are lacking, this study attempts to present quantitative measurement results by reconstructing 3D computed tomography images. Using a 3D reconstruction software, the surrounding bones were removed, including the spinous process, transverse process, and lamina of the cervical spine so that the neural foramen were well observed. Using Image J, a region of interest including the neural foramen area of the 3D image was set, and the number of pixels of the neural foramen area was measured. The neural foramen area was calculated by multiplying the number of measured pixels by the pixel size. In order to measure the widest area of the neural foramen, it was measured between 40-50 degrees in the opposite direction and 15-20 degrees toward the head. The measured cervical neural foramen area showed consistent measurement values. The largest measured area of the right neural foramen C5-6 was 12.21 ㎟, and after 2 years, the area was measured to be 9.95 ㎟, indicating that 18% stenosis had progressed. Since 3D reconstruction using axial CT scan images, no additional radiation exposure is required, and the area of stenosis can be objectively presented. In addition, it is good to explain to patients with neural stenosis while viewing 3D images, and it is considered a good method to be used in the evaluation of the progression of stenosis and post-operative evaluation.

• Journal of the Korean Society of Radiology
• /
• v.15 no.3
• /
• pp.301-306
• /
• 2021

The use of tin filters as a way to reduce the medical radiation in computed tomography (CT). However, due to the changed X-ray spectrum with the use of tin filters, disease diagnosis could be affected because it appears as images of different impressions from previous images. Therefore, this study evaluates the changes in images when using tin filter and high pitch in chest low-dose CT. In this study, images were acquired in groups of three for comparison. Group 1 did not apply to tin filter, and used the existing pitch 0.8. Group 2 used a tin filter, pitch 0.8, Group 3 used a tin filter, and pitch 2.5. To compare the image quality, the natural image quality evaluator (NIQE) and the blind/referenceless image quality evaluator (BRISQUE) were used among the blind quality evaluation factors depended on a no-reference basis. As a result, the NIQE values were low in the order of Group 1, Group 3, and Group 2. BRISQUE values were low in the order of Group 3, Group 2 and Group 1. This study confirms the superiority of images of tin filter and high pitch techniques in chest low-dose CT, which is considered to be a fundamental study for acquiring accurate images of patients with difficult breathing control.

• Journal of the Korean Society of Radiology
• /
• v.16 no.3
• /
• pp.203-209
• /
• 2022

One of the typical methods for lowering radiation dose while maintaining image quality of computed tomography (CT) is the use of model-based iterative reconstruction (MBIR). This study is to evaluate the image quality by adjusting the strength of the advanced modeled iterative reconstruction (ADMIRE), which is well known as a representative model of MBIR. The study was conducted using phantom, and CT images were obtained while adjusting the strength of ADMIRE in units of 1 to 5. Quantitative evaluation includes noise levels using coefficient of variation (COV) and contrast to noise ratio (CNR), as well as natural image quality evaluation (NIQE) and blind/referenceless image spatial quality evaluator (BRISQUE). As a result, in both noise level and blind quality evaluation results, the higher the strength of ADMIRE, the better the results were derived. In particular, it was confirmed that COV and CNR were improved 1.89 and 1.75 times at ADMIRE 5 compared to ADMIRE 1, respectively, and NIQE and BRISQUE were proved to be improved 1.35 and 1.22 times at ADMIRE 5 compared to ADMIRE 1, respectively. In conclusion, this study was proved that the reconstruction strength of ADMIRE had a great influence on the noise level and overall image quality evaluation of CT images.