• Journal of the Korean Society of Radiology
• /
• v.83 no.2
• /
• pp.344-359
• /
• 2022

Purpose To develop a denoising convolutional neural network-based image processing technique and investigate its efficacy in diagnosing breast cancer using low-dose mammography imaging. Materials and Methods A total of 6 breast radiologists were included in this prospective study. All radiologists independently evaluated low-dose images for lesion detection and rated them for diagnostic quality using a qualitative scale. After application of the denoising network, the same radiologists evaluated lesion detectability and image quality. For clinical application, a consensus on lesion type and localization on preoperative mammographic examinations of breast cancer patients was reached after discussion. Thereafter, coded low-dose, reconstructed full-dose, and full-dose images were presented and assessed in a random order. Results Lesions on 40% reconstructed full-dose images were better perceived when compared with low-dose images of mastectomy specimens as a reference. In clinical application, as compared to 40% reconstructed images, higher values were given on full-dose images for resolution (p < 0.001); diagnostic quality for calcifications (p < 0.001); and for masses, asymmetry, or architectural distortion (p = 0.037). The 40% reconstructed images showed comparable values to 100% full-dose images for overall quality (p = 0.547), lesion visibility (p = 0.120), and contrast (p = 0.083), without significant differences. Conclusion Effective denoising and image reconstruction processing techniques can enable breast cancer diagnosis with substantial radiation dose reduction.

• Journal of the Korean Society of Radiology
• /
• v.82 no.5
• /
• pp.1196-1206
• /
• 2021

Purpose To construct a standard dataset of contrast-enhanced CT images of liver tumors to test the performance and safety of artificial intelligence (AI)-based algorithms for clinical decision support systems (CDSSs). Materials and Methods A consensus group of medical experts in gastrointestinal radiology from four national tertiary institutions discussed the conditions to be included in a standard dataset. Seventy-five cases of hepatocellular carcinoma, 75 cases of metastasis, and 30-50 cases of benign lesions were retrieved from each institution, and the final dataset consisted of 300 cases of hepatocellular carcinoma, 300 cases of metastasis, and 183 cases of benign lesions. Only pathologically confirmed cases of hepatocellular carcinomas and metastases were enrolled. The medical experts retrieved the medical records of the patients and manually labeled the CT images. The CT images were saved as Digital Imaging and Communications in Medicine (DICOM) files. Results The medical experts in gastrointestinal radiology constructed the standard dataset of contrast-enhanced CT images for 783 cases of liver tumors. The performance and safety of the AI algorithm can be evaluated by calculating the sensitivity and specificity for detecting and characterizing the lesions. Conclusion The constructed standard dataset can be utilized for evaluating the machine-learning-based AI algorithm for CDSS.

• Journal of the Korean Society of Radiology
• /
• v.18 no.3
• /
• pp.187-201
• /
• 2024

Osteoporosis is a major health issue globally, often remaining undetected until a fracture occurs. To facilitate early detection, deep learning (DL) models were developed to classify osteoporosis using abdominal computed tomography (CT) scans. This study was conducted using retrospectively collected data from 3,012 contrast-enhanced abdominal CT scans. The DL models developed in this study were constructed for using image data, demographic/clinical information, and multi-modality data, respectively. Patients were categorized into the normal, osteopenia, and osteoporosis groups based on their T-scores, obtained from dual-energy X-ray absorptiometry, into normal, osteopenia, and osteoporosis groups. The models showed high accuracy and effectiveness, with the combined data model performing the best, achieving an area under the receiver operating characteristic curve of 0.94 and an accuracy of 0.80. The image-based model also performed well, while the demographic data model had lower accuracy and effectiveness. In addition, the DL model was interpreted by gradient-weighted class activation mapping (Grad-CAM) to highlight clinically relevant features in the images, revealing the femoral neck as a common site for fractures. The study shows that DL can accurately identify osteoporosis stages from clinical data, indicating the potential of abdominal CT scans in early osteoporosis detection and reducing fracture risks with prompt treatment.

• Korean Journal of Heritage: History & Science
• /
• v.57 no.1
• /
• pp.110-123
• /
• 2024

Onyanghaenggung Palace(temporary palace at Onyang) is an important cultural heritage that can substantially confirm the king's visiting at hot springs based on literature records such as <Ongungyeonggoedae(溫宮靈槐臺)>, <Oncheonhaenggungdo(溫泉行宮圖)> of 『Ongungsasil(溫宮事實)』, <Younggoedaedo(靈槐臺圖)>, 『Younggoedaegi(靈槐臺記)』 and cultural properties such as Yeonggoedae(靈槐臺) and Shinjeong Monument(神井碑). Through a photo taken by Hermann Sander in 1906, it can be confirmed that the hot springs(Tangsil building) at Onyanghaenggung Palace during the Joseon Dynasty was maintained until the early Japanese colonial period. The purpose of this study is to estimate the compositions of the hot springs(Tangsil building) in Onyanghaenggung Palace based on literature records and <Oncheonhaenggungdo>(1795). To achieve these purposes, we firstly examined the changes in Onyanghaenggung Palace and the hot springs (Tangsil building); secondly, the bathing behaviors of kings were reviewed; thirdly, we organized the architectural composition of the hot springs (Tangsil building) according to "Ongung Repair" of 『Ongungsasil (溫宮事實)』; and fourthly, by comparing Sander's photo in the early days of Japanese colonial rule, the architectural composition of the hot springs (Tangsil building) in the late Joseon Dynasty was examined. The results of this study are as follows. First, the hot springs(Tangsil building) of Onyanghaenggung Palace were continuously connected to the Onjeongsil(溫井室) in the reign of King Hyeonjong and maintained until 『Hoseo-eupji』 (1871) in the late Joseon Dynasty. It matches the photograph taken by Hermann Sander(1906) and <1912 Onyang Hot Springs in Asan City>(1912) of Korea Copyright Commission during the early Japanese colonial period. Second, the various king's bathing methods during the Joseon Dynasty were adopted such as washing, spilling and bathing head while sitting on a bathing platform or chair, or exposing the steam of hot spring water, dipping feet into the water and a half-body soaking bath below the navel immersed in water. Third, the stone bathtubs of hot springs(Tangsil building) are composed of the upper bath which was hot spring water gushes out from the northwest, bends to the east, enters the middle bath, and bends to the south to come out to the outside to gather in the lower bath. Around the stone bathtubs, pebble stones brought in from Taean were laid on the floor of the hot springs(Tangsil building). From the above considerations, the compositions of the Tangsil building in Onyang Temporary Palace is based on the king's approach from the main royal building, the king's bathing method and bathing tools, the bathing behavior of enlisted medical officers and bathing assistants, and each rooms mentioned in "Ongung Repair". By comparing it with Hermann Sander's photo, the architectural compositions of the hot springs(Tangsil building) can be estimated.

• Journal of the Korean Society of Radiology
• /
• v.84 no.5
• /
• pp.1123-1133
• /
• 2023

Purpose Our study aimed to evaluate the association between automated quantified body composition on CT and pulmonary function or quantitative lung features in patients with chronic obstructive pulmonary disease (COPD). Materials and Methods A total of 290 patients with COPD were enrolled in this study. The volume of muscle and subcutaneous fat, area of muscle and subcutaneous fat at T12, and bone attenuation at T12 were obtained from chest CT using a deep learning-based body segmentation algorithm. Parametric response mapping-derived emphysema (PRM^emph), PRM-derived functional small airway disease (PRM^fSAD), and airway wall thickness (AWT)-Pi10 were quantitatively assessed. The association between body composition and outcomes was evaluated using Pearson's correlation analysis. Results The volume and area of muscle and subcutaneous fat were negatively associated with PRM^emph and PRM^fSAD (p < 0.05). Bone density at T12 was negatively associated with PRM^emph (r = -0.1828, p = 0.002). The volume and area of subcutaneous fat and bone density at T12 were positively correlated with AWT-Pi10 (r = 0.1287, p = 0.030; r = 0.1668, p = 0.005; r = 0.1279, p = 0.031). However, muscle volume was negatively correlated with the AWT-Pi10 (r = -0.1966, p = 0.001). Muscle volume was significantly associated with pulmonary function (p < 0.001). Conclusion Body composition, automatically assessed using chest CT, is associated with the phenotype and severity of COPD.

• Tuberculosis and Respiratory Diseases
• /
• v.64 no.2
• /
• pp.80-86
• /
• 2008

Background: Park et al. developed the Korean reference equation for the measurement of diffusing capacity in 1985. However, the equation has not been widely used in Korea and foreign reference equations have been popularly used. We intended to compare the clinical usefulness and the accuracy of the the Korean reference equation (Park's equation) with that of the foreign equation (Burrows' equation) that is commonly used in Korea. Methods: 1. Evaluation of clinical usefulness; Among 1,584 patients who underwent diffusing capacity ($D_LCO$) at the Asan Medical Center from July to December 2006, group A subjects included 276 patients who had different interpretations of $D_LCO$ in trials employing Burrows' equation and Park's equation. Clinical assessment was decided by consensus of two respiratory physicians. In order to evaluate the clinical usefulness of Burrows' equation and Park's equation, agreement of clinical assessment and $D_LCO$ interpretation were measured. 2. Evaluation of accuracy; Group B subjects were 81 patients with interstitial lung disease (ILD) and 39 normal subjects. The 81 ILD patients were diagnosed following a surgical lung biopsy. The accuracy of diagnosing ILD as well as sensitivity and specificity were evaluated according to the use of the reference equations (Burrows' equation and Park's equation) for $D_LCO$. Results: Agreement between clinical assessment and interpretation of $D_LCO$ was 22% for the use of Burrows' equation and 78% for the use of Park's equation. The sensitivity and specificity of the Burrows' equation for diagnosing ILD were 64.2% and 100%. The sensitivity and specificity of the Park's equation for diagnosing ILD were 90.1% and 100%. The sensitivity of the Park's equation for diagnosing ILD was significantly higher than that of Burrows' equation (p<0.001). Conclusion: The Korean reference equation (Park's equation) was more clinically useful and had higher sensitivity for diagnosing ILD than the foreign reference equation (Burrows' equation).

• The Korean Journal of Nuclear Medicine Technology
• /
• v.14 no.2
• /
• pp.117-121
• /
• 2010

Purpose: In nuclear medicine study, there are two methods, preset count method and preset time method, to acquire static images. We usually use preset count method for static image in $^{99m}Tc$-DMSA renal scan, but occasionally use preset time method. In case of using preset count method, we always acquire same counts but it causes a difference of scan time. In case of using preset time method, it takes same scan time to acquire images but it causes different counts. Therefore, the purpose of this study is to investigate any differences of function and formal information in both kidney by acquisition counts Materials and Methods: From January 11, 2010 to March 31, 2010, we analyzed the 30 patients (M: 11, W: 19). who were examined by $^{99m}Tc$-DMSA scan and have one side of functioning kidney relatively between 40~60%. And the patients who have cold and hot region in image were analyzed but we did not accept images of patients when it was hard to divide kidney into cortex. There was no division between subjects and age of subjects is $14.83{\pm}22.07$ old. We used the BrightView gamma camera from PHILIPS. To analyze function and formal of kidney, we used JET stream release 3.0 version from PHILIPS. Using SPSS 12.0 program, we compared descriptive statistics and paired T-test. Images were acquired sequentially in the same parameters, but there are three methods which different from acquisition time and scan time, 100 kcounts, 300 kcounts and 7 minutes method (exceed 300 kcounts). To assess function and formal information of kidney, we measured renal relative function, geometric mean and size of kidney and analyzed each difference. Results: In case of renal relative function in both kidney, 100 kcounts method was $50.52{\pm}3.64%$. 300 kcounts method was $50.38{\pm}3.66%$ and 7 minutes method was $49.91{\pm}3.40%$ and there were no statistical significant differences between each method. In case of geometric mean, 100 kcounts method was $50.08{\pm}3.25%$. 300 kcounts method was $49.89{\pm}3.40%$ and 7 minutes method was $49.91{\pm}3.24%$. And also, there were no statistical significant differences. When comparing size of kidney, 100 kcounts method was $8.23{\pm}1.96$ cm. 300 kcounts method was $8.12{\pm}1.90$ cm and 7 minutes method was $8.35{\pm}1.97$ cm. In case of right kidney, 100 kcounts method was $7.91{\pm}1.88$ cm. 300 kcounts method was $8.12{\pm}1.90$ cm and 7 minutes method was $8.25{\pm}1.96$ cm. From those values, we recognized that there were significant differences each method (p<0.05). Conclusion: From results of this study, there were no statistical differences in renal relative function and geometric mean by acquisition counts. However, in shape of kidney, the more acquisition counts are increasing, the more size of kidney is getting big. And there were statistical significant differences. Therefore, to perform reliable quantitative result, preset count method is more desirable than preset time method. Especially, in case of a follow-up test, if we use preset time method, it will cause differences of formal results in kidney due to acquisition counts each time we examine patients.

• Journal of Chest Surgery
• /
• v.30 no.2
• /
• pp.164-171
• /
• 1997

Heart transplantation is now accepted as a definitive therapeutic modality in patients with terminal hear failure. The first successful heart transplantation in humans was done in 1967 and the first case in Korea was performed in november, 1992. Since the first case in 1992, more than 50 cases have been performed in Korea. A total of 20 patients underwent orthotopic heart transplantation since November, 1992 in Asan Medicla Center. The purpose of this study is to evaluate the early results and the follow-up course of 20 cases of heart transplantation done in Asan Medical Center. The average age of 20 patients was 39.9$\pm$11.8 years old(20~58). The mean follow-up duration was 14.4$\pm$11.2 months(1~41). All patients are alive till now. The blood type was identical in 14 and compatible in 6 patients. ihe original heart disease was dilated cardiomyopathy in 16, valvular heart disease in 2, ischemic cardiomyopathy in 1, and giant cell myocarditis in 1 patient. HLA cross matching for recipient and donor was done in 18 cases and the results were negative for T-cell and B-cell in 16 patients, pos tive for warm B-cell in 2 patients. Among 6 loci of A, B, and DR, one locus was matched in 8 cases, 2 loci in 5 cases, and 3 loci matched in 1 case. The number of acute allograft rejection averaged 2.8$\pm$0.5 (0~6) per case and the number of acute allograft rejection requiring treatment averaged 1.0$\pm$0.9 (1~3) per case. The time interval from operation to the first acute rejection requiring treatment was 35.5$\pm$20.4 days (5~60). Acute humoral rejection was suspected strongly in 1 case and was successfully treated. The left ventricular ejection fraction measured by echocardiography and/or MUGA scan was dramatically increased from 17.5$\pm$6.8 (9~32)% to 58.9$\pm$2.0 (55~62)% after heart transplantation. Temporary pacing was needed in 5 patients over 24 hours but normal sinus rhythm appeared within 7 days in all cases. One patient has been taken permanent pacemaker implantation due to complete AV block appearing 140 days after heart transplantaion. One patient had cyclosporine-associated n urotoxicity during the immediate postoperative period and was recovered after 27 hours. The heart transplantation of Asan Medical Center is on a developing stage but the early result is comparable to that of well established centers in other countries, even though the long-term follow-up result must be reevaluated. We can conclude that the heart transplantion is a promising therapeutic option in patients with terminal heart failure.

• The Korean Journal of Nuclear Medicine Technology
• /
• v.13 no.3
• /
• pp.67-75
• /
• 2009

Purpose: As to the analytical method of data, the various programs in which it is used for the quantitative rating of the Tl-201 myocardial perfusion SPECT are reported that there is a difference. Therefore, the measured value error of the mutual program is expected to be generated even if the quantitative analysis is made against data of the same patient. Using quantitative index that able to represent myocardial perfusion defect level, we aimed to determine correlation among three myocardial perfusion analysis programs 4DM (4DMSPECT), QPS (Quantitative Perfusion SPECT), ECT (Emory Cardiac Toolbox) that be used generally in most departments of Nuclear Medicine. Materials and Methods: We analyzed the 145 patients who were examined by Tl-201 gated myocardial perfusion SPECT in department of nuclear medicine at Asan Mediacal Center from December 1th 2008 to February 14th 2008. We sorted as normal group and abnormal group. Normal group consist of 80 patients (Male/Female=38/42, age:$65.1{\pm}9.9$) who have low possibility of cardiovascular disease. And abnormal group consist of 65 patients (Male/Female=45/20, age:$63.0{\pm}8.7$) who were diagnosed cardiovascular disease with reversible perfusion defect or fixed perfusion defect through myocardial perfusion SPECT results. Using the 4DM, QPS, and ECT programs, the total defect extent (TDE) such as LAD, LCX, RCA and the summed stress score (SSS) have been analysed for their correlations and statistical comparison with the paried t-test for the quantitative indices analysed from each group. Results: The correlation of 4DM:QPS, QPS:ECT, ECT:4DM each group result from 145 patients is 0.84, 0.86, 0.82 at SSS, 0.87, 0.84, 0.87 at TDE, and both index showed good correlation. In paired t-test and Bland-Altman analysis results showed no statistically significant difference in the comparison of QPS:ECT at the mean SSS and TDE, 4DM:QPS, ECT:4DM comparative analysis results showed statistically significant difference at SSS and TDE index. The correlation of 4DM:QPS, QPS:ECT, ECT:4DM program results from abnormal group (65 patients) is 0.72, 0.72, 0.70 at SSS and 0.77, 0.70, 0.77 at TDE and TDE and SSS has a good correlation. In abnormal group, paired t-test and Bland-Altman analysis results showed no statistically significant difference at QPS:ECT SSS (p=0.89) and TDE (p=0.23) comparison, 4DM:QPS, ECT:4DM comparative analysis results showed statistically significant difference at SSS and TDE index (p<0.01). In normal group (80 patients), paired t-test and Bland-Altman analysis results showed no statistically significant difference at QPS:ECT SSS (p=0.95) and TDE (p=0.73) comparison. And 4DM:QPS, ECT:4DM comparative analysis results showed statistically significant difference at SSS and TDE index (p<0.01). Conclusions: The perfusion defect of the Tl-201 myocardial perfusion SPECT was analyzed in not only the patient in whom it has the cardiovascular disease but also the patient in whom the possibility of the cardiovascular disease is few. In the comparison of the all group research, the mean of the TDE and SSS, 4DM was lower than QPS and ECT progrms. Each program showed good correlation and the results showed statistically significant difference. However, in this way, it is determined to be compatible about the analysis value in which the large-scale side between the programs uses each program a difference in a clinical in the Bland-Altman analyzed result in spite of the good correlation and cannot use. but, this analyzed result will be able to be usefully used as the reference material for the clinical read and is expected.

• The Korean Journal of Nuclear Medicine Technology
• /
• v.13 no.3
• /
• pp.56-60
• /
• 2009

Purpose: The various studies and efforts to develop program are in progress in the field of nuclear medicine for the purpose of reducing scan time. The Oncoflash is one of the programs used in whole body bone scan which allows to maintain the image quality while to reduce scan time. When Those applications are used in clinical setting, both the image quality and reduction of scan time should be considered, therefore, the purpose of this study was to determine the criteria for proper scan speed. Materials and Methods: The subjects of this study were the patients who underwent whole body bone scan at the departments of nuclear medicine in the Asan Medical Center located in Seoul from 1st to 10th, July, 2008. The whole body bone images obtained in the scan speed of 30cm/min were classified by the total counts into under 800 K, and over 800 K, 900 K, 1,000 K, 1,500 K, and 2,000 K. The image quality were assessed qualitatively and the percentages of those of 1,000K and under of total counts were calculated. The FWHM before and after applying the Oncoflash were analyzed using images obtained in $^{99m}Tc$ Flood and 4-Quadrant bar phantom in order to compare the resolution according to the amount of total counts by the application of the Oncoflash. Considering the counts of the whole body bone scan, the dosed 2~5 mCi were used. 152 patients underwent the measurement in which the counts of Patient Postioning Monitor (PPM) were measured with including head and the parts of chest which the starting point of whole body bone scan from 7th to 26th, August, 2008. The correlations with total counts obtained in the scan speed of 30cm/min among them were analyzed (The exclusion criteria were after over six hours of applying isotopes or low amount of doses). Results: The percentage of the whole body bone image which has the geometric average of total counts of under 1,000K among them obtained in the scan speed of 30cm/min were 17.6%(n=58) of 329 patients. The qualitative analysis of the image groups according to the whole body counts showed that the images of under 1,000K were assessed to have coarse particles and increased noises. The analysis on the FWHM of the images before and after applying the Oncoflash showed that, in the case of PPM counts of under 3.6 K, FWHM values after applying the Oncoflash were higher than that before applying the Oncoflash, whereas, in the case of that of over 3.6 K, the FWHM after applying the Oncoflash were not higher than that before applying the Oncoflash. The average of total counts at 2.5~3.0 K, 3.1~3.5 K, 3.6~4.0 k, 4.1~4.5 K, 4.6~5.0 K, 5.1~6.0 K, 6.1~7.0 K, and 7.1 K over (in PPM) were $965{\pm}173\;K$, $1084{\pm}154\;K$, $1242{\pm}186\;K$, $1359{\pm}170\;K$, $1405{\pm}184\;K$, $1640{\pm}376\;K$, $1,771{\pm}324\;K$, and $1,972{\pm}385\;K$, respectively and the correlations between the counts in PPM and the total counts of image obtained in the scan speed of 30 cm/min demonstrated strong correlation (r=.775, p<.01). Conclusions: In the case of PPM coefficient over 3.6 K, the image quality obtained in the scan speed of 30cm/min and after applying the Oncoflash was similar to that obtained in the scan speed of 15 cm/min. In the case of total counts over 1,000 K, it is expected to reduce scan time without any damage on the image quality. In the case of total counts under 1,000 K, however, the image quality were decreased even though the Oncoflash is applied, so it is recommended to perform the re-image in the scan speed of 15 cm/min.