Jong-Keun Kim
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Jong-Keun Kim has a total of 45 co-authors. Here are the top co-authors by number of shared publications:
- Hyuk-Soo Han
Department of Orthopedic Surgery, Seoul National University Hospital, Seoul, Republic of Korea
3 shared publications Shinae Hwang
Department of Pharmacology, Chonnam National University Medical School, Hwasun, Korea
3 shared publications- Byung Woo Cho
Department of Orthopaedic Surgery, Gangnam Severance Hospital, Yonsei University College of Medicine, Seoul, Republic of Korea
2 shared publications Gaeun Kang
Division of Clinical Pharmacology , Chonnam National University Hospital, Gwangju, Korea
2 shared publications- Gaeun Kang
Department of Pharmacology, Chonnam National University Medical School, Hwasun 58128, Korea
2 shared publications - Hyuck Min Kwon
Department of Orthopaedic Surgery, Severance Hospital, Yonsei University College of Medicine, Seoul, Republic of Korea
2 shared publications
Latest Publications
- Open Access
- 2 March 2026
- Published by JMIR Publications Inc. in JMIR Medical Informatics
- Vol. 14, e86155
Enhanced Magnetic Resonance Imaging–Based Knee Cartilage Segmentation Using a Swin-UNet Conditional Generative Adversarial Network: Development and Validation Study
Background: Accurate segmentation of cartilage from magnetic resonance imaging (MRI) is crucial for the diagnosis and surgical planning of knee osteoarthritis. However, manual segmentation is time-consuming, and conventional computed tomography–based surgical systems are limited by their inability to visualize cartilage. Objective: This study aimed to develop a clinically targeted deep learning framework, the Swin-UNet conditional generative adversarial network (cGAN), for the automatic segmentation of femoral and tibial cartilage in MRI. We then evaluated its performance against conventional UNet, UNet cGAN, and Swin-UNet baseline models. Methods: Our dataset comprised 232 knee MRI scans. We conducted quantitative experiments on the proposed Swin-UNet cGAN model and compared the results with those of widely used UNet, UNet cGAN, and Swin-UNet models for femoral and tibial cartilage segmentation, using the Dice similarity coefficient, mean intersection over union, 95th percentile Hausdorff distance, and average symmetric surface distance. All performance metrics were statistically analyzed. In addition, the performance of the Swin-UNet cGAN model was evaluated on an external validation dataset. Results: The proposed Swin-UNet cGAN achieved the highest mean Dice similarity coefficient and intersection over union scores for both femoral and tibial cartilage segmentation, demonstrating performance statistically comparable to the best-performing baseline (UNet) in the tibia. Regarding distance metrics (average symmetric surface distance and 95th percentile Hausdorff distance), the proposed model significantly outperformed all baselines in the tibia while achieving results comparable to the UNet cGAN in the femur. It also maintained consistently high segmentation performance on both the internal test set and an external validation dataset. Conclusions: These findings indicate that the proposed Swin-UNet cGAN achieves more accurate knee cartilage segmentation than UNet, UNet cGAN, and Swin-UNet, particularly in terms of boundary accuracy, while maintaining promising generalizability performance across both internal testing and external validation cohorts. This MRI-based deep learning approach addresses critical limitations of computed tomography–based patient-specific instrumentation systems by providing cartilage visualization, potentially improving surgical precision and outcomes in total knee arthroplasty.
- 2 March 2026
- Published by JMIR Publications Inc. in JMIR Medical Informatics
- Vol. 14, e86155
Open AccessEnhanced Magnetic Resonance Imaging–Based Knee Cartilage Segmentation Using a Swin-UNet Conditional Generative Adversarial Network: Development and Validation Study
Background: Accurate segmentation of cartilage from magnetic resonance imaging (MRI) is crucial for the diagnosis and surgical planning of knee osteoarthritis. However, manual segmentation is time-consuming, and conventional computed tomography–based surgical systems are limited by their inability to visualize cartilage. Objective: This study aimed to develop a clinically targeted deep learning framework, the Swin-UNet conditional generative adversarial network (cGAN), for the automatic segmentation of femoral and tibial cartilage in MRI. We then evaluated its performance against conventional UNet, UNet cGAN, and Swin-UNet baseline models. Methods: Our dataset comprised 232 knee MRI scans. We conducted quantitative experiments on the proposed Swin-UNet cGAN model and compared the results with those of widely used UNet, UNet cGAN, and Swin-UNet models for femoral and tibial cartilage segmentation, using the Dice similarity coefficient, mean intersection over union, 95th percentile Hausdorff distance, and average symmetric surface distance. All performance metrics were statistically analyzed. In addition, the performance of the Swin-UNet cGAN model was evaluated on an external validation dataset. Results: The proposed Swin-UNet cGAN achieved the highest mean Dice similarity coefficient and intersection over union scores for both femoral and tibial cartilage segmentation, demonstrating performance statistically comparable to the best-performing baseline (UNet) in the tibia. Regarding distance metrics (average symmetric surface distance and 95th percentile Hausdorff distance), the proposed model significantly outperformed all baselines in the tibia while achieving results comparable to the UNet cGAN in the femur. It also maintained consistently high segmentation performance on both the internal test set and an external validation dataset. Conclusions: These findings indicate that the proposed Swin-UNet cGAN achieves more accurate knee cartilage segmentation than UNet, UNet cGAN, and Swin-UNet, particularly in terms of boundary accuracy, while maintaining promising generalizability performance across both internal testing and external validation cohorts. This MRI-based deep learning approach addresses critical limitations of computed tomography–based patient-specific instrumentation systems by providing cartilage visualization, potentially improving surgical precision and outcomes in total knee arthroplasty.
- Open Access
- 20 October 2025
- Published by JMIR Publications Inc. in JMIR Preprints
Enhanced Magnetic Resonance Imaging–Based Knee Cartilage Segmentation Using a Swin-UNet Conditional Generative Adversarial Network: Development and Validation Study (Preprint)
BACKGROUND Accurate segmentation of cartilage from magnetic resonance imaging (MRI) is crucial for the diagnosis and surgical planning of knee osteoarthritis. However, manual segmentation is time-consuming, and conventional computed tomography–based surgical systems are limited by their inability to visualize cartilage. OBJECTIVE This study aimed to develop a clinically targeted deep learning framework, the Swin-UNet conditional generative adversarial network (cGAN), for the automatic segmentation of femoral and tibial cartilage in MRI. We then evaluated its performance against conventional UNet, UNet cGAN, and Swin-UNet baseline models. METHODS Our dataset comprised 232 knee MRI scans. We conducted quantitative experiments on the proposed Swin-UNet cGAN model and compared the results with those of widely used UNet, UNet cGAN, and Swin-UNet models for femoral and tibial cartilage segmentation, using the Dice similarity coefficient, mean intersection over union, 95th percentile Hausdorff distance, and average symmetric surface distance. All performance metrics were statistically analyzed. In addition, the performance of the Swin-UNet cGAN model was evaluated on an external validation dataset. RESULTS The proposed Swin-UNet cGAN achieved the highest mean Dice similarity coefficient and intersection over union scores for both femoral and tibial cartilage segmentation, demonstrating performance statistically comparable to the best-performing baseline (UNet) in the tibia. Regarding distance metrics (average symmetric surface distance and 95th percentile Hausdorff distance), the proposed model significantly outperformed all baselines in the tibia while achieving results comparable to the UNet cGAN in the femur. It also maintained consistently high segmentation performance on both the internal test set and an external validation dataset. CONCLUSIONS These findings indicate that the proposed Swin-UNet cGAN achieves more accurate knee cartilage segmentation than UNet, UNet cGAN, and Swin-UNet, particularly in terms of boundary accuracy, while maintaining promising generalizability performance across both internal testing and external validation cohorts. This MRI-based deep learning approach addresses critical limitations of computed tomography–based patient-specific instrumentation systems by providing cartilage visualization, potentially improving surgical precision and outcomes in total knee arthroplasty.
- 20 October 2025
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessEnhanced Magnetic Resonance Imaging–Based Knee Cartilage Segmentation Using a Swin-UNet Conditional Generative Adversarial Network: Development and Validation Study (Preprint)
BACKGROUND Accurate segmentation of cartilage from magnetic resonance imaging (MRI) is crucial for the diagnosis and surgical planning of knee osteoarthritis. However, manual segmentation is time-consuming, and conventional computed tomography–based surgical systems are limited by their inability to visualize cartilage. OBJECTIVE This study aimed to develop a clinically targeted deep learning framework, the Swin-UNet conditional generative adversarial network (cGAN), for the automatic segmentation of femoral and tibial cartilage in MRI. We then evaluated its performance against conventional UNet, UNet cGAN, and Swin-UNet baseline models. METHODS Our dataset comprised 232 knee MRI scans. We conducted quantitative experiments on the proposed Swin-UNet cGAN model and compared the results with those of widely used UNet, UNet cGAN, and Swin-UNet models for femoral and tibial cartilage segmentation, using the Dice similarity coefficient, mean intersection over union, 95th percentile Hausdorff distance, and average symmetric surface distance. All performance metrics were statistically analyzed. In addition, the performance of the Swin-UNet cGAN model was evaluated on an external validation dataset. RESULTS The proposed Swin-UNet cGAN achieved the highest mean Dice similarity coefficient and intersection over union scores for both femoral and tibial cartilage segmentation, demonstrating performance statistically comparable to the best-performing baseline (UNet) in the tibia. Regarding distance metrics (average symmetric surface distance and 95th percentile Hausdorff distance), the proposed model significantly outperformed all baselines in the tibia while achieving results comparable to the UNet cGAN in the femur. It also maintained consistently high segmentation performance on both the internal test set and an external validation dataset. CONCLUSIONS These findings indicate that the proposed Swin-UNet cGAN achieves more accurate knee cartilage segmentation than UNet, UNet cGAN, and Swin-UNet, particularly in terms of boundary accuracy, while maintaining promising generalizability performance across both internal testing and external validation cohorts. This MRI-based deep learning approach addresses critical limitations of computed tomography–based patient-specific instrumentation systems by providing cartilage visualization, potentially improving surgical precision and outcomes in total knee arthroplasty.
- Open Access
- 1 January 2024
- Published by XMLink in Clinics in Orthopedic Surgery
- Vol. 16  (3) , 405-412
Do the Clinical and Radiological Features of Knees with Mucoid Degeneration of the Anterior Cruciate Ligament Differ According to Knee Osteoarthritis Status?
Kim JI, et al. Clin Orthop Surg. 2023 Aug;15:e64. https://doi.org/10.4055/cios23051
- 1 January 2024
- Published by XMLink in Clinics in Orthopedic Surgery
- Vol. 16  (3) , 405-412
Open AccessDo the Clinical and Radiological Features of Knees with Mucoid Degeneration of the Anterior Cruciate Ligament Differ According to Knee Osteoarthritis Status?
Kim JI, et al. Clin Orthop Surg. 2023 Aug;15:e64. https://doi.org/10.4055/cios23051
- Open Access
- 10 November 2022
- Published by Wiley in Journal of Spectroscopy
- Vol. 2022, 1-9
Simple Determination of Gemifloxacin Levels in Human Plasma using High-Performance Liquid Chromatography-Tandem Mass Spectrometry
Gemifloxacin, a broad-spectrum antibacterial agent of the fluoroquinolone class, is used to treat bacterial infections, including acute bacterial exacerbation of chronic bronchitis and community-acquired pneumonia. This study aimed to develop a simple and robust analysis of gemifloxacin in human plasma by high-performance liquid chromatography-tandem mass spectrometry (LC-MS/MS). The sample was prepared using simple protein precipitation procedures with acetonitrile and separated using the Gemini C18 column with a mobile phase (0.1 formic acid: 0.1 formic acid in acetonitrile=78: 22 (V/V)). Moxifloxacin was used as an internal standard. The mass spectrometer was operated in the positive ion mode using multiple reaction monitoring. Each precursor ion of gemifloxacin and moxifloxacin was monitored at m/z 390.1/402.1, and their product ions were monitored at m/z 372.3/384.4. The calibration curve showed linearity in 0.0055μg/mL with an appropriate correlation coefficient (0.99). The variation coefficient of intra- and interprecision values of gemifloxacin was <15. The intra- and interaccuracy values ranged from 85 to 115, except for the lower limit of quantification (accuracy range: 80120). The proposed method was performed with a simple preparation step, and moxifloxacin, which is easily accessible, was used as the internal standard. These results suggest that the present assay is a practical analytical method and, therefore, can be readily applied for analysis, including in pharmacokinetic studies and therapeutic drug monitoring of gemifloxacin.
- 10 November 2022
- Published by Wiley in Journal of Spectroscopy
- Vol. 2022, 1-9
Open AccessSimple Determination of Gemifloxacin Levels in Human Plasma using High-Performance Liquid Chromatography-Tandem Mass Spectrometry
Gemifloxacin, a broad-spectrum antibacterial agent of the fluoroquinolone class, is used to treat bacterial infections, including acute bacterial exacerbation of chronic bronchitis and community-acquired pneumonia. This study aimed to develop a simple and robust analysis of gemifloxacin in human plasma by high-performance liquid chromatography-tandem mass spectrometry (LC-MS/MS). The sample was prepared using simple protein precipitation procedures with acetonitrile and separated using the Gemini C18 column with a mobile phase (0.1 formic acid: 0.1 formic acid in acetonitrile=78: 22 (V/V)). Moxifloxacin was used as an internal standard. The mass spectrometer was operated in the positive ion mode using multiple reaction monitoring. Each precursor ion of gemifloxacin and moxifloxacin was monitored at m/z 390.1/402.1, and their product ions were monitored at m/z 372.3/384.4. The calibration curve showed linearity in 0.0055μg/mL with an appropriate correlation coefficient (0.99). The variation coefficient of intra- and interprecision values of gemifloxacin was <15. The intra- and interaccuracy values ranged from 85 to 115, except for the lower limit of quantification (accuracy range: 80120). The proposed method was performed with a simple preparation step, and moxifloxacin, which is easily accessible, was used as the internal standard. These results suggest that the present assay is a practical analytical method and, therefore, can be readily applied for analysis, including in pharmacokinetic studies and therapeutic drug monitoring of gemifloxacin.
- Open Access
- 1 January 2022
- Published by Chonnam National University Medical School in Chonnam Medical Journal
- Vol. 58  (1) , 18-23
N-Acetylcysteine Induces Apoptotic, Oxidative and Excitotoxic Neuronal Death in Mouse Cortical Cultures
N-acetylcysteine (NAC) has been used as an antioxidant to prevent oxidative cell death. However, we found NAC itself to induce neuronal death in mouse cortical cultures. Therefore, the current study was performed to investigate the mechanism of neuronal death caused by NAC. Cell death was assessed by measuring lactate dehydrogenase efflux to bathing media after 24-48 h exposure to NAC. NAC (0.1-10 mM) induced neuronal death in a concentration- and exposure time-dependent manner. However, NAC did not injure astrocytes even at a concentration of 10 mM. Also, 10 mM NAC markedly attenuated oxidative astrocyte death induced by 0.5 mM diethyl maleate or 0.25 mM H2O2. The NMDA receptor antagonist MK-801 (10 µM) markedly attenuated the neuronal death caused by 10 mM NAC, while NBQX did not affect the neuronal death. Cycloheximide (a protein synthesis inhibitor, 0.1 µg/mL) and z-VAD-FMK (a caspase inhibitor, 100 µM) also significantly attenuated neuronal death. Apoptotic features such as chromatin condensation, nuclear fragmentation, and caspase 3 activation were observed 1 h after the NAC treatment. The neuronal death induced by 1 or 10 mM NAC was significantly attenuated by the treatment with 100 µM Trolox or 1 mM ascorbic acid. NAC induced the generation of intracellular reactive oxygen species (ROS), as measured by the fluorescent dye 2',7'-dichlorofluorescein diacetate. The ROS generation was almost completely abolished by treatment with Trolox or ascorbic acid. These findings demonstrate that NAC can cause oxidative, apoptotic, and excitotoxic neuronal death in mouse neuronal cultures.
- 1 January 2022
- Published by Chonnam National University Medical School in Chonnam Medical Journal
- Vol. 58  (1) , 18-23
Open AccessN-Acetylcysteine Induces Apoptotic, Oxidative and Excitotoxic Neuronal Death in Mouse Cortical Cultures
N-acetylcysteine (NAC) has been used as an antioxidant to prevent oxidative cell death. However, we found NAC itself to induce neuronal death in mouse cortical cultures. Therefore, the current study was performed to investigate the mechanism of neuronal death caused by NAC. Cell death was assessed by measuring lactate dehydrogenase efflux to bathing media after 24-48 h exposure to NAC. NAC (0.1-10 mM) induced neuronal death in a concentration- and exposure time-dependent manner. However, NAC did not injure astrocytes even at a concentration of 10 mM. Also, 10 mM NAC markedly attenuated oxidative astrocyte death induced by 0.5 mM diethyl maleate or 0.25 mM H2O2. The NMDA receptor antagonist MK-801 (10 µM) markedly attenuated the neuronal death caused by 10 mM NAC, while NBQX did not affect the neuronal death. Cycloheximide (a protein synthesis inhibitor, 0.1 µg/mL) and z-VAD-FMK (a caspase inhibitor, 100 µM) also significantly attenuated neuronal death. Apoptotic features such as chromatin condensation, nuclear fragmentation, and caspase 3 activation were observed 1 h after the NAC treatment. The neuronal death induced by 1 or 10 mM NAC was significantly attenuated by the treatment with 100 µM Trolox or 1 mM ascorbic acid. NAC induced the generation of intracellular reactive oxygen species (ROS), as measured by the fluorescent dye 2',7'-dichlorofluorescein diacetate. The ROS generation was almost completely abolished by treatment with Trolox or ascorbic acid. These findings demonstrate that NAC can cause oxidative, apoptotic, and excitotoxic neuronal death in mouse neuronal cultures.
- Open Access
- 1 January 2020
- Published by Chonnam National University Medical School in Chonnam Medical Journal
- Vol. 56  (1) , 20-26
Fluoxetine Induces Apoptotic and Oxidative Neuronal Death Associated with The Influx of Copper Ions in Cultured Neuronal Cells
Hwang S, et al. Chonnam Med J. 2020 Jan;56(1):20-26. https://doi.org/10.4068/cmj.2020.56.1.20
- 1 January 2020
- Published by Chonnam National University Medical School in Chonnam Medical Journal
- Vol. 56  (1) , 20-26
Open AccessFluoxetine Induces Apoptotic and Oxidative Neuronal Death Associated with The Influx of Copper Ions in Cultured Neuronal Cells
Hwang S, et al. Chonnam Med J. 2020 Jan;56(1):20-26. https://doi.org/10.4068/cmj.2020.56.1.20
- 24 May 2019
- Published by SAGE Publications in Journal of Biomaterials Applications
- Vol. 34  (3) , 386-395
Temperature-sensitive anti-adhesive poloxamer hydrogel decreases fascial adhesion in total knee arthroplasty: A prospective randomized controlled study
A temperature sensitive anti-adhesive poloxamer (TAP) hydrogel has been developed to reduce post-operative adhesion. The purpose of this study was to evaluate the efficacy and safety of TAP hydrogel on fascial adhesion and clinical outcomes including range of motion, pain, and patient satisfaction after total knee arthroplasty. Sixty patients who underwent unilateral primary total knee arthroplasty for knee osteoarthritis were included and randomized to either TAP group or control group. After capsular closure, TAP hydrogel was applied on the fascial layer in 30 patients of the TAP hydrogel group, while normal saline was applied to 30 patients of the control group. The primary variable was range of motion at postoperative seven days, three months, and one year. As secondary outcomes, degree of skin fold, fascial thickness measured with ultra-sonography, postoperative pain visual analogue scale, total cumulative analgesics consumption, clinical scores, and degree of satisfaction were evaluated. There were no significant differences in range of motion after surgery between the two groups. However, the degree of suprapatellar skin folding in the TAP hydrogel group was less than that of the control group at six weeks after surgery (p = 0.041). There were no significant differences in postoperative fascial thickness, pain visual analogue scale, analgesics consumption, clinical scores, or degree of satisfaction. No surgical site complication was observed. The use of TAP hydrogel failed to enhance range of motion. However, application of TAP hydrogel on the fascial layer after capsular closure in total knee arthroplasty may decrease early postoperative fascial adhesion without any complications.
- 24 May 2019
- Published by SAGE Publications in Journal of Biomaterials Applications
- Vol. 34  (3) , 386-395
Temperature-sensitive anti-adhesive poloxamer hydrogel decreases fascial adhesion in total knee arthroplasty: A prospective randomized controlled study
A temperature sensitive anti-adhesive poloxamer (TAP) hydrogel has been developed to reduce post-operative adhesion. The purpose of this study was to evaluate the efficacy and safety of TAP hydrogel on fascial adhesion and clinical outcomes including range of motion, pain, and patient satisfaction after total knee arthroplasty. Sixty patients who underwent unilateral primary total knee arthroplasty for knee osteoarthritis were included and randomized to either TAP group or control group. After capsular closure, TAP hydrogel was applied on the fascial layer in 30 patients of the TAP hydrogel group, while normal saline was applied to 30 patients of the control group. The primary variable was range of motion at postoperative seven days, three months, and one year. As secondary outcomes, degree of skin fold, fascial thickness measured with ultra-sonography, postoperative pain visual analogue scale, total cumulative analgesics consumption, clinical scores, and degree of satisfaction were evaluated. There were no significant differences in range of motion after surgery between the two groups. However, the degree of suprapatellar skin folding in the TAP hydrogel group was less than that of the control group at six weeks after surgery (p = 0.041). There were no significant differences in postoperative fascial thickness, pain visual analogue scale, analgesics consumption, clinical scores, or degree of satisfaction. No surgical site complication was observed. The use of TAP hydrogel failed to enhance range of motion. However, application of TAP hydrogel on the fascial layer after capsular closure in total knee arthroplasty may decrease early postoperative fascial adhesion without any complications.
- Open Access
- 1 January 2019
- Published by XMLink in Journal of Clinical Neurology
- Vol. 15  (4) , 572-574
Association between the Levofloxacin Plasma Concentration and Neurological Adverse Events in an Elderly Patient
- 1 January 2019
- Published by XMLink in Journal of Clinical Neurology
- Vol. 15  (4) , 572-574
Open AccessAssociation between the Levofloxacin Plasma Concentration and Neurological Adverse Events in an Elderly Patient
- Open Access
- 1 January 2018
- Published by XMLink in Chonnam Medical Journal
- Vol. 54  (3) , 159-166
Effects of NADPH Oxidase Inhibitors and Mitochondria-Targeted Antioxidants on Amyloid β1-42-Induced Neuronal Deaths in Mouse Mixed Cortical Cultures
Hwang S, et al. Chonnam Med J. 2018 Sep;54(3):159-166. https://doi.org/10.4068/cmj.2018.54.3.159
- 1 January 2018
- Published by XMLink in Chonnam Medical Journal
- Vol. 54  (3) , 159-166
Open AccessEffects of NADPH Oxidase Inhibitors and Mitochondria-Targeted Antioxidants on Amyloid β1-42-Induced Neuronal Deaths in Mouse Mixed Cortical Cultures
Hwang S, et al. Chonnam Med J. 2018 Sep;54(3):159-166. https://doi.org/10.4068/cmj.2018.54.3.159