SPSung Bae Park

Sung Bae Park

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Sung Bae Park has a total of 141 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Enhancing Spinal Cord and Canal Segmentation in Degenerative Cervical Myelopathy : The Role of Interactive Learning Models with manual Click

    by Sangmin Han, Jae Keun Oh, Wonwoo Cho, Tae Joon Kim, Noah Hong, Sung Bae Park

    Objective : We aim to develop an interactive segmentation model that can offer accuracy and reliability for the segmentation of the irregularly shaped spinal cord and canal in degenerative cervical myelopathy (DCM) through manual click and model refinement.Methods : A dataset of 1444 frames from 294 magnetic resonance imaging records of DCM patients was used and we developed two different segmentation models for comparison : auto-segmentation and interactive segmentation. The former was based on U-Net and utilized a pretrained ConvNeXT-tiny as its encoder. For the latter, we employed an interactive segmentation model structured by SimpleClick, a large model that utilizes a vision transformer as its backbone, together with simple fine-tuning. The segmentation performance of the two models were compared in terms of their Dice scores, mean intersection over union (mIoU), Average Precision and Hausdorff distance. The efficiency of the interactive segmentation model was evaluated by the number of clicks required to achieve a target mIoU.Results : Our model achieved better scores across all four-evaluation metrics for segmentation accuracy, showing improvements of +6.4%, +1.8%, +3.7%, and -53.0% for canal segmentation, and +11.7%, +6.0%, +18.2%, and -70.9% for cord segmentation with 15 clicks, respectively. The required clicks for the interactive segmentation model to achieve a 90% mIoU for spinal canal with cord cases and 80% mIoU for spinal cord cases were 11.71 and 11.99, respectively.Conclusion : We found that the interactive segmentation model significantly outperformed the auto-segmentation model. By incorporating simple manual inputs, the interactive model effectively identified regions of interest, particularly in the complex and irregular shapes of the spinal cord, demonstrating both enhanced accuracy and adaptability.


  • Open Access

    Prediction Model for Low Bone Mineral Density in Cancer Survivors and Age-Matched Controls Using a Causal Bayesian Network: A Nationwide Population-Based Study in Korea

    by Sujeong Han, Sung-Bae Park, Sohee Oh, Bumjo Oh

    Background: With advances in cancer treatment, the number of cancer survivors has increased, bringing attention to long-term complications such as alterations in bone mineral density (BMD). Although survivors are at elevated risk for low BMD, prior studies have focused on specific cancer types and relied on traditional regression models, which are limited in capturing complex inter-variable relationships. This study aimed to examine the causal relationships among factors affecting BMD in cancer survivors and age-matched controls using causal Bayesian network (CBN) modeling. Methods: Data from the 2010-2011 Korea National Health and Nutrition Examination Survey (KNHANES) V were analyzed. We included 227 cancer survivors and 681 age- and sex-matched controls. Associations between BMD and variables such as age, sex, body composition, smoking, fracture history, and vitamin D were assessed using linear regression. A CBN model was then applied to evaluate probabilistic dependencies and potential causal relationships between variables and femoral neck BMD. Results: Among all participants, age, sex, smoking, fracture history, body fat percentage, muscle mass, and cancer history were significantly associated with femoral neck BMD. In cancer survivors, age (β = −0.032, P < .001) and sex (β = −0.680, P < .001) showed negative associations with BMD, whereas higher muscle mass (β = 0.073, P < .001) was a strong positive predictor. Smoking (β = −0.779, P = .005) and previous fractures (β = −0.507, P = .003) were also linked to lower BMD. The CBN model identified direct effects of age and muscle mass on BMD, with indirect effects from sex, smoking, and fracture history. Among women aged >60 years, greater muscle mass appeared particularly protective. Conclusion: Causal Bayesian network modeling identified muscle mass as a key modifiable factor influencing BMD among cancer survivors. These findings highlight the importance of muscle-preserving lifestyle interventions, including resistance exercise and adequate protein intake, in survivorship care. The CBN approach provides a framework for identifying individualized risk pathways and can support personalized bone-health management strategies in clinical practice.

  • Open Access

    Sex-Specific Differences and Risk Factors for Postoperative Urinary Retention after Spine Surgery with TIVA

    by Ansun Jeong, Mira Han, Noah Hong, Hae Chan Song, Tae Joon Kim, Sung Bae Park

    Objective : To identify sex-specific risk factors for postoperative urinary retention (POUR) in patients undergoing spine surgery with total intravenous anesthesia (TIVA) and explore differences between male and female patients.Methods : A retrospective chart review was conducted on 446 adult patients (268 males, 178 females) who underwent spine surgery with TIVA from January 1, 2019 to December 31, 2023. Propensity score matching was applied to adjust for confounding variables, resulting in 328 matched patients (164 males and 164 females). Multivariable logistic regression analyses were performed to identify independent risk factors for POUR. Residual urine volume ≥300 mL was used to define POUR.Results : Male patients had a higher incidence of POUR compared to female patients. Significant risk factors for POUR in males included myelopathy (adjusted odds ratio [aOR], 5.17; 95% confidence interval [CI], 1.11–24.15), benign prostatic hyperplasia (BPH) (aOR, 2.63; 95% CI, 1.02–6.78), and prolonged anesthesia time ≥240 minutes (aOR, 2.54; 95% CI, 1.13–5.72). No significant risk factors were found in female patients. Myelopathy and prolonged anesthesia were associated with an increased risk of POUR across the overall population.Conclusion : Male sex, myelopathy, BPH, and prolonged anesthesia time are significant risk factors for POUR following spine surgery with TIVA. Given these findings, particular attention should be paid to myelopathy, a history of BPH, and anesthesia time in male patients.


  • Open Access

    Identification of Bone Metastasis-Related Gene Networks in Lung Cancer: Implications for Bone Metabolism

    by Jungwoo Kim, Seong-Ho Park, Junho K. Hur, Sung Bae Park

    Background: This study aimed to infer a causal gene network associated with bone metastasis in lung cancer and to validate its reliability through experimental gene expression analysis.Methods: Using DNA microarray data from the Gene Expression Omnibus, we analyzed samples from primary lung cancer and those with bone metastasis. Commonly expressed genes in both groups were identified, and a causal network was inferred using Bayesian network inference with Java Objects based on the Bayesian Dirichlet score. To evaluate the network, we predicted the expression changes of downstream genes following knockdown of a key upstream gene and compared these predictions with mRNA expression levels in fatty acid desaturase 1 (FADS1)-knockdown lung cancer cells.Results: The genes FADS1, cardiotrophin-like cytokine factor 1 (CLCF1), chromosome 4 open reading frame 48, sushi, nidogen and EGF like domains 1, FK506-binding protein 15, and coenzyme Q10A (COQ10A) were identified as directly associated with lung cancer bone metastasis. Among them, FADS1 appeared to have a regulatory role, influencing downstream targets. Notably, CLCF1 and COQ10A showed significantly increased expression in FADS1-knockdown cells, consistent with the network’s predictions.Conclusions: These findings suggest that Bayesian network analysis is a reliable machine learning approach for uncovering causal gene relationships in cancer metastasis. Furthermore, FADS1 may serve as a potential therapeutic target in lung cancer bone metastasis. The validity of this network was supported by in vitro experiments using a lung cancer cell line.

  • Deletion of the POLR2J4 Gene in Breast Cancer Cells Promotes Bone Metastatic Tumor Growth

    by Seong-Ho Park, DA Hyun Yoon, Jungwoo Kim, Serk In Park, Junho K. Hur, Sung Bae Park

    Background/Aim: Multiple genetic alterations accumulate during the progression of bone metastasis. Previously, we identified POLR2J4 as a causal gene associated with breast cancer bone metastasis. This study aimed to investigate the roles of POLR2J4 in breast cancer bone metastasis in a murine model. Materials and Methods: We engineered and knocked out the POLR2J4 gene in luciferase-labeled MDA-MB-231 breast cancer cells using the CRISPR-spCas9 system. Parental or POLR2J4-knockout cells were injected into the systemic circulation of female nude mice. Development and growth of bone metastatic lesions were monitored by weekly bioluminescence imaging. After four weeks, mice were sacrificed for histological analysis of their bone lesions. In addition, transcriptomics analysis using RNA sequencing was performed in POLR2J4 knockout vs. parental cells to investigate gene expression changes. POLR2J4 knockout was confirmed by target sequencing and quantitative RT-PCR. Results: The POLR2J4 knockout group had significantly increased incidence of bone metastasis, compared with the control group (pn=28/group). However, no statistical significance was observed in the metastatic tumor size measured by bioluminescence signal intensities. Bioluminescence-positive metastatic lesions were confirmed for tumor cells microscopically. Among the genes differentially expressed between the knockout and control cells, CD74 was significantly up-regulated. Conclusion: POLR2J4 is a negative regulator of breast cancer bone metastasis, and CD74 up-regulation in the POLR2J4 knockout cells contributes to the pro-metastatic phenotype. Overall, our data warrant further investigation on the role of POLR2J4 as a causal gene of breast cancer bone metastasis.

  • Open Access

    Augmented prediction of vertebral collapse after osteoporotic vertebral compression fractures through parameter-efficient fine-tuning of biomedical foundation models

    by Sibeen Kim, Inkyeong Kim, Woon Tak Yuh, Sangmin Han, Choonghyo Kim, Young San Ko, Wonwoo Cho, Sung Bae Park

    Vertebral collapse (VC) following osteoporotic vertebral compression fracture (OVCF) often requires aggressive treatment, necessitating an accurate prediction for early intervention. This study aimed to develop a predictive model leveraging deep neural networks to predict VC progression after OVCF using magnetic resonance imaging (MRI) and clinical data. Among 245 enrolled patients with acute OVCF, data from 200 patients were used for the development dataset, and data from 45 patients were used for the test dataset. To construct an accurate prediction model, we explored two backbone architectures: convolutional neural networks and vision transformers (ViTs), along with various pre-trained weights and fine-tuning methods. Through extensive experiments, we built our model by performing parameter-efficient fine-tuning of a ViT model pre-trained on a large-scale biomedical dataset. Attention rollouts indicated that the contours and internal features of the compressed vertebral body were critical in predicting VC with this model. To further improve the prediction performance of our model, we applied the augmented prediction strategy, which uses multiple MRI frames and achieves a significantly higher area under the curve (AUC). Our findings suggest that employing a biomedical foundation model fine-tuned using a parameter-efficient method, along with augmented prediction, can significantly enhance medical decisions.


  • Open Access

    Correction: CRISPR activation identifies a novel miR-2861 binding site that facilitates the osteogenesis of human mesenchymal stem cells

    by Seong-Ho Park, Jungwoo Kim, Hee-Jin Yang, Ju Yeon Lee, Chi Heon Kim, Junho K. Hur, Sung Bae Park

  • Open Access

    CRISPR activation identifies a novel miR-2861 binding site that facilitates the osteogenesis of human mesenchymal stem cells

    by Seong-Ho B. Park, Jungwoo Kim, Hee-Jin Yang, Ju Yeon Lee, Chi Heon Kim, Junho K. Hur, Sung Bae Park

    We investigated the regulation of histone deacetylases (HDACs) by miR-2861 in the osteoblastic differentiation of human mesenchymal stem cells (MSCs) and miR-2861 binding site by CRISPR activation (CRISPRa). Transfection of miR-2861 into human MSCs was performed and the effect on osteoblast differentiation was analyzed. Using catalytically inactive Cas12a, the CRISPRa system induced targeted overexpression of endogenous miRNA and repressed the luciferase activities of reporters that contained functional miRNA target sites. The delivery of miR-2861 into MSCs enhanced osteoblast differentiation by decreased expressions of the HDAC1, 4 and 5 genes. The mechanism of HDAC5 repression by miR-2861 in humans has not been fully elucidated. To this end, the HDAC5 mRNA sequence was analyzed and a putative primate-specific miR-2861 binding site was identified in the 3’ untranslated region (3’-UTR). CRISPRa was applied to validate the putative binding site and an increase in endogenous miR-2861 was found to repress the expression of a reporter that contained the novel miR-2861 binding site. The delivery of miR-2861 to human MSCs enhanced osteoblast differentiation. In the 3’-UTR, the HDAC5 repression was mediated by the miR-2861 binding site, and miR-2861 promoted osteoblast differentiation via the inhibition of HDAC5 through a primate-specific miRNA binding site. Therefore, miRNAmiR-2861 with the CRISPRa methods might be a good biomaterial for osteogenesis augmentation.


  • Open Access

    Interactive Segmentation of Compressed Spinal Canal and Cord in Degenerative Cervical Myelopathy

    by Sung Bae Park, Sangmin Han, Woonwoo Cho, Chi Heon Kim

    Study Design: Retrospective Diagnostic Study Objective: We aim to develop an interactive segmentation model that can offer accuracy and reliability for the segmentation of the compressed spinal cord in degenerative cervical myelopathy (DCM). Setting:Boramae Medical Center, Korea. Methods: A dataset of 1,444 frames from 294 MRI records of DCM patients was used and we developed two different segmentation models for comparison: autosegmentation and interactive segmentation. The former was based on U-Net and utilized a pretrained ConvNeXT-tiny as its encoder. For the latter, we employed an interactive segmentation model structured by SimpleClick, a large model that utilizes a vision transformer as its backbone, together with simple fine-tuning. The segmentation performances of the two models were compared in terms of their DICE scores. The efficiency of the interactive segmentation model was evaluated by the number of clicks required to achieve a target mean intersection over union (mIoU). Results: The auto and interactive segmentation models with 10 clicks returned a 0.8226 and 0.9537 DICE score for cases involving canal segmentation and a 0.7363 and 0.7767 DICE score for cases involving cord mask segmentation alone, respectively. The required clicks for the interactive segmentation model to achieve a 90% mIoU for spinal canal with cord cases and 80% mIoU for spinal cord cases were 11.71 and 11.99, respectively. Conclusions: We found that the interactive segmentation model outperformed the autosegmentation model. Simple manual inputs can help the model identify a region of interest in the irregular shape of spinal cord. Sponsorship: No sponsorship

  • Open Access

    A Computed Tomography–Based Fracture Prediction Model With Images of Vertebral Bones and Muscles by Employing Deep Learning: Development and Validation Study

    by Sung Hye Kong, Wonwoo Cho, Sung Bae Park, Jaegul Choo, Jung Hee Kim, Sang Wan Kim, Chan Soo Shin

    Background: With the progressive increase in aging populations, the use of opportunistic computed tomography (CT) scanning is increasing, which could be a valuable method for acquiring information on both muscles and bones of aging populations. Objective: The aim of this study was to develop and externally validate opportunistic CT-based fracture prediction models by using images of vertebral bones and paravertebral muscles. Methods: The models were developed based on a retrospective longitudinal cohort study of 1214 patients with abdominal CT images between 2010 and 2019. The models were externally validated in 495 patients. The primary outcome of this study was defined as the predictive accuracy for identifying vertebral fracture events within a 5-year follow-up. The image models were developed using an attention convolutional neural network–recurrent neural network model from images of the vertebral bone and paravertebral muscles. Results: The mean ages of the patients in the development and validation sets were 73 years and 68 years, and 69.1% (839/1214) and 78.8% (390/495) of them were females, respectively. The areas under the receiver operator curve (AUROCs) for predicting vertebral fractures were superior in images of the vertebral bone and paravertebral muscles than those in the bone-only images in the external validation cohort (0.827, 95% CI 0.821-0.833 vs 0.815, 95% CI 0.806-0.824, respectively; P<.001). The AUROCs of these image models were higher than those of the fracture risk assessment models (0.810 for major osteoporotic risk, 0.780 for hip fracture risk). For the clinical model using age, sex, BMI, use of steroids, smoking, possible secondary osteoporosis, type 2 diabetes mellitus, HIV, hepatitis C, and renal failure, the AUROC value in the external validation cohort was 0.749 (95% CI 0.736-0.762), which was lower than that of the image model using vertebral bones and muscles (P<.001). Conclusions: The model using the images of the vertebral bone and paravertebral muscle showed better performance than that using the images of the bone-only or clinical variables. Opportunistic CT screening may contribute to identifying patients with a high fracture risk in the future.