SYSoon Ho Yoon

Soon Ho Yoon

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

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Soon Ho Yoon has a total of 553 co-authors. Here are the top co-authors by number of shared publications:

  • Jin Mo Goo

    Department of Radiology, Seoul National University Hospital and Seoul National University College of Medicine, 101 Daehak-ro, Jongno-gu, Seoul 03080, Korea

    33 shared publications
  • Jin Mo Goo

    Department of Radiology, Seoul National University Hospital, Seoul National University College of Medicine, 101, Daehak-ro, Jongno-gu, Seoul, 03080, Republic of Korea; Cancer Research Institute, Seoul National University College of Medicine, Seoul, 03080, Republic of Korea; Institute of Radiation Medicine, Seoul National University Medical Research Center, 101, Daehak-ro, Jongno-gu, Seoul, 03080, Republic of Korea

    21 shared publications
  • Hyungjin Kim

    Department of Radiology, Seoul National University Hospital and College of Medicine, 101 Daehak-ro, Jongno-gu, Seoul 03080, Korea

    19 shared publications
  • Chang Min Park

    Department of Radiology, Seoul National University Hospital, Seoul, Republic of Korea

    18 shared publications
  • Eui Jin Hwang

    Department of Radiology, Seoul National University Hospital and College of Medicine, 101 Daehak-ro, Jongno-gu, Seoul 03080, Korea

    12 shared publications
  • Hyungjin Kim

    Department of Radiology, Seoul National University Hospital, Seoul National University College of Medicine, Seoul, Republic of Korea

    11 shared publications

Latest Publications

  • Open Access

    Radiologist and AI performance in detecting mucus plugs on chest CT

    by Soonho Yoon, Seth J. Kligerman, Samuel R. Friedlander, Rūdolfs Latišenko, Jean-Paul Charbonnier, Harm Tiddens, David A. Lynch, Stephen M. Humphries

    Objectives: To compare automated mucus plug detection with radiologist evaluation, assess AI-assisted detection, and examine associations of algorithm-derived airway obstruction burden with COPD severity. Materials and methods: This retrospective analysis included 278 COPDGene participants selected across never-smokers, smokers with preserved ratio-impaired spirometry, and GOLD stages 0–4. An AI tool generated candidate plugs. Two thoracic radiologists independently identified plugs in unaided and AI-assisted sessions. A third thoracic radiologist adjudicated all candidates as the reference. Patient- and plug-level performance, observer agreement, false-positive detection, and rank-transformed associations with pulmonary function, CT parameters, and clinical severity were evaluated. Results: Reference mucus plugs were present in 114/278 participants (41%). Patient-level AI sensitivity was lower than R1 (68.4% vs 86.8%; p < 0.001) but not different from R2 (68.4% vs 79.8%; p = 0.141), whereas specificity was similar across approaches (95.7% vs 98.8%–100.0%). Plug-level AI sensitivity was lower than both readers (34.8% vs 58.9% for R1 and 55.6% for R2; both p < 0.001). AI assistance improved readers’ plug-level sensitivity (63.6–68.6%; p ≤ 0.004). AI-reader agreement was good (κ, 0.62–0.69), and inter-reader agreement improved with AI (κ, 0.70 vs 0.78; p < 0.001). These non-reference-positive AI detections (0.396/scan) were mostly marked bronchial wall thickening or partially occlusive mucus. They were independently associated with worse airflow limitation, CT emphysema, Pi10, and clinical severity after accounting for true-positive burden, improving model fit (Δ R ², 0.03–0.05). Conclusion: AI was less sensitive than radiologists but had similar specificity for mucus plug detection. AI assistance improved reader sensitivity and agreement. Additional algorithm-detected airway abnormalities remained associated with airway disease severity. Key Points: Question Automated mucus plug detection on chest CT may enable standardized COPD assessment, but independent validation against radiologist assessments remains limited . Findings Artificial intelligence showed lower sensitivity than radiologists but similar patient-level specificity on chest CT, and assistance improved sensitivity and inter-reader agreement . Clinical relevance Automated mucus plug analysis may support standardized airway obstruction burden assessment in COPD research and clinical trials while retaining radiologist oversight .


  • Determinants of Reader-LLM Interaction in Thoracic Radiology: Impact of Model Confidence and Reader Expertise

    by Jiyoung Song, Won Gi Jeong, Dae Hee Han, Hongseok Ko, Soon Ho Yoon, Hyungjin Kim, Phakhanith Naruetook, Wai Ling Leong, Soojin Kim, Hee Eun Moon, et al.

    Successful human–large language model collaboration is associated with model confidence and reader expertise in chest imaging, with rationale quality facilitating uptake of correct advice but increasing overreliance on incorrect suggestions.


  • Association of biological age acceleration with muscle aging phenotype in young and middle-aged adults: a prospective cohort study.

    by Kyeezu Kim, Yoosoo Chang, Hye-Mi Jang, Hyungjin Kim, Soonho Yoon, Seungho Ryu

    Loss of muscle mass and quality (myopenia and myosteatosis) are hallmarks of aging and predictors of disability, yet their relationship with biological aging in younger adults is unclear. This prospective cohort study included 165,015 adults (mean age 39.9) who underwent low-dose chest computed tomography (CT) between 2010 and 2020. A longitudinal subset of 61,669 participants had repeated scans (median follow-up 4.1 years). We adopted blood-based Phenotypic Age acceleration (PhenoAgeAccel) and chest radiograph-based biological age acceleration (CXR-AgeAccel). Myopenia and myosteatosis were defined as muscle area and density values more than two standard deviations below young sex-specific reference means. Multivariable multinomial logistic and Cox proportional hazard models estimated associations between biological age acceleration and prevalent or incident muscle abnormalities, adjusting for sociodemographic, lifestyle, and clinical factors. PhenoAgeAccel was associated with higher odds of myopenia (OR 1.16, 95% CI 1.08-1.25), myosteatosis (OR 1.24, 95% CI 1.18-1.29), and combined myopenia-myosteatosis (OR 1.51, 95% CI 1.41-1.63). In longitudinal analyses, PhenoAgeAccel was associated with increased risk of incident myopenia (HR 1.21, 95% CI 1.08-1.35), myosteatosis (HR 1.18, 95% CI 1.09-1.28), and the combined myopenia-myosteatosis (HR 1.29, 95% CI 1.08-1.55). Associations were consistent, though attenuated, with CXR-AgeAccel (OR 1.34, 95% CI 1.20-1.50; HR 1.18, 95% CI 0.96-1.45). Associations persisted among participants < 40 years. Accelerated biological aging measured from routine blood and imaging data was associated with early deterioration in muscle mass and quality. Biological age may represent a practical marker for identifying individuals at risk of early musculoskeletal decline.

  • Open Access

    ARKE: An ontology-driven framework for automated mapping of local radiology procedure terms to the LOINC-RadLex playbook using large language model

    by Sumin Lee, Kyulee Jeon, Joon Seok Lim, Yiju Park, Min Seong Kim, Ju Hyun Jin, Chang Hoon Han, Soon Ho Yoon, Seng Chan You

  • Explainable Biological Age from Automated Chest Radiography–based Organ Quantifications: qCXR-bioage

    by Yoosoo Chang, Suntae Park, Hyungjin Kim, Soon Ho Yoon, Hyun-Suk Jung, Jong-Min Kim, Samuel Cho, Ria Kwon, Ga-Young Lim, Dong Gil Kang, et al.

    Chest radiograph–derived biologic age, estimated through automated organ quantification, outperformed chronological age in predicting mortality among Korean adults undergoing regular health evaluations, with accelerated aging associated with multiple mortality types.


  • Estimation of whole‐body skeletal muscle volume using pectoralis muscle area and anthropometric measurements from chest CT in a Korean population

    by Kiook Baek, Soon Ho Yoon, Jung Hee Hong

    Introduction: Skeletal muscle is a critical organ influencing prognosis and overall health; however, the direct measurements of muscle mass are not easily attainable. This study aimed to estimate whole‐body skeletal muscle volume (WBM) using pectoralis muscle area (PMA) derived from a single chest‐CT slice and basic anthropometric data in a Korean population. Methods: We analyzed PET‐CT data from 201 adults (101 women, 100 men). WBM was segmented using deep learning–based software. The PMA was delineated and was divided into pectoralis major and minor. Sex‐specific LASSO regression models were developed: Model 1 included total PMA, Model 2A pectoralis major, Model 2B pectoralis minor, and Model 2C both muscles as separate predictors while Model 3 was based on L3 muscle area for comparison. All models additionally included age, height, weight, and waist circumference. Results: In females and in males, R² values were 0.76 and 0.75 (Model 1), 0.77 and 0.76 (Model 2C), and 0.84 and 0.81 (Model 3), respectively. Bland‐Altman bias were 0.5% (95% CI: −13.6%, 14.7%), 1.1% (95% CI: −20.4%, 22.6%) (Model 1), 0.5% (95% CI: −13.7%, 14.7%) 1.0% (95% CI: −20.1%, 22.2%) (Model 2C), 0.4% (95% CI: −12.2%, 13.0%) and 1.0% (95% CI: −18.2, 20.3) (Model 3), respectively. Conclusion: WBM can be reliably estimated using PMA measured in a single chest‐CT slice and basic body measurements (age, height, weight, and waist circumference). This method offers a practical and efficient surrogate for muscle assessment based on routinely acquired chest CT images.

  • Open Access

    Response to “Enhancing the Study of Long-Term Pulmonary Sequelae After COVID-19: Methodological Perspectives on Imaging and Clinical Integration”

    by Jiyoung Song, Soon Ho Yoon

    Song J, et al. Korean J Radiol. 2026 Jun;27(6):597-598. https://doi.org/10.3348/kjr.2026.0436

  • Open Access

    Artificial Intelligence-Based Three-Dimensional Cardiovascular CT Segmentation and Quantification for Congenital Heart Disease: What We Have Learned

    by Hyun Woo Goo, Soon Ho Yoon, Sang Joon Park, Seon Young Goo

    Artificial intelligence-based segmentation and quantification is a hot topic in cardiovascular imaging. However, the development of clinically useful artificial intelligence-based models for congenital heart disease is challenging owing to morphologic heterogeneity and limited data for training. Although cardiovascular computed tomography is increasingly used in evaluating patients with congenital heart disease, only a few reports have evaluated artificial intelligence-based segmentation algorithms for congenital heart disease. To be familiarized with this trend, we need to understand state-of-the-art segmentation techniques for cardiovascular digital twins as well as basic concept, current challenges, and potential solutions of artificial intelligence-based methods. Additionally, in this review, our learnings from the literature and our pilot study are summarized, and future directions are suggested.


  • CT Screening Challenges Amid Rising Threat of Lung Cancer in Individuals Who Have Never Smoked

    by Jiyoung Song, Eui Jin Hwang, Soon Ho Yoon, So Yeon Kim, Yeun-Chung Chang, Jin Mo Goo

    Routine low-dose CT screening for lung cancer in individuals who have never smoked is not justified due to uncertain mortality benefits and risks of overdiagnosis, overtreatment, and radiation, highlighting the need for personalized, risk-based strategies.