Hong Seon Lee
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
Hong Seon Lee has a total of 50 co-authors. Here are the top co-authors by number of shared publications:
- Sungjun Kim
Department of Radiology, Gangnam Severance Hospital, Yonsei University College of Medicine, Seoul, Republic of Korea
8 shared publications - Jung Hyun Park
Department of Rehabilitation Medicine, Gangnam Severance Hospital, Rehabilitation Institute of Neuromuscular Disease, Yonsei University College of Medicine, Seoul, Korea
3 shared publications - Young Han Lee
Department of Radiology, Research Institute of Radiological Science and Center for Clinical Imaging Data Science (CCIDS), Yonsei University College of Medicine, Seoul, Korea
3 shared publications - Cheong Shin
Department of Integrative Medicine, The Graduate School, Yonsei University College of Medicine, Seoul, Republic of Korea
2 shared publications - Byoung Wook Choi
Department of Radiology, Research Institute of Radiological Science, Severance Hospital, Yonsei University College of Medicine, Seoul, Korea
1 shared publication Chaeri Park
Department of Radiology, Gangnam Severance Hospital, Yonsei University College of Medicine, Seoul, Republic of Korea
1 shared publication
Latest Publications
- 7 July 2026
- Published by Springer Nature in Journal of Imaging Informatics in Medicine
Radiologist Detection of Deficiencies in LLM-Generated Patient Communications About Radiology Reports: A Multi-Reader Study.
Patients increasingly use large language models (LLMs) to interpret radiology reports, yet the reliability of radiologist oversight in detecting errors in patient-facing LLM responses remains unquantified. This study evaluated the diagnostic accuracy of radiologists in detecting deficiencies in LLM-generated answers to simulated patient questions and quantified the incremental benefit of multi-reader oversight. In this retrospective multi-reader study, three board-certified radiologists independently evaluated 1200 LLM-generated answers (from three different models; 400 per model) to questions derived from 187 MIMIC-IV radiology reports. Answers were scored on a 3-point scale: 1 (clinically significant error), 2 (incomplete/ambiguous), and 3 (accurate). An adjudicated consensus served as the reference standard. Individual reader sensitivity, specificity, and multi-reader miss rates were compared. Of 1200 answers, 113 (9.4%) were deficient by consensus, including 18 clinically significant errors (1.5%) and 95 incomplete/ambiguous responses (7.9%). For clinically significant errors, readers missed 5.6-11.1% of cases. Across all deficiencies, individual reader sensitivity ranged from 81.4% (95% CI, 73.3-87.5) to 96.5% (95% CI, 91.3-98.6), with significant differences (McNemar p < 0.001). A single reader missed an average of 12.4% of deficiencies, whereas dual-reader review reduced the average miss rate to 0.6%; triple-reader review detected all deficiencies. Individual radiologist review of LLM-generated patient communications showed variable sensitivity, with some clinically significant errors remaining undetected. Dual-reader oversight significantly improves the safety and reliability of LLM-based communication tools.
- 7 July 2026
- Published by Springer Nature in Journal of Imaging Informatics in Medicine
Radiologist Detection of Deficiencies in LLM-Generated Patient Communications About Radiology Reports: A Multi-Reader Study.
Patients increasingly use large language models (LLMs) to interpret radiology reports, yet the reliability of radiologist oversight in detecting errors in patient-facing LLM responses remains unquantified. This study evaluated the diagnostic accuracy of radiologists in detecting deficiencies in LLM-generated answers to simulated patient questions and quantified the incremental benefit of multi-reader oversight. In this retrospective multi-reader study, three board-certified radiologists independently evaluated 1200 LLM-generated answers (from three different models; 400 per model) to questions derived from 187 MIMIC-IV radiology reports. Answers were scored on a 3-point scale: 1 (clinically significant error), 2 (incomplete/ambiguous), and 3 (accurate). An adjudicated consensus served as the reference standard. Individual reader sensitivity, specificity, and multi-reader miss rates were compared. Of 1200 answers, 113 (9.4%) were deficient by consensus, including 18 clinically significant errors (1.5%) and 95 incomplete/ambiguous responses (7.9%). For clinically significant errors, readers missed 5.6-11.1% of cases. Across all deficiencies, individual reader sensitivity ranged from 81.4% (95% CI, 73.3-87.5) to 96.5% (95% CI, 91.3-98.6), with significant differences (McNemar p < 0.001). A single reader missed an average of 12.4% of deficiencies, whereas dual-reader review reduced the average miss rate to 0.6%; triple-reader review detected all deficiencies. Individual radiologist review of LLM-generated patient communications showed variable sensitivity, with some clinically significant errors remaining undetected. Dual-reader oversight significantly improves the safety and reliability of LLM-based communication tools.
- Open Access
- 27 May 2026
- Published by MDPI AG in Healthcare
- Vol. 14 (11) , 1490
Patient-Facing Radiology Communication with LLMs: Calibration Deficit and the Metadata Paradox
Background/Objectives: Patients increasingly access radiology reports via online portals and frequently seek clarification. While Large Language Models (LLMs) may facilitate this communication, their clinical safety and reliability in this context remain largely uncharacterized. This study aimed to evaluate performance heterogeneity (the disparity between factual synthesis and interpretive reasoning), the Metadata Paradox (performance degradation triggered by demographic priors), and calibration characteristics in answering simulated patient questions derived from radiology reports. Methods: In this retrospective study, 2000 simulated inquiries were generated from 200 MIMIC-IV radiology reports based on an expert-refined 10-category taxonomy, categorized into factual tasks (e.g., terminology/anatomy) and interpretive tasks (e.g., diagnostic confidence/finding detail). Three LLMs (GPT-4o mini, Grok (v4-0709), Claude 3.5 Sonnet) generated 12,000 answers (with/without metadata). Quality was scored (1–3 scale) by Gemini 2.5 Flash, validated by three independent board-certified radiologists and finalized through four-specialist consensus adjudication (n = 1200). Performance and self-confidence calibration were assessed using Generalized Estimating Equations. Results: The LLM judge showed an overall agreement rate of 90.5% with the adjudicated ground truth. Grok and Claude 3.5 Sonnet significantly outperformed GPT-4o mini (p < 0.001); specifically, GPT-4o mini was associated with a 2.8-fold higher risk of failure compared to Grok (adjusted OR 2.83; 95% CI: 2.28–3.49; p < 0.001) and an absolute risk difference (ARD) of 8.4 percentage points. Accuracy reached its ceiling in factual tasks (Terminology: 98.1%) but was significantly lower in interpretive tasks (Diagnostic Confidence: 82.3%, p < 0.001). Metadata inclusion triggered the ‘Metadata Paradox,’ significantly increasing the risk of failure (OR 1.11; p = 0.044). A substantial calibration deficit (defined as the disconnect between self-confidence and accuracy) was observed; notably, the majority of safety-critical errors (Score 1: clinically significant misinformation; n = 131) were assigned high self-confidence (8/10; GPT-4o mini: 93.8%, Grok: 100%, Claude 3.5 Sonnet: 61.5%). Conclusions: Although LLMs accurately address factual queries, their consistent calibration deficit in safety-critical errors and susceptibility to stochastic stereotyping highlight the necessity of independent verification frameworks.- 27 May 2026
- Published by MDPI AG in Healthcare
- Vol. 14 (11) , 1490
Open AccessPatient-Facing Radiology Communication with LLMs: Calibration Deficit and the Metadata Paradox
Background/Objectives: Patients increasingly access radiology reports via online portals and frequently seek clarification. While Large Language Models (LLMs) may facilitate this communication, their clinical safety and reliability in this context remain largely uncharacterized. This study aimed to evaluate performance heterogeneity (the disparity between factual synthesis and interpretive reasoning), the Metadata Paradox (performance degradation triggered by demographic priors), and calibration characteristics in answering simulated patient questions derived from radiology reports. Methods: In this retrospective study, 2000 simulated inquiries were generated from 200 MIMIC-IV radiology reports based on an expert-refined 10-category taxonomy, categorized into factual tasks (e.g., terminology/anatomy) and interpretive tasks (e.g., diagnostic confidence/finding detail). Three LLMs (GPT-4o mini, Grok (v4-0709), Claude 3.5 Sonnet) generated 12,000 answers (with/without metadata). Quality was scored (1–3 scale) by Gemini 2.5 Flash, validated by three independent board-certified radiologists and finalized through four-specialist consensus adjudication (n = 1200). Performance and self-confidence calibration were assessed using Generalized Estimating Equations. Results: The LLM judge showed an overall agreement rate of 90.5% with the adjudicated ground truth. Grok and Claude 3.5 Sonnet significantly outperformed GPT-4o mini (p < 0.001); specifically, GPT-4o mini was associated with a 2.8-fold higher risk of failure compared to Grok (adjusted OR 2.83; 95% CI: 2.28–3.49; p < 0.001) and an absolute risk difference (ARD) of 8.4 percentage points. Accuracy reached its ceiling in factual tasks (Terminology: 98.1%) but was significantly lower in interpretive tasks (Diagnostic Confidence: 82.3%, p < 0.001). Metadata inclusion triggered the ‘Metadata Paradox,’ significantly increasing the risk of failure (OR 1.11; p = 0.044). A substantial calibration deficit (defined as the disconnect between self-confidence and accuracy) was observed; notably, the majority of safety-critical errors (Score 1: clinically significant misinformation; n = 131) were assigned high self-confidence (8/10; GPT-4o mini: 93.8%, Grok: 100%, Claude 3.5 Sonnet: 61.5%). Conclusions: Although LLMs accurately address factual queries, their consistent calibration deficit in safety-critical errors and susceptibility to stochastic stereotyping highlight the necessity of independent verification frameworks. - Open Access
- 1 January 2026
- Published by XMLink in Yonsei Medical Journal
- Vol. 67
Standardized Autologous Stromal Vascular Fraction for Knee Osteoarthritis: A 48-Week Double-Blind Randomized Controlled Trial
Cho HE, et al. Yonsei Med J. 2026 ;67:e72. https://doi.org/10.3349/ymj.2026.0138
- 1 January 2026
- Published by XMLink in Yonsei Medical Journal
- Vol. 67
Open AccessStandardized Autologous Stromal Vascular Fraction for Knee Osteoarthritis: A 48-Week Double-Blind Randomized Controlled Trial
Cho HE, et al. Yonsei Med J. 2026 ;67:e72. https://doi.org/10.3349/ymj.2026.0138
- 25 October 2025
- Published by Elsevier in Osteoarthritis and Cartilage
- Vol. 34 (1) , 167-174
Age- and sex-specific knee alignment patterns in symptomatic Korean patients: A cross-sectional deep learning analysis of 16,000 knees
- 25 October 2025
- Published by Elsevier in Osteoarthritis and Cartilage
- Vol. 34 (1) , 167-174
Age- and sex-specific knee alignment patterns in symptomatic Korean patients: A cross-sectional deep learning analysis of 16,000 knees
- Open Access
- 1 January 2025
- Published by XMLink in Journal of the Korean Society of Radiology
- Vol. 86 (5) , 608-623
AI for Lesion Detection in Musculoskeletal Radiology
김성준, et al. J Korean Soc Radiol. 2025 Sep;86(5):608-623. https://doi.org/10.3348/jksr.2025.0081
- 1 January 2025
- Published by XMLink in Journal of the Korean Society of Radiology
- Vol. 86 (5) , 608-623
Open AccessAI for Lesion Detection in Musculoskeletal Radiology
김성준, et al. J Korean Soc Radiol. 2025 Sep;86(5):608-623. https://doi.org/10.3348/jksr.2025.0081
- Open Access
- 1 January 2025
- Published by XMLink in The World Journal of Men's Health
- Vol. 43
Systematic Measurement of Membranous Urethral Length Using a Cross-Reference System: A Study on Interrater Variability
Kwon J, et al. World J Mens Health. 2025 Sep;43:e72. https://doi.org/10.5534/wjmh.250139
- 1 January 2025
- Published by XMLink in The World Journal of Men's Health
- Vol. 43
Open AccessSystematic Measurement of Membranous Urethral Length Using a Cross-Reference System: A Study on Interrater Variability
Kwon J, et al. World J Mens Health. 2025 Sep;43:e72. https://doi.org/10.5534/wjmh.250139
- Open Access
- 28 February 2023
- Published by Korean Pain Society in The Korean Journal of Pain
- Vol. 36 (2) , 208-215
Feasibility and reliability of various morphologic features on magnetic resonance imaging for iliotibial band friction syndrome
Jin Kyem Kim, Taeho Kim, Hong Seon Lee, and Dong Kyu Kim. Korean J Pain 2023;36:208-15. https://doi.org/10.3344/kjp.22399
- 28 February 2023
- Published by Korean Pain Society in The Korean Journal of Pain
- Vol. 36 (2) , 208-215
Open AccessFeasibility and reliability of various morphologic features on magnetic resonance imaging for iliotibial band friction syndrome
Jin Kyem Kim, Taeho Kim, Hong Seon Lee, and Dong Kyu Kim. Korean J Pain 2023;36:208-15. https://doi.org/10.3344/kjp.22399
- Open Access
- 31 January 2020
- Published by The Korean Society of Radiology in Journal of the Korean Society of Radiology
- Vol. 81 (1) , 21-40
Optimization of MRI Protocol for the Musculoskeletal System
Magnetic resonance imaging (MRI) is an essential modality for the diagnosis of musculoskeletal system defects because of its higher soft-tissue contrast and spatial resolution. With the recent development of MRI-related technology, faster imaging and various image plane reconstructions are possible, enabling better assessment of three-dimensional musculoskeletal anatomy and lesions. Furthermore, the image quality, diagnostic accuracy, and acquisition time depend on the MRI protocol used. Moreover, the protocol affects the efficiency of the MRI scanner. Therefore, it is important for a radiologist to optimize the MRI protocol. In this review, we will provide guidance on patient positioning; selection of the radiofrequency coil, pulse sequences, and imaging planes; and control of MRI parameters to help optimize the MRI protocol for the six major joints of the musculoskeletal system.
- 31 January 2020
- Published by The Korean Society of Radiology in Journal of the Korean Society of Radiology
- Vol. 81 (1) , 21-40
Open AccessOptimization of MRI Protocol for the Musculoskeletal System
Magnetic resonance imaging (MRI) is an essential modality for the diagnosis of musculoskeletal system defects because of its higher soft-tissue contrast and spatial resolution. With the recent development of MRI-related technology, faster imaging and various image plane reconstructions are possible, enabling better assessment of three-dimensional musculoskeletal anatomy and lesions. Furthermore, the image quality, diagnostic accuracy, and acquisition time depend on the MRI protocol used. Moreover, the protocol affects the efficiency of the MRI scanner. Therefore, it is important for a radiologist to optimize the MRI protocol. In this review, we will provide guidance on patient positioning; selection of the radiofrequency coil, pulse sequences, and imaging planes; and control of MRI parameters to help optimize the MRI protocol for the six major joints of the musculoskeletal system.