Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


See Young Lee has a total of 56 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Efficacy and Safety of a Novel Guidewire for Selective Biliary Cannulation: A Multicenter, Prospective Randomized Controlled Study

    by Eui Joo Kim, Huapyong Kang, Young Heon Han, Yeon Suk Kim, See Young Lee, Sung Ill Jang, Kyong Joo Lee, Dong Hee Koh, Jin Lee, Jae Hee Cho, et al.

    Background/Aims: Guidewire selection is critical for endoscopic retrograde cholangiopancreatography (ERCP) in pancreatobiliary diseases. Our study evaluated the noninferiority of novel guidewire compared with conventional guidewire that is widely used in clinical practice.Methods: A total of 198 patients with a native papilla undergoing ERCP at three tertiary hospitals were randomized in a 1:1 ratio to the test or control group. The primary endpoint was the selective biliary cannulation success rate. The secondary endpoints were the cannulation time, the incidence of difficult cannulation, and adverse events, including post-ERCP pancreatitis (PEP). Difficult cannulation was defined as a cannulation time of >5 minutes, more than five cannulation attempts, or at least one unintentional pancreatic duct cannulation.Results: Among 198 patients analyzed (test: 98; control: 100), baseline characteristics were comparable. Selective biliary cannulation success was 99.0% in the test group and 96.0% in the control group; the between-group difference was 3.0 percentage points (95% CI, -2.2 to 8.9), with the lower CI bound above the prespecified -11% non-inferiority margin. PEP incidence was similar between groups (5.1% vs. 4.0%, p=0.746). However, difficult cannulation occurred more frequently in the test group than in the control group (29.6% vs. 15.0%; p=0.017), and the cannulation time was longer in the test group (5.7 vs. 3.4 minutes; p=0.011).Conclusions: The novel guidewire met the prespecified non-inferiority criterion for selective biliary cannulation success compared with the conventional guidewire. Adverse event rates, including PEP, were similar between the groups. The novel guidewire was associated with a longer cannulation time.

  • Open Access

    Improving Radiology Report Error Detection Using a Multipass Large Language Model: Framework Development and Validation

    by Songsoo Kim, Seungtae Lee, See Young Lee, Joonho Kim, Keechan Kan, Hyunji Lee, Dukyong Yoon

    Background: Large language model (LLM) proofreaders for radiology reports generate many false positives (FPs) due to the low prevalence of errors. Objective: This study aimed to determine whether an optimized LLM framework could improve both precision and cost-efficiency without compromising error detection capability. Methods: In this retrospective study, 1000 radiology reports (radiography, ultrasonography, computed tomography, and magnetic resonance imaging; 250 each) were sampled from the Medical Information Mart for Intensive Care III database. Two public chest radiography corpora (CheXpert and Open-i) served as external test sets. Three LLM frameworks were evaluated: single-prompt detector (framework 1); report extractor plus single-prompt detector (framework 2); and extractor, detector, and FP verifier (framework 3). Precision for each framework was assessed using positive predictive value (PPV) and detected errors per 1000 reports. Overall efficiency was estimated using model inference costs and reviewer labor costs. Results: PPV increased from 0.063 (95% CI 0.036‐0.101) in framework 1 to 0.079 (95% CI 0.049‐0.118) in framework 2 and 0.159 (95% CI 0.090‐0.252) in framework 3 ( P <.001). Despite improved PPV, detected errors remained stable (detected errors per 1000 reports: 12‐14). Human review burden decreased from 192 to 88 reports. Framework 3 also reduced model inference costs to US $5.57 per 1000 reports (vs US $9.72 and US $6.85 for frameworks 1 and 2; 42.6% and 18.5% reductions, respectively). External validation confirmed similar improvements. Qualitative analysis revealed that remaining FPs in framework 3 were largely confined to cases requiring deep clinical context (clinically equivalent rephrasing: 53%; unsupported discrepancy assertions: 43%). By eliminating structural FPs (eg, section mismatches and lexical errors: 0%), the framework effectively shifted the quality assurance burden to a smaller set of ambiguous cases, enabling a targeted human-in-the-loop workflow. Conclusions: The multipass LLM improved the precision and cost-efficiency of radiology report error detection in real-world, low-error prevalence settings. The framework demonstrates the feasibility of synergistic artificial intelligence–radiologist collaboration and provides a cost-effective and scalable approach to artificial intelligence–assisted quality assurance in both radiological practice and research.


  • Open Access

    Improving Radiology Report Error Detection Using a Multi-Pass LLM Framework (Preprint)

    by Songsoo Kim, Seungtae Lee, See Young Lee, Joonho Kim, Keechan Kan, Dukyong Yoon

    BACKGROUND Large language model (LLM) proofreaders for radiology reports generate many false positives (FP) due to the low prevalence of errors. OBJECTIVE This study aimed to determine whether an optimized LLM framework could improve both precision and cost-efficiency without compromising error detection capability. METHODS In this retrospective study, 1,000 radiology reports (radiography, ultrasonography, CT, and MRI; 250 each) were sampled from the Medical Information Mart for Intensive Care III (MIMIC-III) database. Two public chest radiography corpora (CheXpert and Open-i) served as external test sets. Three LLM frameworks were evaluated: single-prompt detector (Framework 1); report extractor plus single-prompt detector (Framework 2); and extractor, detector, and false positive verifier (Framework 3). Precision for each framework was assessed using positive predictive value (PPV) and detected errors per 1,000 reports (DE/1k). Overall efficiency was estimated using model inference computational costs. RESULTS PPV increased from 0.063 [95% CI, 0.036–0.101] in Framework 1 to 0.079 (0.049–0.118) in Framework 2 and 0.159 (0.090–0.252) in Framework 3 (P<.001). Despite improved PPV, detected errors remained stable (DE/1k: 12–14). Human review burden decreased from 192 to 88 reports. Framework 3 also reduced costs to $5.58 per 1,000 reports (vs $9.72 and $6.85 for Frameworks 1 and 2; 42.6% and 18.5% reductions). External validation confirmed similar improvements. CONCLUSIONS A three-pass LLM framework more than doubled precision and halved the cost of radiology report error detection without compromising error detection capability, offering sustainable strategies for AI-assisted quality assurance in both radiological practice and research.

  • Open Access

    Diagnosis and Management of Intraductal Papillary Mucinous Neoplasms: Focusing on Precancerous Lesions

    by See Young Lee, Chan Min Jung, Jung Hyun Jo, Sung Ill Jang, Jae Hee Cho

    See Young Lee, Chan Min Jung, Jung Hyun Jo, Sung Ill Jang, Jae Hee Cho. Journal of Digestive Cancer Research 2025;13:120-30. https://doi.org/10.52927/jdcr.2025.13.2.120


  • Open Access

    Disulfiram/Copper Complex Induces Cytotoxicity in Pancreatic Cancer Cells and 5-Fluorouracil-Resistant Cells through Nuclear Factor E2-Related Factor-2 Suppression and Reactive Oxygen Species Modulation

    by Eun Kyoung Kim, Cheong Ran Je, Sung Ill Jang, Jung Hyun Jo, See Young Lee, Young Ju Lee, Jae Hee Cho

    Pancreatic ductal adenocarcinoma (PDAC) is a challenging cancer to treat and has a poor prognosis and limited treatment options. In this study, the anticancer effects of disulfiram combined with copper (DSF/Cu) on PDAC cells, including those resistant to 5-fluorouracil, was assessed. Human pancreatic cancer cells (BxPC-3 and CFPAC-1) and their 5-fluorouracil-resistant (5FUR) counterparts were treated with DSF/Cu to assess cytotoxicity. Expression levels of nuclear factor E2-related factor-2 (NRF-2) and heme oxygenase-1 (HO-1) were analyzed by reverse transcription quantitative polymerase chain reaction and Western blotting, while intracellular reactive oxygen species (ROS) levels were evaluated using H2DCFDA staining and flow cytometry. The effects of DSF/Cu on protein kinase B (Akt) and mitogen-activated protein kinase (MAPK) signaling pathways were evaluated by Western blot analysis. In vivo efficacy was investigated using a xenograft mouse model, in which mice were orally administered DSF (75 mg/kg) and Cu (2 mg/kg) twice weekly for 5 weeks. We demonstrated that DSF/Cu effectively induced cytotoxicity in both pancreatic cancer cells and their 5FUR counterparts by modulating ROS levels, NRF-2 levels, and associated survival pathways. DSF/Cu treatment significantly decreased NRF-2 expression and reduced ROS levels, specifically in 5FUR cells. DSF/Cu facilitated NRF-2-independent HO-1 expression and differentially modulated Akt and MAPK signaling pathways in pancreatic cancer cells and their 5FUR counterparts. In vivo studies using a xenograft mouse model confirmed the antitumor efficacy of DSF/Cu, as evidenced by reduced tumor volumes and NRF-2 expression. These findings highlight the potential of DSF/Cu as a novel and effective therapeutic strategy for PDAC, specifically for overcoming resistance to standard therapies.


  • Open Access

    A New Staining Method Using Methionyl-tRNA Synthetase 1 Antibody for Endoscopic Ultrasound-Guided Fine-Needle Aspiration Cytology of Pancreatic Cancer

    by Sung Ill Jang, See Young Lee, Ji Hae Nahm, Jae Hee Cho, Jung Hyun Jo, Chan Min Jung, Beom Jin Lim, Jin Hong Lim, Hyung Sun Kim, Su Yun Lee, et al.

    Background/Objectives: Accurate determination of malignancy in pancreatic masses through endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) is crucial for appropriate clinical management and prognostic assessment. However, the diagnostic sensitivity of conventional cytology using Papanicolaou (Pap) staining remains limited, often leading to inconclusive results. In this study, we investigated the diagnostic utility of methionyl-tRNA synthetase 1 (MARS1) through immunohistochemical (IHC) and immunofluorescence (IF) staining as a potential biomarker for pancreatic cancer. IHC analysis was conducted on resected tissue samples from 10 patients, including both pancreatic ductal adenocarcinoma and corresponding non-neoplastic pancreatic tissue. Additionally, cytologic samples were obtained from 198 patients with pancreatic masses who underwent EUS-FNA for diagnostic evaluation. Pap staining and MARS1 IF staining were performed on liquid-based cytology slides derived from EUS-FNA specimens. Results: MARS1 was detected by IHC staining in the 10 surgical specimens diagnosed with pancreatic adenocarcinomas. After Pap staining, 37 patients were excluded because of unsuitable specimens, leaving 161 patients who underwent both Pap and MARS1 IF staining. EUS-FNA specimens from the 151 patients with pancreatic ductal adenocarcinoma were classified by Pap staining as atypia (n = 36), suspicious for malignancy (n = 55), or malignancy (n = 60). MARS1 IF staining was positive in 147 of these patients and negative in 4. MARS1 IF staining distinguished pancreatic cancer in specimens with atypia on Pap staining. The sensitivity for detecting pancreatic cancer was significantly higher for MARS1 IF staining than for conventional Pap staining (97.4% vs. 79.1%, p < 0.0001). Conclusions: The high sensitivity of MARS1 IF staining improved malignancy detection in pancreatic masses. Further prospective studies are required to validate our findings.

  • Open Access

    The Catheter Flushing Method Shows a Similar Diagnostic Yield to the Conventional Method in Brushing Cytology for Biliary Strictures

    by Sung Ill Jang, Ji Hae Nahm, See Young Lee, Seok Jeong, Tae Hoon Lee, Dong Uk Kim, Chang-Il Kwon, Jae Hee Cho, Min Je Sung

    Background/Objectives: Endobiliary brushing is usually performed in the diagnosis of indeterminate biliary strictures; however, in this setting, brush cytology is limited by a low diagnostic yield and sensitivity. Here, we compared the catheter flushing method (CFM) with the conventional cytologic method (CCM) in terms of cellularity and diagnostic performance. Methods: Endobiliary brushings were obtained during endoscopic retrograde cholangiopancreatography (ERCP) from patients with biliary strictures enrolled at six tertiary hospitals. Additionally, the CFM was performed after brushing. Using liquid-based cytologic preparations of samples, we assessed the diagnostic performance of the CCM using Pap staining and the CFM using methionyl-transfer RNA synthetase 1 (MARS1) immunofluorescence staining. Results: From a total of 399 patients (malignant, 253; benign, 146), 374 CCM samples and 361 CFM samples contained adequate cells, with no significant difference in diagnostic yield (93.7% vs. 90.5%, respectively; p = 0.088). The sensitivity of the CFM (90.3%) was significantly higher than that of the CCM (75.1%; p < 0.001), with no significant difference in accuracy between methods (81.2% vs. 82.6%, respectively; p = 0.608). Conclusions: The diagnostic yield of the CFM was comparable to that of the CCM. Additionally, the diagnostic performance of the CFM was comparable to that of the CCM. These findings indicate that the CFM could be an additional brush cytology method for sample collection in patients with indeterminate biliary strictures. Incorporating both the CCM and CFM might be expected to improve the diagnostic yield of brush cytology in the biliary strictures. Further prospective comparative studies between the CCM and CFM using the same staining method are needed to validate these findings.

  • Oral high-carbohydrate solution as an alternative dietary modality in patients with acute pancreatitis

    by See Young Lee, Jaein Lee, Jae Hee Cho, Dong Ki Lee, Yeseul Seong, Sung Ill Jang


  • Open Access

    Gallstone Dissolution Effects of Combination Therapy with n-3 Polyunsaturated Fatty Acids and Ursodeoxycholic Acid: A Randomized, Prospective, Preliminary Clinical Trial

    by See Young Lee, Sung Ill Jang, Jae Hee Cho, Min Young Do, Su Yeon Lee, Arong Choi, Hye Sun Lee, Juyeon Yang, Dong Ki Lee

    See Young Lee, Sung Ill Jang, Jae Hee Cho, Min Young Do, Su Yeon Lee, Arong Choi, Hye Sun Lee, Juyeon Yang, Dong Ki Lee. Gut and Liver -0001;0:. https://doi.org/10.5009/gnl230494