Quality control (QC) in medical imaging is important for improving diagnostic accuracy, optimizing treatment planning, and ensuring patient safety. With the increasing complexity of imaging technologies, consistent and structured QC practices are essential to ensure high-quality healthcare delivery. Korea’s QC initiatives began with regulatory standards for special medical equipment, aiming to institutionalize requirements for basic equipment and personnel.Current Concepts: Currently, Korea’s system primarily focuses on equipment-level control through regular inspections and legal standards. However, standardized protocols for image acquisition, interpretation, and reporting remain inconsistent across institutions. The American College of Radiology provides a benchmark model for a more integrated approach. Technological advances, including artificial intelligence, are increasingly influencing imaging processes. However, these advancements pose new challenges regarding their evaluation and integration into existing QC systems.Discussion and Conclusion: QC in medical imaging should be expanded beyond equipment maintenance to encompass procedural and interpretive standards. Multi-sector collaboration is necessary to refine policies and ensure that emerging technologies improve patient outcomes and healthcare efficiency.

Background and Objectives: Outcomes of deferring percutaneous coronary intervention (PCI) without invasive physiologic assessment for intermediate coronary lesions is uncertain.We sought to compare long-term outcomes between medical treatment and PCI of intermediate lesions without invasive physiologic assessment. Methods: A total of 899 patients with intermediate coronary lesions between 50% and 70% diameter-stenosis were randomized to the conservative group (n=449) or the aggressive group (n=450). For intermediate lesions, PCI was performed in the aggressive group, but was deferred in the conservative group. The primary endpoint was major adverse cardiac events (MACE, a composite of all-cause death, myocardial infarction [MI], or ischemia-driven any revascularization) at 3 years. Results: The number of treated lesions per patient was 0.8±0.9 in the conservative group and 1.7±0.9 in the aggressive group (p=0.001). At 3 years, the conservative group had a significantly higher incidence of MACE than the aggressive group (13.8% vs. 9.3%; hazard ratio [HR], 1.49; 95% confidence interval [CI], 1.00–2.21; p=0.049), mainly driven by revascularization of target intermediate lesion (6.5% vs. 1.1%; HR, 5.69; 95% CI, 2.20–14.73;p<0.001). Between 1 and 3 years after the index procedure, compared to the aggressive group, the conservative group had significantly higher incidence of cardiac death or MI (3.2% vs.0.7%; HR, 4.34; 95% CI, 1.24–15.22; p=0.022) and ischemia-driven any revascularization. Conclusions For intermediate lesions, medical therapy alone, guided only by angiography, was associated with a higher risk of MACE at 3 years compared with performing PCI, mainly due to increased revascularization.

Purpose: This study aimed to evaluate the diagnostic accuracy of three multimodal large language models (LLMs) in radiological image interpretation and to assess the impact of prompt engineering strategies and input conditions. Methods: This study analyzed 67 radiological quiz cases from the Korean Society of Ultrasound in Medicine. Three multimodal LLMs (Claude 3.5 Sonnet, GPT-4o, and Gemini-1.5-Pro-002) were evaluated using six types of prompts (basic [without system prompt], original [specific instructions], chain-of-thought, reflection, multiagent, and artificial intelligence [AI]–generated). Performance was assessed across various factors, including tumor versus non-tumor status, case rarity, difficulty, and knowledge cutoff dates. A subgroup analysis compared diagnostic accuracy between imaging-only inputs and combined imaging-descriptive text inputs. Results: With imaging-only inputs, Claude 3.5 Sonnet achieved the highest overall accuracy (46.3%, 186/402), followed by GPT-4o (43.5%, 175/402) and Gemini-1.5-Pro-002 (39.8%, 160/402). AI-generated prompts yielded superior combined accuracy across all three models, with significant improvements over the basic (7.96%, P=0.009), chain-of-thought (6.47%, P=0.029), and multiagent prompts (5.97%, P=0.043). The integration of descriptive text significantly enhanced diagnostic accuracy for Claude 3.5 Sonnet (46.3% to 66.2%, P<0.001), GPT-4o (43.5% to 57.5%, P<0.001), and Gemini-1.5-Pro-002 (39.8% to 60.4%, P<0.001). Model performance was significantly influenced by case rarity for GPT-4o (rare: 6.7% vs. nonrare: 53.9%, P=0.001) and by knowledge cutoff dates for Claude 3.5 Sonnet (post-cutoff: 23.5% vs. pre-cutoff: 64.0%, P=0.005). Conclusion Claude 3.5 Sonnet achieved the highest diagnostic accuracy in radiological quiz cases, followed by GPT-4o and Gemini-1.5-Pro-002. The use of AI-generated prompts and the integration of descriptive text inputs enhanced model performance.

This study explores the development, roles, and key initiatives of the Regional Environmental Health Centers in Korea, detailing their evolution through four distinct phases and their impact on environmental health policy and local governance. It chronicles the establishment and transformation of these centers from their inception in May 2007, through four developmental stages. Originally named Environmental Disease Research Centers, they were subsequently renamed Environmental Health Centers following legislative changes. The analysis includes the expansion in the number of centers, the transfer of responsibilities to local governments, and the launch of significant projects such as the Korean Children’s Environmental Health Study (Ko-CHENS ). During the initial phase (May 2007–February 2009), the 10 centers concentrated on research-driven activities, shifting from a media-centered to a receptor-centered approach. In the second phase, prompted by the enactment of the Environmental Health Act, six additional centers were established, broadening their scope to address national environmental health issues. The third phase introduced Ko-CHENS, a 20-year national cohort project designed to influence environmental health policy by integrating research findings into policy frameworks. The fourth phase marked a decentralization of authority, empowering local governments and redefining the centers' roles to focus on regional environmental health challenges. The Regional Environmental Health Centers have significantly evolved and now play a crucial role in addressing local environmental health issues and supporting local government policies. Their capacity to adapt and respond to region-specific challenges is essential for the effective implementation of environmental health policies, reflecting geographical, socioeconomic, and demographic differences.

Purpose: This study aimed to evaluate the diagnostic accuracy of three multimodal large language models (LLMs) in radiological image interpretation and to assess the impact of prompt engineering strategies and input conditions. Methods: This study analyzed 67 radiological quiz cases from the Korean Society of Ultrasound in Medicine. Three multimodal LLMs (Claude 3.5 Sonnet, GPT-4o, and Gemini-1.5-Pro-002) were evaluated using six types of prompts (basic [without system prompt], original [specific instructions], chain-of-thought, reflection, multiagent, and artificial intelligence [AI]–generated). Performance was assessed across various factors, including tumor versus non-tumor status, case rarity, difficulty, and knowledge cutoff dates. A subgroup analysis compared diagnostic accuracy between imaging-only inputs and combined imaging-descriptive text inputs. Results: With imaging-only inputs, Claude 3.5 Sonnet achieved the highest overall accuracy (46.3%, 186/402), followed by GPT-4o (43.5%, 175/402) and Gemini-1.5-Pro-002 (39.8%, 160/402). AI-generated prompts yielded superior combined accuracy across all three models, with significant improvements over the basic (7.96%, P=0.009), chain-of-thought (6.47%, P=0.029), and multiagent prompts (5.97%, P=0.043). The integration of descriptive text significantly enhanced diagnostic accuracy for Claude 3.5 Sonnet (46.3% to 66.2%, P<0.001), GPT-4o (43.5% to 57.5%, P<0.001), and Gemini-1.5-Pro-002 (39.8% to 60.4%, P<0.001). Model performance was significantly influenced by case rarity for GPT-4o (rare: 6.7% vs. nonrare: 53.9%, P=0.001) and by knowledge cutoff dates for Claude 3.5 Sonnet (post-cutoff: 23.5% vs. pre-cutoff: 64.0%, P=0.005). Conclusion Claude 3.5 Sonnet achieved the highest diagnostic accuracy in radiological quiz cases, followed by GPT-4o and Gemini-1.5-Pro-002. The use of AI-generated prompts and the integration of descriptive text inputs enhanced model performance.

Background and Objectives: Outcomes of deferring percutaneous coronary intervention (PCI) without invasive physiologic assessment for intermediate coronary lesions is uncertain.We sought to compare long-term outcomes between medical treatment and PCI of intermediate lesions without invasive physiologic assessment. Methods: A total of 899 patients with intermediate coronary lesions between 50% and 70% diameter-stenosis were randomized to the conservative group (n=449) or the aggressive group (n=450). For intermediate lesions, PCI was performed in the aggressive group, but was deferred in the conservative group. The primary endpoint was major adverse cardiac events (MACE, a composite of all-cause death, myocardial infarction [MI], or ischemia-driven any revascularization) at 3 years. Results: The number of treated lesions per patient was 0.8±0.9 in the conservative group and 1.7±0.9 in the aggressive group (p=0.001). At 3 years, the conservative group had a significantly higher incidence of MACE than the aggressive group (13.8% vs. 9.3%; hazard ratio [HR], 1.49; 95% confidence interval [CI], 1.00–2.21; p=0.049), mainly driven by revascularization of target intermediate lesion (6.5% vs. 1.1%; HR, 5.69; 95% CI, 2.20–14.73;p<0.001). Between 1 and 3 years after the index procedure, compared to the aggressive group, the conservative group had significantly higher incidence of cardiac death or MI (3.2% vs.0.7%; HR, 4.34; 95% CI, 1.24–15.22; p=0.022) and ischemia-driven any revascularization. Conclusions For intermediate lesions, medical therapy alone, guided only by angiography, was associated with a higher risk of MACE at 3 years compared with performing PCI, mainly due to increased revascularization.

Background: Dapagliflozin, a sodium-glucose cotransporter 2 inhibitor, reduces hyperglycemia and obesity by inhibiting renal glucose reabsorption. This post hoc study evaluated clinical factors influencing patient response to dapagliflozin. Methods: The analysis focused on patients treated with dapagliflozin (10 mg/day for 52 weeks) within the randomized, double-blind, parallel-group BEYOND trial. Adequate glycemic control (GC) was defined as a reduction in glycated hemoglobin (HbA1c) of ≥ 1.0% or the achievement of an HbA1c level <7.0% at week 52. Significant weight loss (WL) referred to a reduction in body weight of ≥3.0% at week 52. Participants were classified into four groups based on their GC and WL responses: GC+/WL+, GC+/WL−, GC−/WL+, and GC−/WL−. Results: Among dapagliflozin recipients (n=56), at 52 weeks, HbA1c had decreased by 1.0%±0.8% from baseline, while body weight had declined by 2.4±3.1 kg. Overall, 69.6% of participants achieved GC+, and 57.1% achieved WL+. Male sex and shorter diabetes duration were significantly associated with achieving GC+. Conversely, higher estimated glomerular filtration rate was significantly linked to WL+. The only factor significantly associated with both GC+ and WL+ was shorter diabetes duration (odds ratio, 0.81; 95% confidence interval, 0.68 to 0.97; P=0.023). The GC+ and WL+ groups exhibited favorable responses beginning soon after dapagliflozin therapy was initiated. Furthermore, HbA1c decline was more strongly associated with reduction in visceral fat than with WL. Conclusion A short duration of diabetes and early response to treatment appear to represent key factors in maximizing the benefits of dapagliflozin for blood glucose and weight management.

Purpose: Treatment options for hepatocellular carcinoma (HCC) vary according to known guidelines among liver resection (LR), liver transplantation (LT), radiofrequency ablation (RFA), and transarterial chemoembolization (TACE). This study aimed to compare the outcomes of initial treatment for patients with resectable HCC within Milan criteria (MC) via nationwide data. Methods: Patients with resectable HCC (Child-Pugh class A; platelet count, ≥100,000/μL) within MC from the Korean Liver Cancer Association databank were analyzed, retrospectively. Outcomes according to initial treatment and subgroups according to tumor size and number were analyzed. Overall survival (OS) rates after initial treatment were compared. Results: A total of 3,241 patients who underwent LR (n = 1,371), LT (n = 12), RFA (n = 679), or TACE (n = 1,179) were included. The 5-year OS rates differed significantly between the groups (P < 0.05), except for LT (LR, 84.9%; LT, 82.5%;RFA, 76.2%; and TACE, 59.9%). For patients with a single tumor of any size, the 5-year OS rates of the LR group were significantly higher than RFA and TACE groups. For patients with multiple tumors, the 5-year OS rates were 78.2%, 100%, 74.3%, and 53.0% for the LR, LT, RFA, and TACE groups, respectively, but without significant difference between LR and RFA (P = 0.86). Conclusion For resectable HCC within MC, the LR had the highest OS rate for a single tumor of any size. LR and RFA showed no significant differences in OS rate for multiple tumors. LR has a much more optimistic outlook for HCC within MC.

This study explores the development, roles, and key initiatives of the Regional Environmental Health Centers in Korea, detailing their evolution through four distinct phases and their impact on environmental health policy and local governance. It chronicles the establishment and transformation of these centers from their inception in May 2007, through four developmental stages. Originally named Environmental Disease Research Centers, they were subsequently renamed Environmental Health Centers following legislative changes. The analysis includes the expansion in the number of centers, the transfer of responsibilities to local governments, and the launch of significant projects such as the Korean Children’s Environmental Health Study (Ko-CHENS ). During the initial phase (May 2007–February 2009), the 10 centers concentrated on research-driven activities, shifting from a media-centered to a receptor-centered approach. In the second phase, prompted by the enactment of the Environmental Health Act, six additional centers were established, broadening their scope to address national environmental health issues. The third phase introduced Ko-CHENS, a 20-year national cohort project designed to influence environmental health policy by integrating research findings into policy frameworks. The fourth phase marked a decentralization of authority, empowering local governments and redefining the centers' roles to focus on regional environmental health challenges. The Regional Environmental Health Centers have significantly evolved and now play a crucial role in addressing local environmental health issues and supporting local government policies. Their capacity to adapt and respond to region-specific challenges is essential for the effective implementation of environmental health policies, reflecting geographical, socioeconomic, and demographic differences.