The philtrum of the upper lip is important to the facial contour and general appearance of individuals. In patients who have undergone cleft lip surgery, reconstruction of the philtrum is important in restoring the normal appearance of the upper lip and it also helps in diverting people`s attention away from the surgical scar. Several methods of philtral dimple creation have been used, but the postoperative results have not always been satisfactory. Between 1991 and 1997, reconstruction of the philtrum with unilateral cleft nasal deformity was performed in 43 patients, transposing the orbicularis oris muscle of the central upper lip to the depressed alar base. Using this method, efforts were made to achieve reconstruction of the philtrum and correction of alar base depression simultaneously. The results were rated good to excellent by a panel of judges using the ordinary scale method. Although reconstruction of the philtral dimple and correction of the depressed alar base were very successful, reconstruction of philtral column(especially the upper portion) was not satisfactory. Therefore, other procedures such as temporal fascial grafts in the upper portion of the philtral column should also be considered at the time of primary surgery or revisional surgery.
Cicatrix ; Cleft Lip* ; Congenital Abnormalities ; Depression* ; Humans ; Lip ; Transplants

PURPOSE: The purpose of this study was to report on splenic artery aneurysms (SAAs) treated by transcatheter embolization in our single-center institution and to evaluate the clinical outcomes of patients with SAA by aneurysm location. MATERIALS AND METHODS: The original medical records and imaging results of 52 patients with SAA treated in our center between January 1, 1995 and December 31, 2013 were reviewed. Of these cases, 7 patients (13.5%) underwent surgery, 4 patients (7.5%) underwent serial observation, and 1 patient had stent insertion only, leaving 40 patients (78.9%) who underwent endovascular treatment using a coil, with or without N-butyl-2-cyanoacrylate. RESULTS: Aneurysms were located in the distal third of the splenic artery in 27 patients (67.5%), in the middle third in 9 cases (22.5%), and in the proximal third in 4 cases (10%). Of the 40 included patients, 25 were female (62.5%). Twenty-eight patients (70%) were asymptomatic. The mean aneurysm diameter was 2.48 cm (range, 0.8-6.0 cm). Complications involved pancreatitis (n=1) and early spleen infarction (n=29: <1/3 in 14, 1/3-2/3 in 10, and >2/3 in 5). Postembolization syndrome was noted in 26 patients (65%). There were no significant differences by aneurysm location in the postoperative increase in the values of white blood cells, amylase, lipase, and C-reactive protein (P=0.067, P=0.881, P=0.891, and P=0.188, respectively). CONCLUSION: At our institution, endovascular management is safe, has high technical success, and represents the first-line treatment for SAA, regardless of aneurysm location.
Amylases ; Aneurysm* ; C-Reactive Protein ; Embolization, Therapeutic ; Enbucrilate ; Female ; Humans ; Infarction ; Leukocytes ; Lipase ; Medical Records ; Pancreatitis ; Spleen ; Splenic Artery* ; Stents

Numerous choices exist for closing any wound, so the surgical challenge is that of selecting the optimal method. It is necessary to balance multiple factors, including recipient site requirements, donor site morbidity, operative complexity, and patient factors. Limiting the donor site morbidity is emphasized in the aphorism "Never rob Peter to pay Paul unless Peter can afford it. Certainly, documented cases exist in which donor site morbidity exceeds the original recipient problem, necessitating a second procedure to reconstruct the donor site. The flap survived and the wound was closed, but the donor site was often worse than the original defect. Numerous donor site complications are often overlooked while one concentrates on the successful flap transfer. The standard method for grafting donor wound after harvesting of a flap uses thick split-thickness skin grafts. This method, however, creates an additional comlication-prone wound at the donor sites. Donor sites for grafting can be painful and may develop infection, hypertrophic scarring, blistering. The problem of donor sites scar hypertrophy occurs most frequently when a graft is taken at more than 0.012 inch thick, leaving a residual dermal bed is too thin. AlloDerm processed allograft dermis was developed as a permanent dermal transplant for full thickness wounds. Between 1997 and 1999, we have applied AlloDerm grafts and ultra-thin autografts on 11 patients with donor sites after harvesting flaps. All the composite AlloDerm /autograft were noted to be firmly adherent except 2 cases, which showed focal loss of the grafts and was healed after second graft. AlloDerm exhibited a high percentage take and supported an overlying ultra thin split-thickness skin autograft, applied simultaneously. By providing a dermal replacement, the grafted dermal matrix permitted a thin autograft from the donor site. The ultra-thin autografts leave donor sites that heal faster and with fewer complication. AlloDerm dermal transplants exhibit excellent elastisity and good pigmentation with minimal scarring or wound contracture. Sensory reinnervation after the composite AlloDerm/autograft was not fully recovered. The reason was that these grafts were placed on the bone or tendon exposed sites which were not sufficiently well- innervated graft bed. The high reproducibility of excellent results with this composite graft, coupled with the reduced trauma and rapid healing of donor sites associated with ultra-thin autograft STSG, has made composite grafting with the use of AlloDerm dermal transplants our new method of choice for treatment of donor defects of flap surgery.
Allografts* ; Autografts ; Blister ; Cicatrix ; Cicatrix, Hypertrophic ; Contracture ; Dermis ; Humans ; Hypertrophy ; Pigmentation ; Skin ; Tendons ; Tissue Donors* ; Transplants ; Wounds and Injuries

OBJECTIVES: In Korea, there have been 10,480 confirmed cases of coronavirus disease 2019 (COVID-19) as of April 11, 2020. We investigated the transmission of COVID-19 in a cluster of cases. METHODS: We analyzed the epidemiological characteristics of 10 confirmed COVID-19 patients in an outbreak that started at Spa facility A in a local community in Korea on March 28, 2020 and traced them through April 8, 2020. Epidemiological surveys and diagnostic tests were conducted for each contact, and the secondary attack rate was estimated. RESULTS: There were 3 male confirmed patients (30.0%) and 7 female confirmed patients (70.0%), and their mean age was 53.5 years (range, 2.0 to 73.0). Two patients (20.0%) were asymptomatic. The incubation period was between 3 days and 12 days. Three confirmed patients were infected at female’s Spa facility A and 7 confirmed patients were second, third, and fourth generations of transmission. Seven confirmed patients contracted COVID-19 through presymptomatic contact. In total, 192 contacts were identified, with a secondary attack rate of 3.6%. Eighty-three contacts (43.2%) were aged 40-59 years, and the secondary attack rate was the highest (12.1%) in those aged ≥60 years. Most exposures (n=156, 81.3%) involved casual contact. The number of visitors using the female’s spa facility was 58, including 3 confirmed patients, resulting in a secondary outbreak rate of 5.9%. CONCLUSIONS This study presents a cluster of cases occurring in a setting with high temperature and humidity. The second, third, and fourth generations were transmitted through presymptomatic contact.

OBJECTIVES: In Korea, there have been 10,480 confirmed cases of coronavirus disease 2019 (COVID-19) as of April 11, 2020. We investigated the transmission of COVID-19 in a cluster of cases. METHODS: We analyzed the epidemiological characteristics of 10 confirmed COVID-19 patients in an outbreak that started at Spa facility A in a local community in Korea on March 28, 2020 and traced them through April 8, 2020. Epidemiological surveys and diagnostic tests were conducted for each contact, and the secondary attack rate was estimated. RESULTS: There were 3 male confirmed patients (30.0%) and 7 female confirmed patients (70.0%), and their mean age was 53.5 years (range, 2.0 to 73.0). Two patients (20.0%) were asymptomatic. The incubation period was between 3 days and 12 days. Three confirmed patients were infected at female’s Spa facility A and 7 confirmed patients were second, third, and fourth generations of transmission. Seven confirmed patients contracted COVID-19 through presymptomatic contact. In total, 192 contacts were identified, with a secondary attack rate of 3.6%. Eighty-three contacts (43.2%) were aged 40-59 years, and the secondary attack rate was the highest (12.1%) in those aged ≥60 years. Most exposures (n=156, 81.3%) involved casual contact. The number of visitors using the female’s spa facility was 58, including 3 confirmed patients, resulting in a secondary outbreak rate of 5.9%. CONCLUSIONS This study presents a cluster of cases occurring in a setting with high temperature and humidity. The second, third, and fourth generations were transmitted through presymptomatic contact.

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.

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.

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.

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.

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.