Purpose: This study explored the feasibility of utilizing a deep learning artificial intelligence (AI) model to detect ileocolic intussusception on grayscale ultrasound images. Methods: This retrospective observational study incorporated ultrasound images of children who underwent emergency ultrasonography for suspected ileocolic intussusception. After excluding video clips, Doppler images, and annotated images, 40,765 images from two tertiary hospitals were included (positive-to-negative ratio: hospital A, 2,775:35,373; hospital B, 140:2,477). Images from hospital A were split into a training set, a tuning set, and an internal test set (ITS) at a ratio of 7:1.5:1.5. Images from hospital B comprised an external test set (ETS). For each image indicating intussusception, two radiologists provided a bounding box as the ground-truth label. If intussusception was suspected in the input image, the model generated a bounding box with a confidence score (0-1) at the estimated lesion location. Average precision (AP) was used to evaluate overall model performance. The performance of practical thresholds for the modelgenerated confidence score, as determined from the ITS, was verified using the ETS. Results: The AP values for the ITS and ETS were 0.952 and 0.936, respectively. Two confidence thresholds, CTopt and CTprecision, were set at 0.557 and 0.790, respectively. For the ETS, the perimage precision and recall were 95.7% and 80.0% with CTopt, and 98.4% and 44.3% with CTprecision. For per-patient diagnosis, the sensitivity and specificity were 100.0% and 97.1% with CTopt, and 100.0% and 99.0% with CTprecision. The average number of false positives per patient was 0.04 with CTopt and 0.01 for CTprecision. Conclusion The feasibility of using an AI model to diagnose ileocolic intussusception on ultrasonography was demonstrated. However, further study involving bias-free data is warranted for robust clinical validation.

Purpose: This study explored the feasibility of utilizing a deep learning artificial intelligence (AI) model to detect ileocolic intussusception on grayscale ultrasound images. Methods: This retrospective observational study incorporated ultrasound images of children who underwent emergency ultrasonography for suspected ileocolic intussusception. After excluding video clips, Doppler images, and annotated images, 40,765 images from two tertiary hospitals were included (positive-to-negative ratio: hospital A, 2,775:35,373; hospital B, 140:2,477). Images from hospital A were split into a training set, a tuning set, and an internal test set (ITS) at a ratio of 7:1.5:1.5. Images from hospital B comprised an external test set (ETS). For each image indicating intussusception, two radiologists provided a bounding box as the ground-truth label. If intussusception was suspected in the input image, the model generated a bounding box with a confidence score (0-1) at the estimated lesion location. Average precision (AP) was used to evaluate overall model performance. The performance of practical thresholds for the modelgenerated confidence score, as determined from the ITS, was verified using the ETS. Results: The AP values for the ITS and ETS were 0.952 and 0.936, respectively. Two confidence thresholds, CTopt and CTprecision, were set at 0.557 and 0.790, respectively. For the ETS, the perimage precision and recall were 95.7% and 80.0% with CTopt, and 98.4% and 44.3% with CTprecision. For per-patient diagnosis, the sensitivity and specificity were 100.0% and 97.1% with CTopt, and 100.0% and 99.0% with CTprecision. The average number of false positives per patient was 0.04 with CTopt and 0.01 for CTprecision. Conclusion The feasibility of using an AI model to diagnose ileocolic intussusception on ultrasonography was demonstrated. However, further study involving bias-free data is warranted for robust clinical validation.

Purpose: This study explored the feasibility of utilizing a deep learning artificial intelligence (AI) model to detect ileocolic intussusception on grayscale ultrasound images. Methods: This retrospective observational study incorporated ultrasound images of children who underwent emergency ultrasonography for suspected ileocolic intussusception. After excluding video clips, Doppler images, and annotated images, 40,765 images from two tertiary hospitals were included (positive-to-negative ratio: hospital A, 2,775:35,373; hospital B, 140:2,477). Images from hospital A were split into a training set, a tuning set, and an internal test set (ITS) at a ratio of 7:1.5:1.5. Images from hospital B comprised an external test set (ETS). For each image indicating intussusception, two radiologists provided a bounding box as the ground-truth label. If intussusception was suspected in the input image, the model generated a bounding box with a confidence score (0-1) at the estimated lesion location. Average precision (AP) was used to evaluate overall model performance. The performance of practical thresholds for the modelgenerated confidence score, as determined from the ITS, was verified using the ETS. Results: The AP values for the ITS and ETS were 0.952 and 0.936, respectively. Two confidence thresholds, CTopt and CTprecision, were set at 0.557 and 0.790, respectively. For the ETS, the perimage precision and recall were 95.7% and 80.0% with CTopt, and 98.4% and 44.3% with CTprecision. For per-patient diagnosis, the sensitivity and specificity were 100.0% and 97.1% with CTopt, and 100.0% and 99.0% with CTprecision. The average number of false positives per patient was 0.04 with CTopt and 0.01 for CTprecision. Conclusion The feasibility of using an AI model to diagnose ileocolic intussusception on ultrasonography was demonstrated. However, further study involving bias-free data is warranted for robust clinical validation.

Purpose: This study explored the feasibility of utilizing a deep learning artificial intelligence (AI) model to detect ileocolic intussusception on grayscale ultrasound images. Methods: This retrospective observational study incorporated ultrasound images of children who underwent emergency ultrasonography for suspected ileocolic intussusception. After excluding video clips, Doppler images, and annotated images, 40,765 images from two tertiary hospitals were included (positive-to-negative ratio: hospital A, 2,775:35,373; hospital B, 140:2,477). Images from hospital A were split into a training set, a tuning set, and an internal test set (ITS) at a ratio of 7:1.5:1.5. Images from hospital B comprised an external test set (ETS). For each image indicating intussusception, two radiologists provided a bounding box as the ground-truth label. If intussusception was suspected in the input image, the model generated a bounding box with a confidence score (0-1) at the estimated lesion location. Average precision (AP) was used to evaluate overall model performance. The performance of practical thresholds for the modelgenerated confidence score, as determined from the ITS, was verified using the ETS. Results: The AP values for the ITS and ETS were 0.952 and 0.936, respectively. Two confidence thresholds, CTopt and CTprecision, were set at 0.557 and 0.790, respectively. For the ETS, the perimage precision and recall were 95.7% and 80.0% with CTopt, and 98.4% and 44.3% with CTprecision. For per-patient diagnosis, the sensitivity and specificity were 100.0% and 97.1% with CTopt, and 100.0% and 99.0% with CTprecision. The average number of false positives per patient was 0.04 with CTopt and 0.01 for CTprecision. Conclusion The feasibility of using an AI model to diagnose ileocolic intussusception on ultrasonography was demonstrated. However, further study involving bias-free data is warranted for robust clinical validation.

Purpose: This study explored the feasibility of utilizing a deep learning artificial intelligence (AI) model to detect ileocolic intussusception on grayscale ultrasound images. Methods: This retrospective observational study incorporated ultrasound images of children who underwent emergency ultrasonography for suspected ileocolic intussusception. After excluding video clips, Doppler images, and annotated images, 40,765 images from two tertiary hospitals were included (positive-to-negative ratio: hospital A, 2,775:35,373; hospital B, 140:2,477). Images from hospital A were split into a training set, a tuning set, and an internal test set (ITS) at a ratio of 7:1.5:1.5. Images from hospital B comprised an external test set (ETS). For each image indicating intussusception, two radiologists provided a bounding box as the ground-truth label. If intussusception was suspected in the input image, the model generated a bounding box with a confidence score (0-1) at the estimated lesion location. Average precision (AP) was used to evaluate overall model performance. The performance of practical thresholds for the modelgenerated confidence score, as determined from the ITS, was verified using the ETS. Results: The AP values for the ITS and ETS were 0.952 and 0.936, respectively. Two confidence thresholds, CTopt and CTprecision, were set at 0.557 and 0.790, respectively. For the ETS, the perimage precision and recall were 95.7% and 80.0% with CTopt, and 98.4% and 44.3% with CTprecision. For per-patient diagnosis, the sensitivity and specificity were 100.0% and 97.1% with CTopt, and 100.0% and 99.0% with CTprecision. The average number of false positives per patient was 0.04 with CTopt and 0.01 for CTprecision. Conclusion The feasibility of using an AI model to diagnose ileocolic intussusception on ultrasonography was demonstrated. However, further study involving bias-free data is warranted for robust clinical validation.

Background: Adrenocortical carcinomas (ACCs) are rare tumors with aggressive but varied prognosis. Stage, Grade, Resection status, Age, Symptoms (S-GRAS) score, based on clinical and pathological factors, was found to best stratify the prognosis of European ACC patients. This study assessed the prognostic performance of modified S-GRAS (mS-GRAS) scores including modified grade (mG) by integrating mitotic counts into the Ki67 index (original grade), in Korean ACC patients. Methods: Patients who underwent surgery for ACC between January 1996 and December 2022 at three medical centers in Korea were retrospectively analyzed. mS-GRAS scores were calculated based on tumor stage, mG (Ki67 index or mitotic counts), resection status, age, and symptoms. Patients were divided into four groups (0–1, 2–3, 4–5, and 6–9 points) based on total mS-GRAS score. The associations of each variable and mS-GRAS score with recurrence and survival were evaluated using Cox regression analysis, Harrell’s concordance index (C-index), and the Kaplan–Meier method. Results: Data on mS-GRAS components were available for 114 of the 153 patients who underwent surgery for ACC. These 114 patients had recurrence and death rates of 61.4% and 48.2%, respectively. mS-GRAS score was a significantly better predictor of recurrence (C-index=0.829) and death (C-index=0.747) than each component (P<0.05), except for resection status. mS-GRAS scores correlated with shorter progression-free survival (P=8.34E-24) and overall survival (P=2.72E-13). Conclusion mS-GRAS scores showed better prognostic performance than tumor stage and grade in Asian patients who underwent surgery for ACC.

Background/Aims: H2 receptor antagonists (H2RA) have been used to treat gastritis by inhibiting gastric acid. Proton pump inhibitors (PPIs) are more potent acid suppressants than H2RA.However, the efficacy and safety of low-dose PPI for treating gastritis remain unclear. The aim was to investigate the efficacy and safety of low-dose PPI for treating gastritis. Methods: A double-blind, noninferiority, multicenter, phase 3 clinical trial randomly assigned 476 patients with endoscopic erosive gastritis to a group using esomeprazole 10 mg (DW1903) daily and a group using famotidine 20 mg (DW1903R1) daily for 2 weeks. The full-analysis set included 319 patients (DW1903, n=159; DW1903R1, n=160) and the per-protocol set included 298 patients (DW1903, n=147; DW1903R1, n=151). The primary endpoint (erosion improvement rate) and secondary endpoint (erosion and edema cure rates, improvement rates of hemorrhage, erythema, and symptoms) were assessed after the treatment. Adverse events were compared. Results: According to the full-analysis set, the erosion improvement rates in the DW1903 and DW1903R1 groups were 59.8% and 58.8%, respectively. According to the per-protocol analysis, the erosion improvement rates in the DW1903 and DW1903R1 groups were 61.9% and 59.6%, respectively. Secondary endpoints were not significantly different between two groups except that the hemorrhagic improvement rate was higher in DW1903 with statistical tendency. The number of adverse events were not statistically different. Conclusions DW1903 of a low-dose PPI was not inferior to DW1903R1 of H2RA. Thus, lowdose PPI can be a novel option for treating gastritis (ClinicalTrials.gov Identifier: NCT05163756).

Objective: The Scales for Outcomes in Parkinson’s Disease–Cognition (SCOPA-Cog) was developed to assess cognition in patients with Parkinson’s disease (PD). In this study, we aimed to evaluate the validity and reliability of the Korean version of the SCOPACog (K-SCOPA-Cog). Methods: We enrolled 129 PD patients with movement disorders from 31 clinics in South Korea. The original version of the SCOPA-Cog was translated into Korean using the translation-retranslation method. The test–retest method with an intraclass correlation coefficient (ICC) and Cronbach’s alpha coefficient were used to assess reliability. Spearman’s rank correlation analysis with the Montreal Cognitive Assessment-Korean version (MOCA-K) and the Korean Mini-Mental State Examination (K-MMSE) were used to assess concurrent validity. Results: The Cronbach’s alpha coefficient was 0.797, and the ICC was 0.887. Spearman’s rank correlation analysis revealed a significant correlation with the K-MMSE and MOCA-K scores (r = 0.546 and r = 0.683, respectively). Conclusion Our results demonstrate that the K-SCOPA-Cog has good reliability and validity.

Background/Aims: To compare the effects of abatacept and conventional synthetic disease modifying anti-rheumatic drugs (csDMARDs) on the progression and development of rheumatoid arthritis-associated interstitial lung disease (RA-ILD). Methods: This multi-center retrospective study included RA patients receiving abatacept or csDMARDs who underwent at least two pulmonary function tests and/or chest high-resolution computed tomography (HRCT). We compared the following outcomes between the groups: progression of RA-ILD, development of new ILD in RA patients without ILD at baseline, 28-joint Disease Activity Score with the erythrocyte sedimentation rate (DAS28-ESR), and safety. Longitudinal changes were compared between the groups by using a generalized estimating equation. Results: The study included 123 patients who were treated with abatacept (n = 59) or csDMARDs (n = 64). Nineteen (32.2%) and 38 (59.4%) patients treated with abatacept and csDMARDs, respectively, presented with RA-ILD at baseline. Newly developed ILD occurred in one patient receiving triple csDMARDs for 32 months. Among patients with RA-ILD at baseline, ILD progressed in 21.1% of cases treated with abatacept and 34.2% of cases treated with csDMARDs during a median 21-month follow-up. Longitudinal changes in forced vital capacity and diffusing capacity for carbon monoxide were comparable between the two groups. However, the abatacept group showed a more significant decrease in DAS28-ESR and glucocorticoid doses than csDMARDs group during the follow-up. The safety of both regimens was comparable. Conclusions Abatacept and csDMARDs showed comparable effects on the development and stabilization of RA-ILD. Nevertheless, compared to csDMARDs, abatacept demonstrated a significant improvement in disease activity and led to reduced glucocorticoid use.

Purpose: To develop and evaluate a deep learning technique to automatically segment bone structures in zero echo time (ZTE) for skull magnetic resonance imaging (MRI) in children. Materials and Methods: From January to December 2021, 38 bone ZTE MRIs from infants and children (age range, 1–31 months) were collected for model development.Mask images were generated by manually segmenting the craniofacial bone using a commercial segmentation program. Among them, 35 ZTE series were used to train the three-dimensional (3D)-nnUnet deep learning model and the remaining three series were used for model validation. A temporally different dataset of 19 ZTE bone MRIs obtained in May 2022 from infants and children (age range, 3–168 months) was used to determine the model’s performance. Dice similarity coefficient was calculated for each test case.From 3D volume rendering images, segmentation accuracy, overall image quality, and visibility of cranial sutures were subjectively evaluated on a 5-point scale and compared with ground truth data from manual segmentation. Reasons for segmentation failure were analyzed using axially segmented ZTE images. Results: For the test set, the mean Dice similarity coefficient was 0.985 ± 0.019. The segmentation accuracy was lower than the ground truth without showing a statistically significant difference between the two (3.39 ± 1.11 vs. 3.73 ± 0.77, p = 0.055). The overall image quality and suture visibility showed no significant difference (3.34 ± 0.75 vs.3.42 ± 0.69, p = 0.317; 3.55 ± 0.97 vs. 3.60 ± 0.95, p = 0.157). Common reasons for low segmentation accuracy were well-pneumatized sinuses, metal artifacts, skin at the vertex level, and bones too thin. Conclusion The deep learning-based automatic segmentation technique of bone ZTE MRIs showed comparable segmentation performance to manual segmentation. Using the deep learning-based segmentation results, acceptable 3D-volume rendering images of craniofacial bones were generated.