Objective: To prospectively evaluate the effect of accelerated deep learning-based reconstruction (Accel-DL) on improving brain magnetic resonance imaging (MRI) quality and reducing scan time compared to that in conventional MRI. Materials and Methods: This study included 150 participants (51 male; mean age 57.3 ± 16.2 years). Each group of 50 participants was scanned using one of three 3T scanners from three different vendors. Conventional and Accel-DL MRI images were obtained from each participant and compared using 2D T1- and T2-weighted and 3D gradient-echo sequences. Accel-DL acquisition was achieved using optimized scan parameters to reduce the scan time, with the acquired images reconstructed using U-Net-based software to transform low-quality, undersampled k-space data into high-quality images. The scan times of Accel-DL and conventional MRI methods were compared. Four neuroradiologists assessed the overall image quality, structural delineation, and artifacts using Likert scale (5- and 3-point scales). Inter-reader agreement was assessed using Fleiss’ kappa coefficient. Signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were calculated, and volumetric quantification of regional structures and white matter hyperintensities (WMHs) was performed. Results: Accel-DL showed a mean scan time reduction of 39.4% (range, 24.2%–51.3%). Accel-DL improved overall image quality (3.78 ± 0.71 vs. 3.36 ± 0.61, P < 0.001), structure delineation (2.47 ± 0.61 vs. 2.35 ± 0.62, P < 0.001), and artifacts (3.73 ± 0.72 vs. 3.71 ± 0.69, P = 0.016). Inter-reader agreement was fair to substantial (κ = 0.34–0.50). SNR and CNR increased in Accel-DL (82.0 ± 23.1 vs. 31.4 ± 10.8, P = 0.02; 12.4 ± 4.1 vs. 4.4 ± 11.2, P = 0.02). Bland-Altman plots revealed no significant differences in the volumetric measurements of 98.2% of the relevant regions, except in the deep gray matter, including the thalamus. Five of the six lesion categories showed no significant differences in WMH segmentation, except for leukocortical lesions (r = 0.64 ± 0.29). Conclusion Accel-DL substantially reduced the scan time and improved the quality of brain MRI in both spin-echo and gradientecho sequences without compromising volumetry, including lesion quantification.

Objective: To test the performance of an artificial intelligence-based computer-aided diagnosis (AI-CAD) designed for fullfield digital mammography (FFDM) when applied to synthetic mammography (SM). Materials and Methods: We analyzed 501 women (mean age, 57 ± 11 years) who underwent preoperative mammography and breast cancer surgery. This cohort consisted of 1002 breasts, comprising 517 with cancer and 485 without. All patients underwent digital breast tomosynthesis (DBT) and FFDM during the preoperative workup. The SM is routinely reconstructed using DBT. Commercial AI-CAD (Lunit Insight MMG, version 1.1.7.2) was retrospectively applied to SM and FFDM to calculate the abnormality scores for each breast. The median abnormality scores were compared for the 517 breasts with cancer using the Wilcoxon signed-rank test. Calibration curves of abnormality scores were evaluated. The discrimination performance was analyzed using the area under the receiver operating characteristic curve (AUC), sensitivity, and specificity using a 10% preset threshold. Sensitivity and specificity were further analyzed according to the mammographic and pathological characteristics.The results of SM and FFDM were compared. Results: AI-CAD demonstrated a significantly lower median abnormality score (71% vs. 96%, P < 0.001) and poorer calibration performance for SM than for FFDM. SM exhibited lower sensitivity (76.2% vs. 82.8%, P < 0.001), higher specificity (95.5% vs.91.8%, P < 0.001), and comparable AUC (0.86 vs. 0.87, P = 0.127) than FFDM. SM showed lower sensitivity than FFDM in asymptomatic breasts, dense breasts, ductal carcinoma in situ, T1, N0, and hormone receptor-positive/human epidermal growth factor receptor 2-negative cancers but showed higher specificity in non-cancerous dense breasts. Conclusion AI-CAD showed lower abnormality scores and reduced calibration performance for SM than for FFDM.Furthermore, the 10% preset threshold resulted in different discrimination performances for the SM. Given these limitations, off-label application of the current AI-CAD to SM should be avoided.

Background/Aims: Adiponectin, a hormone primarily produced by adipocytes, typically shows an inverse relationship with body mass index (BMI). However, some studies have reported a positive correlation between the two. Thus, this study aimed to examine the relationship between adiponectin level and BMI in diabetic patients, focusing on the impact of past obesity on current adiponectin levels. Methods: We conducted an observational study analyzing data from 323 diabetic patients at Kyungpook National University Hospital. Based on past and current BMIs, participants were categorized into never-obese (nn, n = 106), previously obese (on, n = 43), and persistently obese (oo, n = 73) groups based on a BMI threshold of 25 kg/m2. Adiponectin level and BMI were key variables. Kaplan–Meier analysis assessed their impact on all-cause mortality up to August 2023, with survival differences based on adiponectin quartiles and follow-up starting from patient enrollment (2010–2015). Results: The analysis revealed a significant inverse correlation between adiponectin level and past maximum BMI. The on group exhibited approximately 10% lower adiponectin levels compared to the nn group. This association remained significant after adjusting for current BMI, age, and sex, highlighting the lasting influence of previous obesity on adiponectin levels. Furthermore, survival analysis indicated that patients in the lowest adiponectin quartile had reduced survival, with a statistically significant trend (p = 0.062). Conclusions Findings of this study suggest that lower adiponectin levels, potentially reflecting past obesity, are associated with decreased survival in diabetic patients, underscoring a critical role of adiponectin in long-term health outcomes.

Objective: To prospectively evaluate the effect of accelerated deep learning-based reconstruction (Accel-DL) on improving brain magnetic resonance imaging (MRI) quality and reducing scan time compared to that in conventional MRI. Materials and Methods: This study included 150 participants (51 male; mean age 57.3 ± 16.2 years). Each group of 50 participants was scanned using one of three 3T scanners from three different vendors. Conventional and Accel-DL MRI images were obtained from each participant and compared using 2D T1- and T2-weighted and 3D gradient-echo sequences. Accel-DL acquisition was achieved using optimized scan parameters to reduce the scan time, with the acquired images reconstructed using U-Net-based software to transform low-quality, undersampled k-space data into high-quality images. The scan times of Accel-DL and conventional MRI methods were compared. Four neuroradiologists assessed the overall image quality, structural delineation, and artifacts using Likert scale (5- and 3-point scales). Inter-reader agreement was assessed using Fleiss’ kappa coefficient. Signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were calculated, and volumetric quantification of regional structures and white matter hyperintensities (WMHs) was performed. Results: Accel-DL showed a mean scan time reduction of 39.4% (range, 24.2%–51.3%). Accel-DL improved overall image quality (3.78 ± 0.71 vs. 3.36 ± 0.61, P < 0.001), structure delineation (2.47 ± 0.61 vs. 2.35 ± 0.62, P < 0.001), and artifacts (3.73 ± 0.72 vs. 3.71 ± 0.69, P = 0.016). Inter-reader agreement was fair to substantial (κ = 0.34–0.50). SNR and CNR increased in Accel-DL (82.0 ± 23.1 vs. 31.4 ± 10.8, P = 0.02; 12.4 ± 4.1 vs. 4.4 ± 11.2, P = 0.02). Bland-Altman plots revealed no significant differences in the volumetric measurements of 98.2% of the relevant regions, except in the deep gray matter, including the thalamus. Five of the six lesion categories showed no significant differences in WMH segmentation, except for leukocortical lesions (r = 0.64 ± 0.29). Conclusion Accel-DL substantially reduced the scan time and improved the quality of brain MRI in both spin-echo and gradientecho sequences without compromising volumetry, including lesion quantification.

Objective: To test the performance of an artificial intelligence-based computer-aided diagnosis (AI-CAD) designed for fullfield digital mammography (FFDM) when applied to synthetic mammography (SM). Materials and Methods: We analyzed 501 women (mean age, 57 ± 11 years) who underwent preoperative mammography and breast cancer surgery. This cohort consisted of 1002 breasts, comprising 517 with cancer and 485 without. All patients underwent digital breast tomosynthesis (DBT) and FFDM during the preoperative workup. The SM is routinely reconstructed using DBT. Commercial AI-CAD (Lunit Insight MMG, version 1.1.7.2) was retrospectively applied to SM and FFDM to calculate the abnormality scores for each breast. The median abnormality scores were compared for the 517 breasts with cancer using the Wilcoxon signed-rank test. Calibration curves of abnormality scores were evaluated. The discrimination performance was analyzed using the area under the receiver operating characteristic curve (AUC), sensitivity, and specificity using a 10% preset threshold. Sensitivity and specificity were further analyzed according to the mammographic and pathological characteristics.The results of SM and FFDM were compared. Results: AI-CAD demonstrated a significantly lower median abnormality score (71% vs. 96%, P < 0.001) and poorer calibration performance for SM than for FFDM. SM exhibited lower sensitivity (76.2% vs. 82.8%, P < 0.001), higher specificity (95.5% vs.91.8%, P < 0.001), and comparable AUC (0.86 vs. 0.87, P = 0.127) than FFDM. SM showed lower sensitivity than FFDM in asymptomatic breasts, dense breasts, ductal carcinoma in situ, T1, N0, and hormone receptor-positive/human epidermal growth factor receptor 2-negative cancers but showed higher specificity in non-cancerous dense breasts. Conclusion AI-CAD showed lower abnormality scores and reduced calibration performance for SM than for FFDM.Furthermore, the 10% preset threshold resulted in different discrimination performances for the SM. Given these limitations, off-label application of the current AI-CAD to SM should be avoided.

Background/Aims: Adiponectin, a hormone primarily produced by adipocytes, typically shows an inverse relationship with body mass index (BMI). However, some studies have reported a positive correlation between the two. Thus, this study aimed to examine the relationship between adiponectin level and BMI in diabetic patients, focusing on the impact of past obesity on current adiponectin levels. Methods: We conducted an observational study analyzing data from 323 diabetic patients at Kyungpook National University Hospital. Based on past and current BMIs, participants were categorized into never-obese (nn, n = 106), previously obese (on, n = 43), and persistently obese (oo, n = 73) groups based on a BMI threshold of 25 kg/m2. Adiponectin level and BMI were key variables. Kaplan–Meier analysis assessed their impact on all-cause mortality up to August 2023, with survival differences based on adiponectin quartiles and follow-up starting from patient enrollment (2010–2015). Results: The analysis revealed a significant inverse correlation between adiponectin level and past maximum BMI. The on group exhibited approximately 10% lower adiponectin levels compared to the nn group. This association remained significant after adjusting for current BMI, age, and sex, highlighting the lasting influence of previous obesity on adiponectin levels. Furthermore, survival analysis indicated that patients in the lowest adiponectin quartile had reduced survival, with a statistically significant trend (p = 0.062). Conclusions Findings of this study suggest that lower adiponectin levels, potentially reflecting past obesity, are associated with decreased survival in diabetic patients, underscoring a critical role of adiponectin in long-term health outcomes.

Background: Barriers to treatment with intravenous thrombolysis (IVT) for patients with acute ischemic stroke (AIS) in South Korea remain incompletely characterized. We analyze a nationwide prospective cohort to determine patient-level features associated with delayed presentation and non-treatment of potential IVT-eligible patients. Methods: We identified consecutive patients with AIS from 01/2011 to 08/2023 from a multicenter and prospective acute stroke registry in Korea. Patients were defined as IVT candidates if they presented within 4.5 hours from the last known well, had no lab evidence of coagulopathy, and had National Institute of Health Stroke Scale (NIHSS) ≥ 4. Multivariable generalized linear mixed regression models were used to investigate the associations between their characteristics and the IVT candidates or the use of IVT among the candidates. Results: Among 84,103 AIS patients, 41.0% were female, with a mean age of 69 ± 13 years and presentation NIHSS of 4 [interquartile range, 1–8]. Out of these patients, 13,757 (16.4%) were eligible for IVT, of whom 8,179 (59.5%) received IVT. Female sex (adjusted risk ratio [RR], 0.90; 95% confidence interval [CI], 0.86–0.94) and lower years of education (adjusted RR, 0.90; 95% CI, 0.84–0.97 for 0–3 years, compared to ≥ 13 years) were associated with a decreased likelihood of presenting as eligible for IVT after AIS; meanwhile, young age (adjusted RR, 1.12; 95% CI, 1.01–1.24 for ≤ 44 years, compared to 75–84 years) was associated with an increased likelihood of being an IVT candidate. Among those who were eligible for IVT, only age was significantly associated with the use of IVT (adjusted RR, 1.09; 95% CI, 1.03–1.16 for age 65–74 and adjusted RR, 0.83; 95% CI, 0.76–0.90 for ≥ 85 years, respectively). Conclusion Most patients with AIS present outside IVT eligibility in South Korea, and only 60% of eligible patients were ultimately treated. We identified increased age, female sex and lower education as key features on which to focus interventions for improving IVT utilization.

Background: Tinnitus is a multifactorial condition with no universally accepted assessment guidelines. The Korean Tinnitus Study Group previously established consensus statements on the definition, classification, and diagnostic tests for tinnitus. As a continuation of this effort, this study aims to establish expert consensus on tinnitus assessment and treatment outcome evaluation, specifically tailored to the Korean clinical context. Methods: A modified Delphi method involving 26 otology experts from across Korea was used. A two-round Delphi survey was conducted to evaluate statements related to tinnitus assessment before and after treatment. Statements were rated on a scale of 1 to 9 for the level of agreement. Consensus was defined as ≥ 70% agreement (score of 7–9) and ≤ 15% disagreement (score of 1–3). Statistical measures such as content validity ratio and Kendall’s coefficient of concordance (W) were calculated to assess agreement levels. Results: Of the 46 assessment-related statements, 17 (37%) reached consensus, though overall pre-treatment assessments showed weak agreement (Kendall’s W = 0.319). Key areas of agreement included the use of the visual analogue scale, numeric rating scale, and validated questionnaires for pre-treatment evaluation. Five statements, such as the use of computed tomography, magnetic resonance imaging, and angiography for diagnosing pulsatile tinnitus, achieved over 90% agreement. For treatment outcome measurements, 8 of 12 statements (67%) reached a consensus, with moderate agreement (Kendall’s W = 0.513). Validated questionnaires and psychoacoustic tests were recommended for evaluating treatment effects within 12 weeks. While standardized imaging for pulsatile tinnitus and additional clinical tests were strongly recommended, full consensus was not achieved across all imaging modalities. Conclusion This study provides actionable recommendations for tinnitus assessment and treatment evaluation, emphasizing the use of standardized tools and individualized approaches based on patient needs. These findings offer a practical framework to enhance consistency and effectiveness in tinnitus management within Korean clinical settings.

Background: Tinnitus is a bothersome condition associated with various mechanisms of action. Although treatment methods vary according to these mechanisms, standardized guidelines would benefit both patients and clinicians. We conducted a Delphi study, a method that collects expert opinions through multiple rounds of questionnaires, to reach a consensus on tinnitus treatment with professional experts. Methods: A two-round modified Delphi survey was conducted to develop a clinical consensus on tinnitus treatment. The experts scored each statement on a scale of 1 (highest disagreement) to 9 (highest agreement) for their level of agreement on tinnitus treatment.Consensus was defined when 75% or more of the participants scored 7–9, and 15% or less scored 1–3. To ensure reliability of the responses, the content validity ratio and Kendall’s coefficient of concordance were evaluated. Results: Approximately 19 of 31 statements reached a consensus. All 3 statements reached a consensus regarding the candidates for treatment. Regarding treatment, 3 of 8 statements on medication, 2 of 2 statements on tinnitus retraining therapy/cognitive behavioral therapy, and 5 of 7 statements on auditory rehabilitation reached a positive consensus. Although all 6 statements regarding miscellaneous treatment reached a consensus, most were negatively agreed. For treatment with neuromodulation, none of the 5 statements reached a consensus. Conclusion The experts reached a high level of consensus on treatment candidates, tinnitus retraining therapy/cognitive behavioral therapy, and auditory rehabilitation in this modified Delphi study. The results of this study can provide beneficial and practical information for clinicians regarding the treatment of tinnitus.