Hyunsook Hong's research while affiliated with Seoul National University Hospital and other places

... Recent guidelines of the Korean Society of Radiology (KSR) align with the views of these organizations [3]. The findings in the newly published study by Dr. Choi et al. [4] urges rethinking global and domestic fasting practices. ...

... Consistent findings were also observed by Li et al. [29]. Patients with SN had more than two times death or recurrence risk than those with PSN after adjusting for other confounding factors [30]. Notably, we observed that even for patients with lymph node metastasis, the SN group still had a poorer DFS than the PSN group. ...

... In routine clinical practice, radiologists often misdiagnose intracranial aneurysms due to their small size, complexity of intracranial vasculature [5], and heavy workload [6], which may impose risk for patients with intracranial aneurysms, which can cause nontraumatic subarachnoid hemorrhage (SAH) with a high rate of mortality and disability [7][8][9]. AI has demonstrated improved reader performance on limited retrospective datasets [10][11][12][13], while critics point out that AI systems may be less helpful than retrospective data would suggest and almost all previous AI studies were performed in an open-label design (comparing "AI + readers" vs AI alone or readers alone), which would introduce Hawthorne effect and automation biases [14][15][16]. There is a lack of high-level evidence for the real-world evaluation of AI systems, especially prospective, real-world, doubleblinded randomized controlled trials (RCTs), which are the highest standard evidence in this field. ...

... We used this measurement approach instead of using whole-body composition results obtained on chest CT because not all patients had exactly the same CT scan coverage. Indeed, multi-slice averaged body composition analysis results at the level of T12-L1 are known to be well correlated with whole-body composition [31]. Therefore, in this study, the subcutaneous and visceral fat areas were summed and defined as body fat area to represent whole-body fat. ...

... In recent years, the combination of deep learning and spectral CT has been used more and more in the diagnosis of liver cancer. Bae et al. (2023) through combined low monoenergetic images and deep learning denoising, could obtain high conspicuity of HCC and acceptable image noise while lowering the amount of contrast medium. They thought that the findings on noninferiority for the 50 keV deep learning denoising images in terms of image quality and lesion conspicuity, compared with model-based iterative reconstruction images of the standardcontrast-dose group, may contribute to the wider clinical use ...

... Kim H et al. [24] also used deep learning methods for calculating total lung capacity for predicting survival in idiopathic pulmonary fibrosis. The mean absolute difference and within-subject SD between ground truth and estimated lung volume were 0.69 L and 0.73 L, respectively. ...

... Moreover, CT has aimed to be used as an opportunistic screening modality for various conditions beyond the desired clinical indication and is a potentially useful modality for diagnosing type 2 diabetes mellitus, but has not yet been used as such [16]. Yet their high cost, limited accessibility, high-dose radiation exposure (in CT), and labor-intensive segmentation procedure limit their usage in clinical research [17]. To overcome these shortcomings, we took advantage of muscle images obtained from DBT to creatively enhance its use beyond breast tissue evaluation as an opportunistic screening tool for T2DM. ...

... There is some controversy between existing guidelines, especially for early-stage NSCLC, as the detection rate of distant metastases is very low [3,4,[7][8][9][10][11]. In general, an initial routine brain MRI is unnecessary for patients with GGNs and subsolid nodules [2] because preoperative staging does not have prognostic benefit for survival [12,13]. ...

... Therefore, future studies should aim to enlist a broader range of respondents to resolve the problem of the lack of generalizability of the study results. To increase the validity of the study, a larger sample size is necessary (Hong et al., 2021). Hence, confidence in this study may be further improved. ...

... The Chinese national guideline (trial version 5) temporarily used typical CT findings of viral pneumonia for making a clinical diagnosis of suspected COVID-19 cases in Wuhan and Hubei [68]. Given the pooled diagnostic CT performance, we may imagine why the first-line diagnosis/triage use of CT could be inevitably considered a practical supplemental option, although CT delivered radiation exposure [69] and it was unknown how well CT interpretation for typical COVID-19 findings was standardized in Wuhan and Hubei. Chest CT findings can substantially overlap between COVID-19, influenza, and organizing pneumonia [70], and our findings on the pooled CT performance should be cautiously interpreted. ...