Keunwoo Park

Keunwoo Park

Auto-generated profile, based on publicly available information

Keunwoo Park

Department of Biomedical Engineering, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Republic of Korea

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
10 October 2022 in PLOS ONE
Verified
Journal Article
Open access
PDF file
YCYeonjoo Choi
Yeonjoo Choi
JCJihye Chae
Jihye Chae
KPKeunwoo Park
Keunwoo Park
JHJaehee Hur
Jaehee Hur
JKJihoon Kweon
Jihoon Kweon
JAJoong Ho Ahn
Joong Ho Ahn
Backgrounds and objective: Evaluating the tympanic membrane (TM) using an otoendoscope is the first and most important step in various clinical fields. Unfortunately, most lesions of TM have more than one diagnostic name. Therefore, we built a database of otoendoscopic images with multiple diseases and investigated the impact of concurrent diseases on the classification performance of deep learning networks. Study design: This retrospective study investigated the impact of concurrent diseases in the tympanic membrane on diagnostic performance using multi-class classification. A customized architecture of EfficientNet-B4 was introduced to predict the primary class (otitis media with effusion (OME), chronic otitis media (COM), and ’None’ without OME and COM) and secondary classes (attic cholesteatoma, myringitis, otomycosis, and ventilating tube). Results: Deep-learning classifications accurately predicted the primary class with dice similarity coefficient (DSC) of 95.19%, while misidentification between COM and OME rarely occurred. Among the secondary classes, the diagnosis of attic cholesteatoma and myringitis achieved a DSC of 88.37% and 88.28%, respectively. Although concurrent diseases hampered the prediction performance, there was only a 0.44% probability of inaccurately predicting two or more secondary classes (29/6,630). The inference time per image was 2.594 ms on average. Conclusion: Deep-learning classification can be used to support clinical decision-making by accurately and reproducibly predicting tympanic membrane changes in real time, even in the presence of multiple concurrent diseases.

10 November 2026 in Scientific Reports
Verified
Journal Article
Open access
PDF file
KPKeunwoo Park
Keunwoo Park
JCJihye Chae
Jihye Chae
JKJihoon Kweon
Jihoon Kweon
JAJoong Ho Ahn
Joong Ho Ahn
Text-to-image synthesis in medicine remains challenging because clinically relevant visual attributes are often not fully specified by text prompts, which can limit the controllability and diversity of generated images. We propose Controllable Latent Space of Non-Prompt Attributes (CLNA), a framework that models non-prompt visual attributes within a continuous latent space for diagnosis-conditioned image synthesis, implemented with a style encoder and dual cross-attention within a Stable Diffusion backbone. We evaluated CLNA on otoendoscopic image synthesis for three diagnostic classes: normal tympanic membrane, otitis media with effusion, and chronic otitis media. Compared with the baseline Stable Diffusion model, CLNA improved image fidelity in the domain-adapted feature space, reducing Fréchet Inception Distance from 52.91 to 10.11 against the held-out internal test set, and improved feature-space recall from 57.67% to 90.22%. In classification experiments, replacing 10–40% of the real training data with CLNA-generated images at \(\:\gamma\:=0.5\) gave F1 scores of 93.57–94.08% on the internal test set (vs. 94.46% averaged across 10 real-only runs) and 84.81–85.30% on the external test set (vs. 83.76% from a single real-only run). With complete replacement of the training data by generated images, CLNA outperformed the baseline model on both the internal test set (78.53% vs. 70.25%) and the external test set (72.37% vs. 54.26%). When the synthetic-only training set was increased 10-fold, CLNA further improved F1 scores to 83.21% on the internal test set and 76.77% on the external test set. These findings suggest that CLNA can control the magnitude of sampled latent variation through a style parameter, thereby improving diversity in otoendoscopic text-to-image synthesis. CLNA-generated images could support downstream classifier training under reduced use of real images.

30 October 2026 in JMIR Medical Informatics
Verified
Journal Article
Open access
SYSe Eun Yi
Se Eun Yi
JCJihye Chae
Jihye Chae
KPKeunwoo Park
Keunwoo Park
YCYeon Joo Choi
Yeon Joo Choi
JKJihoon Kweon
Jihoon Kweon
JAJoong Ho Ahn
Joong Ho Ahn
Background: Accurate diagnosis of middle ear diseases remains challenging due to the specialized nature of otoscopic interpretation and the global shortage of otolaryngologists. Objective: We aimed to evaluate the impact of a deep learning–based AI support system on the diagnostic performance of clinicians with varying levels of expertise, specifically focusing on its ability to bridge the gap between trainees and experts. Methods: In this retrospective reader study, a deep learning model (EfficientNet-B4) was developed using a large-scale dataset of 7335 endoscopic images. The model was designed for multi-class classification, simultaneously identifying primary categories (chronic otitis media, otitis media with effusion, or none [indicating the absence of the first two]) and secondary features (attic retraction, myringitis, otomycosis, and ventilation tube). To assess clinical impact, diagnostic performance was compared among 3 otology professors, 5 senior residents, and 5 junior residents using an independent clinical assessment set of 100 images. Participants performed diagnoses both with and without AI-generated predictions, with a time constraint of 1 minute per image to simulate real-world clinical pressure. Results: In 5-fold cross-validation, the deep learning model achieved 94.8% accuracy for the primary classification. The stand-alone AI achieved diagnostic accuracies ranging from 94.0% to 99.0% in the clinical assessment set. AI support significantly enhanced diagnostic accuracy across all groups for primary classification, with improvements of 11.0% for otology professors, 28.8% for senior residents, and 32.2% for junior residents ( P <.001). Accuracy increased from 78.7% to 89.7% for otology professors, from 65.0% to 93.8% for senior residents, and from 54.4% to 86.6% for junior residents. The accuracy gap between the most experienced (otology professors) and least experienced (junior residents) groups was reduced from 24.3% to 3.1%. For secondary categories, significant improvements were observed in diagnosing attic retraction and myringitis among residents. AI-supported accuracy for attic retraction was higher in senior residents than that of the stand-alone AI, although the 95% CIs overlapped substantially, so the difference was not statistically significant. Conclusions: Our findings demonstrate that deep learning–based AI support substantially improves the diagnostic accuracy of middle ear diseases, particularly for less experienced clinicians. By narrowing the performance gap between trainees and experts, this system shows high potential as a valuable adjunct for clinical decision-making and medical education in otology.