Emine Bensu Erdemir

Dr. Emine Bensu Erdemir


Research Focus & Metrics

Academic Career Length

1Year (2026-Present)

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

Representative Publications

View All Publications
5 September 2026
Verified
Conference Paper
Closed access
EEEmine Bensu Erdemir
Emine Bensu Erdemir
GDGorkem Durak
Gorkem Durak
SGSule Gumurdu
Sule Gumurdu
DSDeniz Sobayapan
Deniz Sobayapan
SESukru Mehmet Erturk
Sukru Mehmet Erturk
ZBZuleyha Bingol
Zuleyha Bingol
Idiopathic pulmonary fibrosis (IPF) shows marked heterogeneity in fibrotic lung involvement and clinical outcomes, limiting the prognostic accuracy of physiology-based tools such as the GAP index. Artificial intelligence (AI)–based quantitative CT enables objective volumetric assessment of fibrotic burden and may enhance risk stratification. We developed an AI-driven three-dimensional quantitative CT framework in multidisciplinary discussion–confirmed IPF patients. An ensemble of deep learning models (LungMask, nnU-Net, 3D-ResUNet, Swin-UNETR) was used to segment lung parenchyma and fibrotic components, allowing calculation of the pathological lung volume–to–total lung volume ratio. This imaging biomarker was integrated into the GAP index to generate a radiology-enhanced prognostic model. Survival and ROC analyses were performed. Among 116 patients (mean age 64.2 years), higher pathological lung volume and lower preserved lung volume were significantly associated with mortality (p<0.05). The AI-derived pathological lung volume ratio showed stronger prognostic performance than GAP alone (AUC 0.795 vs 0.712). Integration with GAP further improved discrimination (AUC 0.805) and enabled clear survival stratification, with median survival of 100 months in low-risk versus 48 months in high-risk patients (p<0.001). AI-integrated quantitative CT enhances prognostic precision of the GAP index while preserving clinical applicability in IPF. erj;68/suppl_70/PA6239/F1 F1 F1

13 November 2026 in Preprints.org
Verified
Preprint
Open access
PDF file
EEEmine Bensu Erdemir
Emine Bensu Erdemir
Gorkem DurakGorkem Durak
Gorkem Durak
BKBetul Buse Karabıyık
Betul Buse Karabıyık
Muhammed Enes TasciMuhammed Enes Tasci
Muhammed Enes Tasci
SESukru Mehmet Erturk
Sukru Mehmet Erturk
ZBZuleyha Bingol
Zuleyha Bingol
Background: Idiopathic pulmonary fibrosis (IPF) is a devastating chronic lung disease with an incompletely understood course. Limited quantitative tools complicate consistent patient assessment, and visual evaluation of radiological progression varies between observers. This study aimed to assess baseline and one-year CT scans using artificial intelligence (AI)–assisted 3D quantitative Computed Tomography (QCT), integrating radiological changes with clinical findings to objectively characterize disease progression and examine its association with prognosis. Methods: In this retrospective cohort of 83 patients with IPF, we evaluated patients using an AI-assisted quantitative CT framework for prognostic assessment. Whole-lung segmentation was performed using a pretrained deep learning model (LungMask), followed by semi-automated segmentation of fibrotic abnormalities with radiologist-guided refinement. Quantitative imaging biomarkers, including honeycombing volume, non-honeycomb fibrotic lung volume, preserved lung volume, and the pathological lung volume-to-total lung volume ratio, were derived from 3D QCT analysis. Baseline and one-year measurements and their changes were evaluated in relation to mortality, pulmonary function test, gender, age, physiology score (GAP) index, and composite physiologic index (CPI). Between-group comparisons were performed using the Mann–Whitney U test, paired comparisons using the Wilcoxon signed-rank test, and correlations using Spearman’s rank correlation analysis. Statistical analyses were conducted using SPSS version 27.0. Results: A total of 83 patients with IPF were included in the study (mean age: 64.1 ± 8.5 years; 69.9% male). First-year mortality was 14.5% (n = 12), and overall mortality was 37.3% (n = 31) over a median follow-up of 48 months. Compared with the survival group, the mortality group had greater baseline non-honeycomb fibrotic lung volume (1002.2 ml ± 532.3 vs. 628.4 ± 460.2), a higher pathological-to-total lung volume ratio, comprising honeycomb and non-honeycomb fibrosis (0.34 ± 0.16 vs. 0.18 ± 0.15), and lower preserved lung volume (2317.9 ± 961.4 vs. 3531.6 ± 1579.2; all p < 0.001). Over one year, honeycomb and non-honeycomb fibrotic lung volume ratios increased, while the preserved lung volume ratio decreased in both groups (all p < 0.05). Patients with a 10% forced vital capacity (FVC) decline had a greater increase in the non-honeycomb fibrotic-to-total lung volume ratio (0.08 ± 0.10 vs. 0.02 ± 0.20; p = 0.044). The baseline pathological-to-total lung volume ratio correlated with baseline GAP (r = 0.343, p = 0.002), baseline CPI (r = 0.431, p < 0.001), and one-year CPI (r = 0.560, p < 0.001). One-year QCT changes did not differ significantly between the mortality and survival groups. Conclusion: 3D semi-automated, AI-assisted QCT provided an objective assessment of radiological progression in IPF. Greater baseline fibrotic involvement and lower preserved lung volume were associated with mortality, and quantitative imaging measures correlated with GAP and CPI. Although one-year QCT changes did not differ by mortality status, increased non-honeycomb fibrotic involvement was associated with FVC decline. These findings support QCT as a complementary tool for assessing disease extent and progression alongside functional and prognostic indices.