Abstract 4370143: Plaque Eccentricity and Delayed Flow as Predictors of Vulnerable Lesions in Unstable Angina: Insights from Dynamic Angiography, IVUS, and Deep Learning for Personalized Management

L Loc Vu (School of Medicine, Tan Tao University, Long An, Viet Nam) D Dung Ho (Thong Nhat Hospital, Hochiminh City, Viet Nam) K Khiem Ngo (UTRGV, Harlingen, Texas, United States) M Minh Le T Thuan Huynh (Thong Nhat Hospital, Hochiminh City, Viet Nam) Q Quan Nguyen (Princeton University) T Trien Tran (Methodist Hospital, Merrillville, Indiana, United States) B Bao Huy Le (Methodist Hospital, Laporte, Indiana, United States) T Thien Tan Tri Tai Truyen T Thach Nguyen (Interventional Cardiology Department, Methodist Hospital and St. Mary Medical Center, Merrillville, Indiana, United States)

Abstract

Background: In coronary hemodynamics, laminar flow is a protective mechanism, and turbulent flow is a detrimental factor. In clinical practice, for patients with unstable angina, the necessity of intervention for moderate lesions (50-70% stenosis) is always questioned. In this study, we comprehensively evaluated the risk factors that may lead these plaques to progress to myocardial infarction using angiography, IVUS, and Deep Learning. Hypothesis: Vulnerable lesions are usually associated with plaque eccentricity and stagnant flow, causing chest pain. Methods: Patients with a single lesion admitted with diagnosed of unstable angina underwent the dynamic angiographic technique with IVUS support. First, the coronary artery was fully injected with contrast. After the injection stopped, the blood (white) began streaming in to replace the contrast (black). The flow characteristics and movements will be recorded through the disappearance of the contrast opacity with 15 frames/second. In aditionally, the arterial phase (AP) was calculated as the time from frames with full contrast to frames with washed-out contrast. Moreover, IVUS was performed for evaluate plaque features (eccentricity, calcification and vulnerable). At the same time, deep learning (DL) models were built based on independent datasets (225 angiogram videos). The DL program was constructed using a combination of U-Net and DenseNet-121. The segmentation model and a convolutional neural network were used to detect the starting frame, ending frame of AP. Results: Fifty patients met inclusion criteria (72% males) with a mean age of 66.2 ± 9.5 years. IVUS analysis revealed that 84% (42 patients) had plaque eccentricity without (or minimal) calcification. Contrast stagnation was observed for a longer duration in eccentric coronary regions compared to non-eccentric plaque regions, as quantified by DL analysis of the arterial phase (24.5 +/- 1.6 frames vs 30.1+/- 1.4 frames, p < 0.05). After stenting, the stagnant flow was eliminated, and flow was restored to 25 ± 2.1 frames. Conclusion: Plaque eccentricity along with a prolonged arterial phase (stagnant flow) are causes of chest pain in unstable angina. This is a personalized indication for PCI, even if they are moderate lesions.<div id="gtx-trans" style="position: absolute; left: 393px; top: 368.844px;"><div class="gtx-trans-icon"> </div></div>

Article Details

Journal Circulation
Volume / Issue Vol. 152, Issue Suppl_3
Published November 04, 2025
ISSN 0009-7322
Publisher Lippincott Williams & Wilkins

Journal Info

Circulation

Lippincott Williams & Wilkins

ISSN: 0009-7322 Health Sciences

Authors (10)

L

Loc Vu

School of Medicine, Tan Tao University, Long An, Viet Nam

D

Dung Ho

Thong Nhat Hospital, Hochiminh City, Viet Nam

K

Khiem Ngo

UTRGV, Harlingen, Texas, United States

M

Minh Le

T

Thuan Huynh

Thong Nhat Hospital, Hochiminh City, Viet Nam

Q

Quan Nguyen

Princeton University

T

Trien Tran

Methodist Hospital, Merrillville, Indiana, United States

B

Bao Huy Le

Methodist Hospital, Laporte, Indiana, United States

T

Thien Tan Tri Tai Truyen

T

Thach Nguyen

Interventional Cardiology Department, Methodist Hospital and St. Mary Medical Center, Merrillville, Indiana, United States