Abstract WP134: Quantitatively Monitoring Regression or Progression in Intracranial Atherosclerotic Plaques Using 3D Vessel Wall Imaging and Deep-learning-based Vessel Wall Analysis

    loading  Checking for direct PDF access through Ovid

Abstract

Introduction: Intracranial atherosclerotic disease (ICAD) is one of the most common causes of ischemic stroke worldwide. Despite aggressive medical management, the rate of recurrent stroke is 13% at 1 year. Intracranial vessel wall imaging (VWI) is a noninvasive, “looking-beyond-the-lumen” imaging method that can directly characterize the geometric and signal features of ICAD lesions. The present work sought to assess the feasibility of quantitatively monitoring regression or progression of ICAD plaques using VWI-based methods.

Methods: Eight ischemic stroke patients (1F, 7M; age 27-66 ys) treated with intensive medical therapy underwent initial (4 days - 4 months of onset) and follow-up 3D VWI (6-13 months). Images were randomized and reviewed by two neuroradiologists to determine the culprit lesion in each subject. A custom-designed deep-learning-based intracranial vessel analysis method was used to segment vessel wall boundaries and quantify the following features of the culprit lesion, including peak normalized wall index (NWI), plaque volume, pre-contrast plaque-wall contrast ratio (CR), and post-contrast plaque enhancement ratio (ER).

Results: No subjects except for subject #4 had a recurrent stroke during the follow-up period. The four plaque features exhibited different change patterns as shown in Figure. Subject #1 and #4 demonstrated an increase in plaque ER, volume, and peak NWI; subject #7 demonstrated an increase in plaque CR, volume and peak NWI. Other patients had a decrease or no change in these features.

Conclusions: In this work, the interrogated plaque features demonstrated regression in most of patients after intensive medical therapy. Elevated values in some of these features appeared positively associated with stroke recurrence. Temporal changes in these features may have a strong indication on culprit lesions’ response to medical therapy. In conclusion, it is feasible to quantitatively monitor plaque-level treatment response.

Related Topics

    loading  Loading Related Articles