Assessing the transferability of micro-computed tomography (CT)-based plaque segmentation to clinical CT using...

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Radiology

Background:

• Noninvasive plaque characterization is limited in peripheral arterial obstructive disease (PAOD).
• Clinical CT relies on lesion length and stenosis, not plaque composition, for treatment planning.
• Ex vivo micro-computed tomography (micro-CT) offers quasihistological plaque assessment.

Purpose of the Study:

• To evaluate the transferability of plaque structural information from micro-CT to clinical CT.
• To assess the efficacy of a super-resolution (SR) framework for enhancing CT plaque characterization.
• To define the limitations of current CT technology for plaque analysis in PAOD.

Main Methods:

• Popliteal artery segments from six PAOD patients were analyzed using micro-CT and histology.
• Convolutional neural networks were trained on micro-CT data for plaque segmentation.
• Low-resolution clinical CT images were upsampled using a Laplacian pyramid SR approach.

Main Results:

• Segmentation of calcified plaque components on micro-CT yielded Dice scores of 0.58–0.67 (low-to-intermediate agreement).
• SR-CT image segmentation showed nonrandom identification of calcified structures but with significant heterogeneity.
• The SR approach enabled exploratory assessment but did not overcome the fundamental resolution gap.

Conclusions:

• Super-resolution enables exploratory assessment of plaque information transfer but does not fully bridge the resolution gap in CT.
• Current CT-based plaque characterization is limited, necessitating further advancements.
• Findings provide a framework for evaluating future imaging technologies in PAOD.