Fully Automated Lipid Pool Detection Using Near Infrared Spectroscopy.

Area of Science:

• Cardiovascular Imaging
• Medical Device Technology
• Computational Pathology

Background:

• Detecting vulnerable plaque, prone to rupture, remains a challenge for cardiologists.
• Standard angiography cannot identify lipid core-containing plaque.
• Near-infrared spectroscopy integrated with intravascular ultrasound (NIRS-IVUS) offers plaque characterization capabilities.

Purpose of the Study:

• To develop a fully automated algorithm for lipid pool detection on NIRS images.
• To enable better characterization of atherosclerotic plaque.
• To provide a tool for interpreting NIRS-IVUS data.

Main Methods:

• An algorithm comprising four stages: preprocessing, artifact segmentation, lipid area detection, and Lipid Core Burden Index (LCBI) calculation.
• Analysis of 31 NIRS chemograms using the proposed algorithm and a commercial system.
• Comparison of total LCBI, maximal LCBI in 4 mm blocks, and maximal LCBI in 2 mm blocks.

Main Results:

• The proposed algorithm demonstrated good agreement and correlation with a commercial system.
• Intraclass correlation (ICC) and Bland-Altman plots confirmed the reliability of the algorithm.
• Accurate calculation of key LCBI metrics for plaque assessment.

Conclusions:

• The developed algorithm provides fully automated lipid pool detection on NIRS images.
• This tool is suitable for offline data analysis in cardiology.
• The algorithm is designed for future augmentation and integration into new projects.