Five multiresolution-based calcium volume measurement techniques from coronary IVUS videos: A comparative approach

Area of Science:

• Medical Imaging
• Computational Cardiology
• Image Processing

Background:

• Fast intravascular ultrasound (IVUS) video processing is crucial for accurate calcium volume computation during percutaneous coronary intervention (PCI) planning.
• Nonlinear multiresolution techniques are commonly used to accelerate video processing by down-sampling frames, but can impact image quality.

Purpose of the Study:

• To evaluate and compare the performance of four different segmentation methods combined with five multiresolution techniques for IVUS calcium volume measurement.
• To identify the optimal combination for fast and accurate calcium volume computation in PCI planning.

Main Methods:

• Four segmentation methods (Threshold-based, Fuzzy c-Means (FCM), K-means, Hidden Markov Random Field (HMRF)) were tested with five multiresolution techniques (bilinear, bicubic, wavelet, Lanczos, Gaussian pyramid), creating 20 combinations.
• Performance was assessed using 38,760 IVUS frames from 19 patients, evaluating computational time, calcium volume accuracy, image quality degradation, and quality assessment ratios.

Main Results:

• Fuzzy c-Means (FCM) segmentation combined with wavelet-based multiresolution demonstrated the best performance, achieving significant improvements in computational time (77.15% for FCM, 74.07% for wavelet).
• Wavelet multiresolution yielded the highest precision-of-merit (PoM), Jaccard Index, and Dice Similarity, indicating superior accuracy and image quality preservation.
• A bias correction approach was introduced, enhancing mean calcium volume similarity, with bicubic interpolation showing the largest improvement (4.13%) after correction.

Conclusions:

• The study successfully demonstrated time improvements in IVUS calcium volume computation without compromising image quality.
• The combination of FCM segmentation and wavelet-based multiresolution emerged as the optimal system among the 20 evaluated combinations for efficient and accurate calcium volume measurement.