Segmentation of common and internal carotid arteries from 3D ultrasound images based on adaptive triple loss
Mingjie Jiang1, Yuan Zhao1, Bernard Chiu1
1Department of Electrical Engineering, City University of Hong Kong, Hong Kong, Hong Kong.
Insights
This study introduces a novel AI segmentation method for 3D ultrasound carotid images, accurately measuring vessel wall thickness to aid atherosclerosis research.
Area of Science:
- Medical imaging analysis
- Artificial intelligence in healthcare
- Cardiovascular disease research
Background:
- Accurate measurement of carotid artery wall volume and thickness is crucial for assessing atherosclerosis.
- Existing segmentation techniques often fail to segment both common and internal carotid arteries (CCA and ICA).
- Accelerated quantification of vessel wall volume (VWV) and vessel-wall-plus-plaque thickness (VWT) is needed for clinical research.
Purpose of the Study:
- To develop an accurate and efficient method for segmenting media-adventitia boundaries (MAB) and lumen-intima boundaries (LIB) in both CCA and ICA from 3D ultrasound (3DUS) volumes.
- To enable faster quantification of VWV and VWT for anti-atherosclerotic treatment evaluation.
Main Methods:
- A two-channel U-Net deep learning model was employed for segmentation.
- A novel adaptive triple Dice loss (ADTL) function was utilized to drive the segmentation process.
- Training data augmentation incorporated longitudinal interpolation for artery continuity, and test-time augmentation (TTA) with majority voting was applied for robust segmentation.
Main Results:
- The proposed method achieved high accuracy, with Dice similarity coefficients (DSC) of 95.1% ± 4.1% for CCA MAB and 91.6% ± 6.6% for CCA LIB.
- Segmentation of the ICA also yielded strong results: 94.2% ± 3.3% for MAB and 89.0% ± 8.1% for LIB.
- Both ADTL and TTA significantly improved segmentation accuracy for most boundaries, demonstrating their effectiveness.
Conclusions:
- The developed two-channel U-Net with ADTL and TTA provides accurate and efficient segmentation of CCA and ICA from 3DUS data.
- This approach has the potential to significantly accelerate the clinical application of 3DUS measurements in carotid atherosclerosis research.
- The method facilitates more precise assessment of anti-atherosclerotic therapies.
Purpose:
Vessel wall volume (VWV) and localized vessel-wall-plus-plaque thickness (VWT) measured from three-dimensional (3D) ultrasound (US) carotid images are sensitive to anti-atherosclerotic effects of medical/dietary treatments. VWV and VWT measurements require the lumen-intima (LIB) and media-adventitia boundaries (MAB) at the common and internal carotid arteries (CCA and ICA). However, most existing segmentation techniques were capable of segmenting the CCA only. An approach capable of segmenting the MAB and LIB from the CCA and ICA was required to accelerate VWV and VWT quantification.
Methods:
Segmentation for CCA and ICA was performed independently using the proposed two-channel U-Net, which was driven by a novel loss function known as the adaptive triple Dice loss (ADTL) function. The training set was augmented by interpolating manual segmentation along the longitudinal direction, thereby taking continuity of the artery into account. A test-time augmentation (TTA) approach was applied, in which segmentation was performed three times based on the input axial images and its flipped versions; the final segmentation was generated by pixel-wise majority voting.
Results:
Experiments involving 224 3DUS volumes produce a Dice similarity coefficient (DSC) of 95.1% ± 4.1% and 91.6% ± 6.6% for the MAB and LIB, in the CCA, respectively, and 94.2% ± 3.3% and 89.0% ± 8.1% for the MAB and LIB, in the ICA, respectively. TTA and ATDL independently contributed to a statistically significant improvement to all boundaries except the LIB in ICA.
Conclusions:
The proposed two-channel U-Net with ADTL and TTA can segment the CCA and ICA accurately and efficiently from the 3DUS volume. Our approach has the potential to accelerate the transition of 3DUS measurements of carotid atherosclerosis to clinical research.
