Carotid artery wall segmentation in ultrasound image sequences using a deep convolutional neural network

Abstract

The objective of this study is the segmentation of the intimamedia complex of the common carotid artery, on longitudinal ultrasound images, to measure its thickness. We propose a fully automatic region-based segmentation method, involving a supervised deep-learning approach based on a dilated U-net network. It was trained and evaluated using a 5-fold cross-validation on a multicenter database composed of 2176 images annotated by two experts. The resulting mean absolute difference (< 120 µm) compared to reference annotations was less than the inter-observer variability (≈ 180 µm). With a 98.7% success rate, i.e., only 1.3% cases requiring manual correction, the proposed method has been shown to be robust and thus may be recommended for use in clinical practice.

Ultrasound-Based Image Analysis for Predicting Carotid Artery Stenosis Risk: A Comprehensive Review of the Problem, Techniques, Datasets, and Future Directions

“…and 84 trans. Segmentation dilated-Unet [77] 30 (Aug applied) Segmentation CNN [110] 2716 images (CUBS) Segmentation dilated-Unet…”

“…Supervised deep-learning approaches based on a dilated U-net network, Unet++, segnet, and Segnet + Unet and Deeplab have been studied by [43,49,67,110] as shown in Table 9. The several regions of segmentation performed on the state of the art are as shown in Figure 10.…”

“…The far wall was detected, and then IMC was segmented from the near wall. A multi-stage approach was used with cross-validation [110]. An inference from the evaluation noted that a failure of the first step, which is far wall detection, will fail the segmentation as a whole, and so the second stage is dependent on the success and accuracy of the first stage.…”