| Resumo: | The study presented here addresses segmentation methodologies applied to Doppler ultrasound images of the carotid bifurcation acquired as a sequence of 2D ultrasound transverse images. Initial identification of image elements representing transversal cuts of the artery followed by parameter analysis enables a fast methodology for lumen and artery wall identification and segmentation for the complete image sequence. Different arterial geometries, along with the presence of plaque, artefacts and speckle noise, can modify the expected circular shapes of the arteries making identification extremely difficult. Initial identified parameters are crucial for the analysis of image behavior across the bifurcation suggesting methodologies for image correction and irregularity detection on the artery representation. Statistical analysis of the accuracy of the bifurcation modelling from transverse image segmentation validates the methodology results as compared to results from medical manual segmentation and results from segmentation of longitudinal Doppler ultrasound images. (c) Springer Nature Switzerland AG 2019.
|
|---|