Blood vessel segmentation using multi-scale quadrature filtering

Abstract

The segmentation of blood vessels is a common problem in medical imaging and various applications are found in diagnostics, surgical planning, training and more. Among many different techniques, the use of multiple scales and line detectors is a popular approach. However, the typical line filters used are sensitive to intensity variations and do not target the detection of vessel walls explicitly. In this article, we combine both line and edge detection using quadrature filters across multiple scales. The filter result gives well defined vessels as linear structures, while distinct edges facilitate a robust segmentation. We apply the filter output to energy optimization techniques for segmentation and show promising results in 2D and 3D to illustrate the behavior of our method. The conference version of this article received the best paper award in the bioinformatics and biomedical applications track at ICPR 2008.

Multiorgan Segmentation Using Coherent Propagating Level Set Method Guided by Hierarchical Shape Priors and Local Phase Information

“…The output of this pair of filters is represented by a complex number where the real part is the output of the ridge-picking filter and the imaginary part is the output of the edge-picking filter. The argument of this complex number in the complex plane is referred to as the local phase [10], θ in Fig. 10.6c.…”

“…10.6c. The magnitude (q) of the complex number is called local energy [10]. Local phase is often used as a promising alternative to image gradient for image segmentation [1,11,18].…”

Unsupervised level set parameterization using multi-scale filtering

“…Multi-scale filtering has been repeatedly applied for retrieving localized edge regions [11]- [13]. The key idea is to compute the scale pyramid by subsampling an image in order to investigate the scale with the largest certainty of containing a target edge.…”

Phase-Based Level Set Segmentation of Ultrasound Images

“…This uses local phase information derived from the monogenic signal, which is a multidimensional extension of the analytic signal [2], [3], [4]. Phase information has been used in numerous applications in the literature, such as image registration [5], stereo matching [6], filtering [7], enhancement [8], corner and edges detection [9], [10] and segmentation [11], [12], [13], [14]. Phase-based processing has attracted a lot of attention in image analysis, but probably not still enough in ultrasound image segmentation, see [1].…”