Automated Vessel Segmentation Using Infinite Perimeter Active Contour Model with Hybrid Region Information with Application to Retinal Images

Abstract

Automated detection of blood vessel structures is becoming of crucial interest for better management of vascular disease. In this paper, we propose a new infinite active contour model that uses hybrid region information of the image to approach this problem. More specifically, an infinite perimeter regularizer, provided by using L(2) Lebesgue measure of the γ -neighborhood of boundaries, allows for better detection of small oscillatory (branching) structures than the traditional models based on the length of a feature's boundaries (i.e., H(1) Hausdorff measure). Moreover, for better general segmentation performance, the proposed model takes the advantage of using different types of region information, such as the combination of intensity information and local phase based enhancement map. The local phase based enhancement map is used for its superiority in preserving vessel edges while the given image intensity information will guarantee a correct feature's segmentation. We evaluate the performance of the proposed model by applying it to three public retinal image datasets (two datasets of color fundus photography and one fluorescein angiography dataset). The proposed model outperforms its competitors when compared with other widely used unsupervised and supervised methods. For example, the sensitivity (0.742), specificity (0.982) and accuracy (0.954) achieved on the DRIVE dataset are very close to those of the second observer's annotations.

Exploiting the Retinal Vascular Geometry in Identifying the Progression to Diabetic Retinopathy Using Penalized Logistic Regression and Random Forests

“…Although many state-of-the-art automated tools have been proposed in literature, utilizing many different methods e.g. wavelets and edge location refinement both to segment and measure retinal vessels using image profiles, computed across a spline fit of each detected centerline [5], an infinite active contour model, using an infinite perimeter regularizer and multiple region information [29] or using neighbourhood estimator before filling filter [2], still they cannot be used in large studies for evaluating the progression of the disease. Their consistency and accuracy/precision as well as the measurement errors across datasets with different image quality , do not allow us to find these subtle changes that occur inside the vasculature over time, and which we are trying to identify in the same retinas.…”

Retinal Vessel Segmentation via Structure Tensor Coloring and Anisotropy Enhancement

“…The area under a receiver operating characteristic curve (AUC) is commonly used for the measuring the performance of the classifiers for different thresholds which is not applicable for unsupervised methods. Another definition such as AUC = (Sensitivity + Specificity)/2 is more suitable for unsupervised methods like the proposed study [45,46].…”

Iterative deep convolutional encoder-decoder network for medical image segmentation

“…In hand-craft models, AT [8], Multi-scale top-hat [11], Liu's method [12], and IPACHI method [2] were used. In a deep learning based model, U-net [4] was also used.…”

“…In [1], the optimized shape model was used to extract features and boundary information for distinguishing tooth region from background regions. In [2], infinite perimeter active contour model was used to perform automated retinal vessel segmentation. Authors of [3] proposed multi-scale classification-based lesion segmentation method for dermoscopic images.…”