Boundary delineation for hepatic hemangioma in ultrasound images

Abstract

Hemangioma is one of the most common benign congenital complications of the human body which can arise in interior organs and external limbs. The main aim of this work is to present a new method for automatic detection of liver hemangioma and its boundaries in ultrasound images, using image processing techniques. Overall there are two phases, the preprocessing procedure and the boundary delineation phase. The preprocessing phase includes three main stages: 1. Image contrast enhancement using Difference of Offset Gaussian (DoOG) method, 2. Applying Canny edge filtering, 3. Applying an adaptive threshold in order to detect the ROI (hemangioma). Following, the snake algorithm is used to segment the hemangioma region in the second phase. For the quantitative assessment of the proposed method for the segmentation stage, the results derived via the proposed algorithms have been compared with the corresponding segmented regions determined by an expert using three similarity criteria. The results showed 73 percent similarity without pre-processing and 90 percent similarity with pre-processing.

Algorithm guided outlining of 105 pancreatic cancer liver metastases in Ultrasound

“…Most of these publications discuss the difficulties of segmentation performed using ultrasound images and propose new ways to overcome difficulties like speckle noise existence 33 . One way is to enhance the contrast of the lesion in comparison to the surrounding tissue by using additional intravenous contrast dye [34][35][36] . This time-consuming method truly helps to delineate focal liver lesion, but harbors the risk of allergic reaction to the contrast dye and significantly increases the costs and length of the examination.…”

In-depth assessment of an interactive graph-based approach for the segmentation for pancreatic metastasis in ultrasound acquisitions of the liver with two specialists in Internal Medicine

“…The 2017 Biomedical Engineering International Conference (BMEiCON-2017) Others working in the segmentation [16]- [21] area of US images, are for example Hao et al [22], Bahrami et al [23], Bakas et al [24], Gatos et al [25], Jain and Kumar [26], Quan et al [27] and Ciurte et al [28]. Hao et al present a region growing that is performed in a multi-feature vector space.…”