BROWSE
Detail
Conference Paper
Vessel Segmentation Model using Automated Threshold Algorithm from Lower Leg MRI
Cited 6 time in 
Share 
Share
- Authors
- Ji Young Lee, Jin Yeong Mun, Mohammad Taheri, Seong Ho Son, Sung Shin
- Issue Date
- 2017-09
- Citation
- Research in Adaptive and Convergent Systems (RACS) 2017, pp.120-125
- Language
- English
- Type
- Conference Paper
- Abstract
- Blood vessel segmentation has been developed in the liver, heart, and retinal images due to accurate description and analysis of vascular structure plays a crucial role in clinical routine. Since the varicose vein, deep vein thrombosis, and occlusive arterial diseases are related to vascular structure in the lower leg, blood vessel segmentation in lower limbs is also clinically important. In this paper, we proposed a feature-based adaptive threshold model for automatically extracting vessel in the lower leg Magnetic Resonance Images (MRIs). The proposed model is divided into 2 stages. The first stage, pre-processing, included partial volume reduction, contrast equalization, and removing background noises. The second stage is segmentation stage. Fuzzy C-mean clustering, Hough transform in feature extraction technique, and threshold algorithm were included in the second stage. Automatic threshold value determination algorithm is enhanced by using the Hough transform in feature extraction technique. The proposed model has been implemented for showing accuracy (ACC) compared with a manually generated ground truth from domain experts. Results show that proposed model has the accuracy with the average 98.43%, which is higher than existing model, Adaptive Vein Segmentation (AVS) method as a reference [1].
- KSP Keywords
- Adaptive threshold model, Automatic threshold, Deep Vein Thrombosis(DVT), Extraction technique, Feature extractioN, Feature-based, First stage, Fuzzy C-mean Clustering, Ground Truth, Hough transform, Magnetic resonance(MR)