Hybrid Level Set Model for Parathyroid Gland Segmentation Based on Local Entropy of Images

Lin Mao; Liqiang Zhao; Ming’an Yu; Ying Wei; Ying Wang

doi:10.3788/AOS201939.1217001

Acta Optica Sinica, Volume. 39, Issue 12, 1217001(2019)

Hybrid Level Set Model for Parathyroid Gland Segmentation Based on Local Entropy of Images

Lin Mao¹, Liqiang Zhao¹, Ming’an Yu², Ying Wei², and Ying Wang^3,4、*

Author Affiliations

¹College of Information Science and Technology, Beijing University of Chemical Technology, Beijing 100029, China

²Interventional Ultrasound Medicine, China-Japan Friendship Hospital, Beijing 100029, China

³Institute of Microelectronics, Chinese Academy of Sciences, Beijing 100029, China

⁴University of Chinese Academy of Sciences, Beijing 100049, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(7)

Fig. 1. Intensity homogeneous image, parathyroid ultrasound image, and their local entropy images. (a) Intensity homogeneous image; (b) local entropy of intensity homogeneous image; (c) parathyroid ultrasound image; (d) local entropy of parathyroid ultrasound image

Download full size

Fig. 2. Evolution curves for iterations 20 with different global weights. (a) Original contour; (b) α=0.1; (c) α=0.5; (d) α=0.9; (e) α_i_,_j=exp[-h(i,j)]

Download full size

Fig. 3. Parathyroid ultrasound images with different intensity. (a)(b) Parathyroid lesion with clear edge; (c) parathyroid lesion with weak edge; (d)(e) parathyroid lesion with fuzzy edge; (f) parathyroid lesion connected to the other tissue

Download full size

Fig. 4. Segmentation results of different types of parathyroid images by three models. (a) Parathyroid ROI with original contour; (b) segmentation results of the model in Ref. [6]; (c) segmentation results of the model in Ref. [12]; (d) segmentation results of the proposed model

Download full size

Fig. 5. Segmentation results of the proposed model at different scales. (a) Small scale; (b) large scale; (c) two-scales

Download full size

Table 1. Parathyroid segmentation results with different models

View table

Table 1. Parathyroid segmentation results with different models

Test image	Model in Ref. [6]			Model in Ref. [12]			Proposed model
Test image	DSC	Iteration	Running time /s	DSC	Iteration	Running time /s	DSC	Iteration	Running time /s
a	0.907	846	413.94	0.851	477	211.54	0.949	258	127.74
b	0.724	1000	727.54	0.970	1000	676.31	0.977	996	633.97
c	0.931	1000	726.11	0.942	1000	670.08	0.957	1000	706.18
d	0.809	670	452.68	0.911	704	434.31	0.923	263	201.14
e	0.725	1000	467.76	0.848	1000	426.52	0.856	600	287.46
f	0.809	1000	343.30	0.778	1000	339.06	0.932	380	175.68

Table 2. Segmentation results of the proposed model at different scales
View table
Table 2. Segmentation results of the proposed model at different scales
Parameter Small scale Large scale Two scales
Iteration 380 380 380
DSC 0.921 0.904 0.932
Running time /s 144.22 165.45 140.27

Tools

Get Citation

Copy Citation Text

Lin Mao, Liqiang Zhao, Ming’an Yu, Ying Wei, Ying Wang. Hybrid Level Set Model for Parathyroid Gland Segmentation Based on Local Entropy of Images[J]. Acta Optica Sinica, 2019, 39(12): 1217001

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites