Limits...
New region-scalable discriminant and fitting energy functional for driving geometric active contours in medical image segmentation.

Wang X, Niu Y, Tan L, Zhang SX - Comput Math Methods Med (2014)

Bottom Line: The region-scalable discriminant and fitting energy functional is defined to capture the image intensity characteristics in local and global regions for driving the evolution of active contour.The discriminant term in the model aims at separating background and foreground in scalable regions while the fitting term tends to fit the intensity in these regions.The new model utilizes intensity information in the local and global regions as much as possible; so it not only handles better intensity inhomogeneity, but also allows more robustness to noise and more flexible initialization in comparison to the original global region and regional-scalable based models.

View Article: PubMed Central - PubMed

Affiliation: Key Laboratory of Optoelectronic Technology and Systems of Ministry of Education, College of Optoelectronic Engineering, Chongqing University, Chongqing 400044, China.

ABSTRACT
We propose a novel region-based geometric active contour model that uses region-scalable discriminant and fitting energy functional for handling the intensity inhomogeneity and weak boundary problems in medical image segmentation. The region-scalable discriminant and fitting energy functional is defined to capture the image intensity characteristics in local and global regions for driving the evolution of active contour. The discriminant term in the model aims at separating background and foreground in scalable regions while the fitting term tends to fit the intensity in these regions. This model is then transformed into a variational level set formulation with a level set regularization term for accurate computation. The new model utilizes intensity information in the local and global regions as much as possible; so it not only handles better intensity inhomogeneity, but also allows more robustness to noise and more flexible initialization in comparison to the original global region and regional-scalable based models. Experimental results for synthetic and real medical image segmentation show the advantages of the proposed method in terms of accuracy and robustness.

Show MeSH

Related in: MedlinePlus

Segmentation results on an X-ray vessel image by CV (a), RSF (b), LGIF (c), and our method (d) with different initial contours. The initial and the final zero level sets are plotted as the green and the red contours, respectively (color online).
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4109108&req=5

fig3: Segmentation results on an X-ray vessel image by CV (a), RSF (b), LGIF (c), and our method (d) with different initial contours. The initial and the final zero level sets are plotted as the green and the red contours, respectively (color online).

Mentions: In this subsection, we employ our method to segment typical medical images with different modalities and compare it to related methods. Figure 3 shows the results of an X-ray image of blood vessels. The parameters are settled as w = 0.95~0.99; k = 0.1~0.3; ν = 0.001∗2552 ~ 0.005∗2552; Δt = 0.1; and μ = 0.2/Δt. As illustrated in these figures, with different initial contours, the CV model fails to segment the objects, especially in the weak boundary of the vessel. The RSF model is not only sensitive to the initial contours, but also less capable of excluding the false objects. As shown in the figures in the first and second columns, even when the initial contour is settled on the right vessel object, it still captures the wrong contours in the background as pseudo object. The LGIF method performs similar to the RSF method because the local intensity fitting-based energy functional plays more important role than the CV model in segmenting this image. The vessel and additional small background noise are extracted as objects. Additional postprocessing should be used for removing them. This fact limits its availability in practice. In contrast, our method is less sensitive to the initial contour and it captures the vessel accurately in all the cases. The robustness of our method is furtherly verified in the experiment on segmenting a vessel image with heavy intensity inhomogeneity. As shown in Figure 4, the segmentation results of the RSF model are different along with the different initial contours. Remarkable missegmentation appears in the three cases. The LGIF model extracts the vessel in all these cases, but with the false vessel segmented from background, which means that the simple combination of local and global intensity fitting is hard to segment the object in noisy background. In contrast, the proposed model not only gives accurate results but also shows robustness to different initial contours. For example, in the third row of Figure 3 and the first row of Figure 4, even though the initial contours do not contain any foreground objects, our method can still obtain precise segmentation results. These two experiments illustrate the advantages of our proposed model in handling real image: the ability to handle weak boundaries and complex background and robustness to noise.


New region-scalable discriminant and fitting energy functional for driving geometric active contours in medical image segmentation.

Wang X, Niu Y, Tan L, Zhang SX - Comput Math Methods Med (2014)

Segmentation results on an X-ray vessel image by CV (a), RSF (b), LGIF (c), and our method (d) with different initial contours. The initial and the final zero level sets are plotted as the green and the red contours, respectively (color online).
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4109108&req=5

fig3: Segmentation results on an X-ray vessel image by CV (a), RSF (b), LGIF (c), and our method (d) with different initial contours. The initial and the final zero level sets are plotted as the green and the red contours, respectively (color online).
Mentions: In this subsection, we employ our method to segment typical medical images with different modalities and compare it to related methods. Figure 3 shows the results of an X-ray image of blood vessels. The parameters are settled as w = 0.95~0.99; k = 0.1~0.3; ν = 0.001∗2552 ~ 0.005∗2552; Δt = 0.1; and μ = 0.2/Δt. As illustrated in these figures, with different initial contours, the CV model fails to segment the objects, especially in the weak boundary of the vessel. The RSF model is not only sensitive to the initial contours, but also less capable of excluding the false objects. As shown in the figures in the first and second columns, even when the initial contour is settled on the right vessel object, it still captures the wrong contours in the background as pseudo object. The LGIF method performs similar to the RSF method because the local intensity fitting-based energy functional plays more important role than the CV model in segmenting this image. The vessel and additional small background noise are extracted as objects. Additional postprocessing should be used for removing them. This fact limits its availability in practice. In contrast, our method is less sensitive to the initial contour and it captures the vessel accurately in all the cases. The robustness of our method is furtherly verified in the experiment on segmenting a vessel image with heavy intensity inhomogeneity. As shown in Figure 4, the segmentation results of the RSF model are different along with the different initial contours. Remarkable missegmentation appears in the three cases. The LGIF model extracts the vessel in all these cases, but with the false vessel segmented from background, which means that the simple combination of local and global intensity fitting is hard to segment the object in noisy background. In contrast, the proposed model not only gives accurate results but also shows robustness to different initial contours. For example, in the third row of Figure 3 and the first row of Figure 4, even though the initial contours do not contain any foreground objects, our method can still obtain precise segmentation results. These two experiments illustrate the advantages of our proposed model in handling real image: the ability to handle weak boundaries and complex background and robustness to noise.

Bottom Line: The region-scalable discriminant and fitting energy functional is defined to capture the image intensity characteristics in local and global regions for driving the evolution of active contour.The discriminant term in the model aims at separating background and foreground in scalable regions while the fitting term tends to fit the intensity in these regions.The new model utilizes intensity information in the local and global regions as much as possible; so it not only handles better intensity inhomogeneity, but also allows more robustness to noise and more flexible initialization in comparison to the original global region and regional-scalable based models.

View Article: PubMed Central - PubMed

Affiliation: Key Laboratory of Optoelectronic Technology and Systems of Ministry of Education, College of Optoelectronic Engineering, Chongqing University, Chongqing 400044, China.

ABSTRACT
We propose a novel region-based geometric active contour model that uses region-scalable discriminant and fitting energy functional for handling the intensity inhomogeneity and weak boundary problems in medical image segmentation. The region-scalable discriminant and fitting energy functional is defined to capture the image intensity characteristics in local and global regions for driving the evolution of active contour. The discriminant term in the model aims at separating background and foreground in scalable regions while the fitting term tends to fit the intensity in these regions. This model is then transformed into a variational level set formulation with a level set regularization term for accurate computation. The new model utilizes intensity information in the local and global regions as much as possible; so it not only handles better intensity inhomogeneity, but also allows more robustness to noise and more flexible initialization in comparison to the original global region and regional-scalable based models. Experimental results for synthetic and real medical image segmentation show the advantages of the proposed method in terms of accuracy and robustness.

Show MeSH
Related in: MedlinePlus