Limits...
An active contour model for the segmentation of images with intensity inhomogeneities and bias field estimation.

Huang C, Zeng L - PLoS ONE (2015)

Bottom Line: The proposed model first appeared as a two-phase model and then extended to a multi-phase one.The experimental results demonstrate the advantages of our model in terms of accuracy and insensitivity to the location of the initial contours.In particular, our method has been applied to various synthetic and real images with desirable results.

View Article: PubMed Central - PubMed

Affiliation: Key Laboratory of Optoelectronic Technology and System of the Education Ministry of China, Chongqing University, Chongqing, 400044, China; Engineering Research Center of Industrial Computed Tomography Nondestructive Testing of the Education Ministry of China, Chongqing University, Chongqing, 400044, China.

ABSTRACT
Intensity inhomogeneity causes many difficulties in image segmentation and the understanding of magnetic resonance (MR) images. Bias correction is an important method for addressing the intensity inhomogeneity of MR images before quantitative analysis. In this paper, a modified model is developed for segmenting images with intensity inhomogeneity and estimating the bias field simultaneously. In the modified model, a clustering criterion energy function is defined by considering the difference between the measured image and estimated image in local region. By using this difference in local region, the modified method can obtain accurate segmentation results and an accurate estimation of the bias field. The energy function is incorporated into a level set formulation with a level set regularization term, and the energy minimization is conducted by a level set evolution process. The proposed model first appeared as a two-phase model and then extended to a multi-phase one. The experimental results demonstrate the advantages of our model in terms of accuracy and insensitivity to the location of the initial contours. In particular, our method has been applied to various synthetic and real images with desirable results.

No MeSH data available.


Experimental results of our model.(a), (e) The original image with red initial contours. (b), (f) The final segmentation results of our model. (c), (g) The gray images of ∣d∣. (d), (h) The estimation of the bias field in our model.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4383562&req=5

pone.0120399.g004: Experimental results of our model.(a), (e) The original image with red initial contours. (b), (f) The final segmentation results of our model. (c), (g) The gray images of ∣d∣. (d), (h) The estimation of the bias field in our model.

Mentions: The next experiment considers the segmentation of the same image in Fig. 1 (as shown in Fig. 4). The T-shaped image is a real image with intensity inhomogeneity, which size is 127×96. The initial active contours are set inside the object domain and contain the background. Our method outperforms Li’s model (the code was downloaded from [44]) in some cases. As the local regional difference is considered, incorrect estimations of the true image J in the local region can be corrected in each iteration, which is insensitive to the initial contours in our experiments. Fig. 4 indicates that even the initial contours located inside the objects contain background, the segmentation results and bias field estimate are nearly the same. We choose the absolute value of the local difference d, which is shown in gray images (Fig. 4(c) and Fig. 4(g)) to describe the level between the measured image I and estimated bJ. In Figs. 4(c) and 4(g), ∣d∣ is often large in highlight regions or regions with similar intensities; in these regions, the difference between I and bJ must be corrected to obtain better estimations of b and J. By the quantitative comparison using the above metrics in the second row of Fig. 2 and Fig. 4, the values of JS and DSC in our method are bigger than Li’s method, the value of RFP are equal show that the region Sb/O of Li’s method and our method are the same, while the value of RFN in our model (0.0068) are smaller than Li’s method (0.3186) mean the region Sm/O of our method achieves more accurate segmentation results (see Table 1).


An active contour model for the segmentation of images with intensity inhomogeneities and bias field estimation.

Huang C, Zeng L - PLoS ONE (2015)

Experimental results of our model.(a), (e) The original image with red initial contours. (b), (f) The final segmentation results of our model. (c), (g) The gray images of ∣d∣. (d), (h) The estimation of the bias field in our model.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4383562&req=5

pone.0120399.g004: Experimental results of our model.(a), (e) The original image with red initial contours. (b), (f) The final segmentation results of our model. (c), (g) The gray images of ∣d∣. (d), (h) The estimation of the bias field in our model.
Mentions: The next experiment considers the segmentation of the same image in Fig. 1 (as shown in Fig. 4). The T-shaped image is a real image with intensity inhomogeneity, which size is 127×96. The initial active contours are set inside the object domain and contain the background. Our method outperforms Li’s model (the code was downloaded from [44]) in some cases. As the local regional difference is considered, incorrect estimations of the true image J in the local region can be corrected in each iteration, which is insensitive to the initial contours in our experiments. Fig. 4 indicates that even the initial contours located inside the objects contain background, the segmentation results and bias field estimate are nearly the same. We choose the absolute value of the local difference d, which is shown in gray images (Fig. 4(c) and Fig. 4(g)) to describe the level between the measured image I and estimated bJ. In Figs. 4(c) and 4(g), ∣d∣ is often large in highlight regions or regions with similar intensities; in these regions, the difference between I and bJ must be corrected to obtain better estimations of b and J. By the quantitative comparison using the above metrics in the second row of Fig. 2 and Fig. 4, the values of JS and DSC in our method are bigger than Li’s method, the value of RFP are equal show that the region Sb/O of Li’s method and our method are the same, while the value of RFN in our model (0.0068) are smaller than Li’s method (0.3186) mean the region Sm/O of our method achieves more accurate segmentation results (see Table 1).

Bottom Line: The proposed model first appeared as a two-phase model and then extended to a multi-phase one.The experimental results demonstrate the advantages of our model in terms of accuracy and insensitivity to the location of the initial contours.In particular, our method has been applied to various synthetic and real images with desirable results.

View Article: PubMed Central - PubMed

Affiliation: Key Laboratory of Optoelectronic Technology and System of the Education Ministry of China, Chongqing University, Chongqing, 400044, China; Engineering Research Center of Industrial Computed Tomography Nondestructive Testing of the Education Ministry of China, Chongqing University, Chongqing, 400044, China.

ABSTRACT
Intensity inhomogeneity causes many difficulties in image segmentation and the understanding of magnetic resonance (MR) images. Bias correction is an important method for addressing the intensity inhomogeneity of MR images before quantitative analysis. In this paper, a modified model is developed for segmenting images with intensity inhomogeneity and estimating the bias field simultaneously. In the modified model, a clustering criterion energy function is defined by considering the difference between the measured image and estimated image in local region. By using this difference in local region, the modified method can obtain accurate segmentation results and an accurate estimation of the bias field. The energy function is incorporated into a level set formulation with a level set regularization term, and the energy minimization is conducted by a level set evolution process. The proposed model first appeared as a two-phase model and then extended to a multi-phase one. The experimental results demonstrate the advantages of our model in terms of accuracy and insensitivity to the location of the initial contours. In particular, our method has been applied to various synthetic and real images with desirable results.

No MeSH data available.