Levels Propagation Approach to Image Segmentation: Application to Breast MR Images

Abstract

Accurate segmentation of a breast tumor region is fundamental for treatment. Magnetic resonance imaging (MRI) is a widely used diagnostic tool. In this paper, a new semi-automatic segmentation approach for MRI breast tumor segmentation called Levels Propagation Approach (LPA) is introduced. The introduced segmentation approach takes inspiration from tumor propagation and relies on a finite set of nested and non-overlapped levels. LPA has several features: it is highly suitable to parallelization and offers a simple and dynamic possibility to automate the threshold selection. Furthermore, it allows stopping of the segmentation at any desired limit. Particularly, it allows to avoid to reach the breast skin-line region which is known as a significant issue that reduces the precision and the effectiveness of the breast tumor segmentation. The proposed approach have been tested on two clinical datasets, namely RIDER breast tumor dataset and CMH-LIMED breast tumor dataset. The experimental evaluations have shown that LPA has produced competitive results to some state-of-the-art methods and has acceptable computation complexity.

Keywords: Breast; Image segmentation; Levels; MRI; Propagation; Tumor.

Fig. 1

Tumor growing steps: initiation, promotion and progression [33]

Fig. 2

A representation of four nested levels that surround the origin pixel

Fig. 3

Flowchart of the proposed LPA

Fig. 4

Example showing the impact of the value of the parameter α on the final extraction of a non-enhanced tumor

Fig. 5

Variation of the threshold values through the levels for the image of Fig. 4a according to α = 0.8 and α = 0.95

Fig. 6

Example showing the impact of the value of the parameter α on the final extraction of an enhanced tumor

Fig. 7

Variation of the threshold values through the levels for the image of Fig. 6a according to α = 0.8 and α = 0.95

Fig. 8

An example of an image from the RIDER breast tumor MRI dataset in which the intensity of the skin-line region is closely similar to the intensity of the tumor region

Fig. 9

Example of segmentation results of LPA on RIDER dataset. With (b) and (c) (resp. (e) and (f)) are the ground truth and the extracted tumor of (a) (resp. (d))

Fig. 10

Example illustrating the segmentation results of a non-enhanced tumor using MASRG, Level set, and LPA. With a is the original MRI image, b is the ground truth, c is the segmented image by MASRG, d is the segmented image using Level set, and e is the segmented image by LPA

Fig. 11

Example showing the segmentation results of an enhanced tumor using MASRG, Level set, and LPA. With a is the original MRI image, b is the ground truth, c is the segmented image by MASRG, d is the segmented image using Level set, and e is the segmented image by LPA