A Split-and-Merge-Based Uterine Fibroid Ultrasound Image Segmentation Method in HIFU Therapy

Abstract

High-intensity focused ultrasound (HIFU) therapy has been used to treat uterine fibroids widely and successfully. Uterine fibroid segmentation plays an important role in positioning the target region for HIFU therapy. Presently, it is completed by physicians manually, reducing the efficiency of therapy. Thus, computer-aided segmentation of uterine fibroids benefits the improvement of therapy efficiency. Recently, most computer-aided ultrasound segmentation methods have been based on the framework of contour evolution, such as snakes and level sets. These methods can achieve good performance, although they need an initial contour that influences segmentation results. It is difficult to obtain the initial contour automatically; thus, the initial contour is always obtained manually in many segmentation methods. A split-and-merge-based uterine fibroid segmentation method, which needs no initial contour to ensure less manual intervention, is proposed in this paper. The method first splits the image into many small homogeneous regions called superpixels. A new feature representation method based on texture histogram is employed to characterize each superpixel. Next, the superpixels are merged according to their similarities, which are measured by integrating their Quadratic-Chi texture histogram distances with their space adjacency. Multi-way Ncut is used as the merging criterion, and an adaptive scheme is incorporated to decrease manual intervention further. The method is implemented using Matlab on a personal computer (PC) platform with Intel Pentium Dual-Core CPU E5700. The method is validated on forty-two ultrasound images acquired from HIFU therapy. The average running time is 9.54 s. Statistical results showed that SI reaches a value as high as 87.58%, and normHD is 5.18% on average. It has been demonstrated that the proposed method is appropriate for segmentation of uterine fibroids in HIFU pre-treatment imaging and planning.

Fig 1. Flowchart of the proposed method.

Fig 2. Illustration of the geometry and configuration of the transducers in the HIFU machine.

Fig 3. Illustration of the original image with an ROI.

T: Tumor.

Fig 4. Splitting results.

Red lines represent the tumor boundary depicted by the radiologist. (a) The original image. (b) The image after preprocess. (c) Superpixels.

Fig 5. Diagram of superpixel feature extraction.

Fig 6. Examples of textons.

(a) The image. (b) 20 corresponding textons.

Fig 7. An example of RAM.

Left: The image. Right: corresponding RAM.

Fig 8. Results with different F values.

(a) F = 2. (b) F = 3. (c) F = 4. (d) F = 5. (e) F = 6. (f) F = 7. (g) F = 8. When F = 8, the tumor is segmented. The tumor boundary is depicted with a red color.

Fig 9. Boxplots of statistical results.

Left: SI. Middle: HD. Right: NormHD.

Fig 10. An example of a heterogeneous tumor.

(a) The original image. (b) Standard boundary. (c) Boundary generated by the proposed method. (d) The merged result of the construction of W using the Quadratic-Chi histogram distance. (e) The merged result of the construction of W using the χ ² distance. (f) Visualization of W using the Quadratic-Chi histogram distance. (g) Visualization of W using the χ ² distance. The white arrow points to superpixels inside the tumor.

Fig 11. An example of a tumor with weak boundaries.

(a) The original image. (b) Superpixels. Superpixels A and B are inside the tumor, while superpixel C is outside. (c) The boundary generated by the proposed method. (d) Standard boundary. (e) Histograms for three superpixels A, B, and C.

Fig 12. Results for synthetic images.

(a) Amplitudes of scatters inside tumors have the same distribution. (b) Amplitudes of scatters inside tumors have different distributions. (c) Tumors whose amplitudes of scatters have different distributions have complex shapes. (d) Segmentation of (a). (e) Segmentation of (b). (f) Segmentation of (c).

Fig 13. An example of tumor segmentation with different F values.

(a) The original image. (b) Standard boundary. (c) Segmentation result when F = 4. (d) Segmentation result when F = 9.

Fig 14. An example of tumor segmentation with a weak boundary in different ROIs.

(a) The original image. (b) The first ROI image. (c) Segmentation result of (b). (d) The second ROI image. (e) Segmentation result of (d).

Fig 15. An example of tumor segmentation with hyper-echoic appearance in different ROIs.

(a) The original image. (b) The first ROI image. (c) Segmentation result of (b). (d) The second ROI image. (e) Segmentation result of (d).

Fig 16. An example of tumor segmentation with great heterogeneous structure in different ROIs.

(a) The original image. (b) The first ROI image. (c) Segmentation result of (b). (d) The second ROI image. (e) Segmentation result of (d).