A Method for Fetal Head Detection In Ultrasound Image

  • Mr Shengli Li, Affiliated Shenzhen Maternity and Child Healthcare Hospital of Nanfang Medical University, China
  • Mr Kai Chen, Affiliated Shenzhen Maternity and Child Healthcare Hospital of Nanfang Medical University, China
  • Ms Pin Tang, Affiliated Shenzhen Maternity and Child Healthcare Hospital of Nanfang Medical University, China
  • Ms Chenhong Wang, Affiliated Shenzhen Maternity and Child Healthcare Hospital of Nanfang Medical University, China
  • Ms Yuan Yao, Affiliated Shenzhen Maternity and Child Healthcare Hospital of Nanfang Medical University, China
  • Mr Huaxuan Wen, Affiliated Shenzhen Maternity and Child Healthcare Hospital of Nanfang Medical University, China
  • Ms Xiulan Chen, Affiliated Shenzhen Maternity and Child Healthcare Hospital of Nanfang Medical University, China

    • A method for detecting fetal skull in ultrasound image was proposed. The main process was as follows. The K-mean algorithm was used to cluster pixels into 3 classes and the bright objects were identified; only bigger connected components in the bright objects were preserved and skeletonized; randomized Hough transform(RHT) was then applied for ellipse detection. The preprocessing method adopted greatly reduced the number of pixels to be examined by RHT and increased detection speed. Furthermore, a new scoring mechanism for RHT was introduced, which takes the head curve's performance in both image and parameter spaces into account and increases detection accuracy.

    Keywords: ultrasound image, fetal skull, ellipse detection, Hough transform