Fully automated tumor segmentation based on improved fuzzy connectedness algorithm in brain MR images

Comput Biol Med. 2011 Jul;41(7):483-92. doi: 10.1016/j.compbiomed.2011.04.010. Epub 2011 May 23.

Abstract

Uncontrollable and unlimited cell growth leads to tumor genesis in the brain. If brain tumors are not diagnosed early and cured properly, they could cause permanent brain damage or even death to patients. As in all methods of treatments, any information about tumor position and size is important for successful treatment; hence, finding an accurate and a fully automated method to give information to physicians is necessary. A fully automatic and accurate method for tumor region detection and segmentation in brain magnetic resonance (MR) images is suggested. The presented approach is an improved fuzzy connectedness (FC) algorithm based on a scale in which the seed point is selected automatically. This algorithm is independent of the tumor type in terms of its pixels intensity. Tumor segmentation evaluation results based on similarity criteria (similarity index (SI), overlap fraction (OF), and extra fraction (EF) are 92.89%, 91.75%, and 3.95%, respectively) indicate a higher performance of the proposed approach compared to the conventional methods, especially in MR images, in tumor regions with low contrast. Thus, the suggested method is useful for increasing the ability of automatic estimation of tumor size and position in brain tissues, which provides more accurate investigation of the required surgery, chemotherapy, and radiotherapy procedures.

MeSH terms

  • Algorithms*
  • Brain / anatomy & histology
  • Brain / pathology
  • Brain Neoplasms / pathology*
  • Databases, Factual
  • Fuzzy Logic*
  • Humans
  • Image Processing, Computer-Assisted / methods*
  • Magnetic Resonance Imaging*