3-D segmentation of articular cartilages by graph cuts using knee MR images from the osteoarthritis initiative

Hackjoon Shim, Soochan Lee, Bohyeong Kim, Cheng Tao, Samuel Chang, Il Dong Yun, Sang Uk Lee, Kent Kwoh, Kyongtae Bae

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Scopus citations


Knee osteoarthritis is the most common debilitating health condition affecting elderly population. MR imaging of the knee is highly sensitive for diagnosis and evaluation of the extent of knee osteoarthritis. Quantitative analysis of the progression of osteoarthritis is commonly based on segmentation and measurement of articular cartilage from knee MR images. Segmentation of the knee articular cartilage, however, is extremely laborious and technically demanding, because the cartilage is of complex geometry and thin and small in size. To improve precision and efficiency of the segmentation of the cartilage, we have applied a semi-automated segmentation method that is based on an s/t graph cut algorithm. The cost function was defined integrating regional and boundary cues. While regional cues can encode any intensity distributions of two regions, "object" (cartilage) and "background" (the rest), boundary cues are based on the intensity differences between neighboring pixels. For three-dimensional (3-D) segmentation, hard constraints are also specified in 3-D way facilitating user interaction. When our proposed semi-automated method was tested on clinical patients' MR images (160 slices, 0.7 mm slice thickness), a considerable amount of segmentation time was saved with improved efficiency, compared to a manual segmentation approach.

Original languageEnglish (US)
Title of host publicationMedical Imaging 2008
Subtitle of host publicationImage Processing
StatePublished - 2008
EventMedical Imaging 2008: Image Processing - San Diego, CA, United States
Duration: Feb 17 2008Feb 19 2008

Publication series

NameProgress in Biomedical Optics and Imaging - Proceedings of SPIE
ISSN (Print)1605-7422


OtherMedical Imaging 2008: Image Processing
Country/TerritoryUnited States
CitySan Diego, CA


  • Graph cuts
  • Knee cartilage
  • MR images
  • Segmentation

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Atomic and Molecular Physics, and Optics
  • Biomaterials
  • Radiology Nuclear Medicine and imaging


Dive into the research topics of '3-D segmentation of articular cartilages by graph cuts using knee MR images from the osteoarthritis initiative'. Together they form a unique fingerprint.

Cite this