- Title: Automatic Segmentation of Menisci in MR Images Using Pattern Recognition and Graph Cuts
- Description: In this Masters thesis a system for automatic segmentation of the menisci in magnetic resonance (MR) images is developed and tested. MR imaging is a non invasive method with no ionization and with good soft tissue contrast and constitutes an important tool to evaluate soft tissue, tendon, ligaments and menisci in the knee. The menisci are located between the femur (thigh bone) and patella (shin bone) in the knee and has several functions, such as shock absorption. Medical imaging systems have been improved during the last decades. Images today are produced with higher resolution and faster acquisition time than before. Processing these images manually is very time consuming. Therefore it is necessary to create computerized tools that can be used by the medical staff in their work with diagnosis of illness and treatments of patients. The approach for constructing the automatically segment system is divided into three subparts. The first task is to extract all bone parts from the images. Secondly, the area where the menisci can be found is localized and the final step is to segment the menisci. This thesis is mainly based on two methods and they are pattern recognition with Laws texture features and Graph Cut. In this thesis the system is tested on five examinations produced in Landskrona hospital from a Philips 1:5T machine using a spin-echo sequence. The result is promising and could with further work become an accurate system for automatically segmenting the menisci.
- Start Date: Aug. 1, 2010
- Finished Date: April 15, 2011
- Supervisor: Karl Åström
- Supervisor: Petter Strandmark
- Student: Fredrik Nedmark, (F-07)
Questions: webmaster
Last update: 2013-04-11
Centre for Mathematical Sciences, Box 118, SE-22100, Lund. Telefon: +46 46-222 00 00 (vx)