Open AccessSegmentation of osteosarcoma in MRI images by K‐means clustering, Chan‐Vese segmentation, and iterative Gaussian filtering
Author(s) -
Nasor Mohamed,
Obaid Walid
Publication year - 2021
Publication title -
iet image processing
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.401
H-Index - 45
eISSN - 1751-9667
pISSN - 1751-9659
DOI - 10.1049/ipr2.12106
Subject(s) - segmentation , artificial intelligence , computer science , sørensen–dice coefficient , pattern recognition (psychology) , image segmentation , magnetic resonance imaging , cluster analysis , dice , precision and recall , blob detection , osteosarcoma , computer vision , edge detection , mathematics , image processing , image (mathematics) , radiology , medicine , statistics , pathology
Unlike other types of tumours, automated osteosarcoma segmentation in magnetic resonance images (MRI) is a challenging task due to its different and unique intensity and texture. This paper presents a technique for segmenting osteosarcoma in MRI images using a combination of image processing techniques which include K‐means clustering, Chan‐Vese segmentation, iterative Gaussian filtering, and Canny edge detection. In addition, the proposed technique involves iterative morphological operations and object counting. The technique was tested using 50 MRI scan images that contain osteosarcoma tumours. The proposed technique was able to segment the osteosarcoma regardless of the variations in their intensities, textures and locations. The performance of the technique was measured by calculating the values for precision, recall, specificity, Dice score coefficient, accuracy and the running time (RT) for all tested cases. The proposed technique achieved 95.96% precision, 86.15% recall, 99.51% specificity, 89.84% Dice score coefficient, 98.02% accuracy, and 191.62 s average running time. This technique can assist clinicians in making treatment plans for patients with osteosarcoma.