Document Type : Original Paper
Medical physics department, Faculty of medicine, Iran university of medical sciences, Tehran, Iran
Medical Physics Department, School of Medicine, Iran university of Medical Sciences,
Introduction: Image classification is a controversial part of image processing especially for tissue classification of the brain in MR images because brain tissue signals and contrasts are close together. More accurate classification of brain MR images is vital for diagnosis in cerebrospinal fluid, gray matter, and white matter.
Materials and Methods: In this study, 20 brain MR Images were classified by DIPY, SPM, and FSL with a Bayesian framework. All the classified images were surveyed in terms of similarity by using two quantity DICE and Jaccard Coefficients.
Results: The SPM classification has shown a suitable classification of cerebrospinal fluid compared to DIPY and FSL. The DICE and JACCARD coefficients of SPM classification were 97.48 ± 0.28 and 92.68 ± 0.94, respectively. The DICE and Jaccard coefficients for white matter were 95.64 ± 0.23 and 86.18 ± 1.64, respectively, and for gray matter were 93.66 ± 0.76 and 83.62 ± 1.92, respectively for DIPY which showed this classification is better than other software packages.
Conclusion: The obtained results showed an appropriate classification of GM and WM regions by the DIPY python library.