Home About Login Current Archives Announcements Editorial Board
Submit Now For Authors Call for Submissions Statistics Contact
Home > Archives > Volume 20, No 7 (2022) > Article

DOI: 10.14704/nq.2022.20.7.NQ33475

Certain Investigation on Brain tumour Segmentation using Discrete orthogonal Moments approach on U-Net

K. Manasa, Dr.V.Krishnaveni

Abstract

Brain tumour is fatal disease, early brain tumor diagnosis improves patient survival ability. Manual Brain tumour segmentation is difficult, expensive and time consuming, so automated brain tumour segmentation garnered more attention. Image moments are best feature descriptor and most extensively used in image processing domain. In this paper discrete orthogonal moments like Tchebichef, Krawtchouck and Dual Hahn image moments and the separable moments Tchebichef-Krawtchouk (TKM), Tcheichef-Dual Hahn (TDHM) and Krawtchouk-Dual Hahn (KDHM) moments are used for feature extraction from MRI brain tumour dataset. Recently Deep learning algorithms outperformed existing models, in that U-Net gained considerable attention in medical image analysis. The extracted moments are given as initial trainable kernel values during convolution in U-Net. The proposed segmentation model uses Brats 2018 Dataset achieving dice score of 0.88, 0.8 and 0.83 for enhanced tumor core, whole tumor and tumor core respectively.

Keywords

Brain tumour diagnosis, image moments, U-Net, MRI segmentation, Tchebichef moments, Krawtchouck Moments, Dual Hahn Moments

Full Text

PDF

References

?>