Graph Cut Energy Minimization in a Probabilistic Learning Framework for 3D Prostate Segmentation in MRI

Abstract : Variations in inter-patient prostate shape, and size and imaging artifacts in magnetic resonance images (MRI) hinders automatic accurate prostate segmentation. In this paper we propose a graph cut based energy minimization of the posterior probabilities obtained in a supervised learning schema for automatic 3D segmentation of the prostate in MRI. A probabilistic classification of the prostate voxels is achieved with a probabilistic atlas and a random forest based learning framework. The posterior probabilities are combined to obtain the likelihood of a voxel being prostate. Finally, 3D graph cut based energy minimization in the stochastic space provides segmentation of the prostate. The proposed method achieves a mean Dice similarity coefficient (DSC) value of 0.91±0.04 and 95% mean Hausdorff distance (HD) of 4.69±2.62 voxels when validated with 15 prostate volumes of a public dataset in a leave-one-patient-out validation framework. The model achieves statistically significant t-test p-value<0.0001 in mean DSC and mean HD values compared to some of the works in literature.
Document type :
Conference papers
Liste complète des métadonnées

Cited literature [8 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-00710964
Contributor : Soumya Ghose <>
Submitted on : Friday, June 22, 2012 - 9:51:20 AM
Last modification on : Wednesday, September 12, 2018 - 1:27:47 AM
Document(s) archivé(s) le : Sunday, September 23, 2012 - 2:21:59 AM

File

ICPR2012_MRI_Soumya2.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-00710964, version 1

Collections

Citation

Soumya Ghose, Jhimli Mitra, Arnau Oliver, Robert Marti, Xavier Llado, et al.. Graph Cut Energy Minimization in a Probabilistic Learning Framework for 3D Prostate Segmentation in MRI. IAPR International Conference on Pattern Recognition, Nov 2012, Tsukba, Japan. ⟨hal-00710964⟩

Share

Metrics

Record views

261

Files downloads

258