Automated MAP-MRF EM labelling for volume determination in PET

Gribben, Hugh; Miller, Paul; Wang, Hongbin; Carson, K.J.; Hounsell, Alan R.; Zatari, A.

doi:10.1109/isbi.2008.4540917

Cited by 9 publications

(13 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For the patient images, the union ground truth is directly used to compare our method with alternative approaches [9,10]. The mutual information curves of our method in patient images are closer to the ground truth curves, except for a few points for patient 2.…”

Section: Accuracy Evaluationmentioning

confidence: 88%

“…In this section, we present the test data and the experiments used to test the accuracy of the proposed hierarchical segmentation algorithm. We compare our approach with two state-of-the-art methods, namely PGVF [9] and MAP-MRF-EM [10].…”

Section: Methodsmentioning

confidence: 99%

“…the optimal segmentation. Using this metric we compare the proposed method and previous segmentation methods [9,10] for both real and phantom PET. Note that computing DSC requires binary input images, but both our segmentation results and the union ground truth for the patient data have soft alpha values.…”

Section: Accuracy Evaluationmentioning

confidence: 99%

“…Hsu et al [9] utilized an edge-based segmentation model called Poisson Gradient Vector Flow (PGVF) to automatically find the abnormality outlines in PET images. Gribben et al [10] used the Markov Random Field Expectation Maximisation (MAP-MRF EM) labelling technique to segment VOIs. Montgomery et al [11] proposed a segmentation method based on a multiscale Markov Random Field Model (MRFM).…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Unsupervised tumour segmentation in PET using local and global intensity-fitting active surface and alpha matting

Zeng¹,

Wang²,

Tiddeman³

et al. 2013

Computers in Biology and Medicine

View full text Add to dashboard Cite

This paper proposes an unsupervised tumour segmentation approach for PET data. The method computes the volumes of interest (VOIs) with sub-voxel precision by considering the limited image resolution and partial volume effects. First, an improved anisotropic diffusion filter is used to remove image noise. A hierarchical local and global intensity active surface modelling scheme is then applied to segment VOIs, followed by an alpha matting step to further refine the segmentation boundary. The proposed method is validated on real PET images of head-and-neck cancer patients with ground truth provided by human experts, as well as custom-designed phantom PET images with objective ground truth. Experimental results show that our method outperforms previous automatic approaches in terms of segmentation accuracy.

show abstract

Section: Accuracy Evaluationmentioning

confidence: 88%

Section: Methodsmentioning

confidence: 99%

Section: Accuracy Evaluationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Unsupervised tumour segmentation in PET using local and global intensity-fitting active surface and alpha matting

Zeng¹,

Wang²,

Tiddeman³

et al. 2013

Computers in Biology and Medicine

View full text Add to dashboard Cite

show abstract

“…Other investigators have proposed the use of Bayesian frameworks, in which a prior knowledge or power transformation is included as part of the estimation as a way of reducing the errors resulting from noise and PVE [16][17][18]. These approaches have been reported to be accurate for segmenting brain tissue, including gray matter, white matter and CSF and might be adequate for segmenting PET images [19].…”

Section: Introductionmentioning

confidence: 99%