| Literature DB >> 26267800 |
Rolf A Heckemann, Christian Ledig, Katherine R Gray, Paul Aljabar, Daniel Rueckert, Joseph V Hajnal, Alexander Hammers.
Abstract
Entities:
Year: 2015 PMID: 26267800 PMCID: PMC4534373 DOI: 10.1371/journal.pone.0135746
Source DB: PubMed Journal: PLoS One ISSN: 1932-6203 Impact factor: 3.240
Fig 1Sample images chosen randomly from each dataset.
Images were visually centred at the level of the commissures approximately in the centre of the left thalamus to acquire a transverse (top rows) and a sagittal (bottom rows) slice. Left column, O, H, L: manual reference masks, X: generated mask (OXsetup). Middle column, O, H, L, X: generated masks (HX setup in the case of X). Right column, O, H, L, X: discrepancies between the masks—green indicates overinclusion, red indicates underinclusion. Individual JCs were 0.9512 (O), 0.9704 (H), 0.9647 (L), and 0.9503 (X).
Fig 2Overview diagram of pincram workflow.
Step numbers in the text correspond to numbered boxes. C : pre-selection fused mask;M : tight margin (boundary neighborhood) mask; F : fuzzy label summed from rank-selected subset; : brain mask generated from F by thresholding and binarization; wide margin mask generated from F by thresholding and binarization