Abstract<\/h2>\n
Segmentation plays a critical role in exposing connections between biological structure and function. The process of label<\/span> fusion<\/span> collects and combines multiple<\/span> observations into a single estimate. Statistically driven techniques provide mechanisms to optimally combine segmentations; yet, optimality hinges upon accurate modeling of rater behavior. Traditional approaches, e.g., Majority Vote and Simultaneous Truth and Performance Level Estimation (STAPLE), have been shown to yield excellent performance in some cases, but do not account for spatial dependences of rater performance (i.e., regional task difficulty). Recently, the COnsensus<\/span> Level, Labeler Accuracy and Truth Estimation (COLLATE) label<\/span> fusion<\/span> technique augmented the seminal STAPLE approach to simultaneously estimate regions of relative consensus<\/span> versus confusion along with rater performance. Herein, we extend the COLLATE framework to account for multiple<\/span> consensus<\/span> levels. Toward this end, we posit a generalized<\/span> model of rater behavior of which Majority Vote, STAPLE, STAPLE Ignoring Consensus<\/span> Voxels, and COLLATE are special cases. The new algorithm is evaluated with simulations and shown to yield improved performance in cases with complex region difficulties. Multi-COLLATE achieve these results by capturing different consensus<\/span> levels. The potential impacts and applications of generative model to label<\/span> fusion<\/span> problems are discussed.<\/p>\n Z. Xu, A. Asman and B. Landman. \u201cGeneralized Statistical Label Fusion using Multiple Consensus Levels.\u201d In Proceedings of the SPIE Medical Imaging Conference. San Diego, California, February 2012 (Oral Presentation) PMC3438516\u2020 Full Text: https:\/\/www.ncbi.nlm.nih.gov\/pubmed\/?term=%E2%80%9CGeneralized+Statistical+Label+Fusion+using+Multiple+Consensus+Levels. Abstract Segmentation plays a critical role in exposing connections between biological structure and function. The process of label fusion collects and…<\/p>\n","protected":false},"author":6300,"featured_media":660,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[3,82,4,65,40],"tags":[99,42],"class_list":["post-659","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-image-segmentation","category-magnetic-resonance-imaging","category-neuroimaging","category-registration","category-reproducibility","tag-collate","tag-staple"],"_links":{"self":[{"href":"https:\/\/my.dev.vanderbilt.edu\/masi\/wp-json\/wp\/v2\/posts\/659","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/my.dev.vanderbilt.edu\/masi\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/my.dev.vanderbilt.edu\/masi\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/my.dev.vanderbilt.edu\/masi\/wp-json\/wp\/v2\/users\/6300"}],"replies":[{"embeddable":true,"href":"https:\/\/my.dev.vanderbilt.edu\/masi\/wp-json\/wp\/v2\/comments?post=659"}],"version-history":[{"count":2,"href":"https:\/\/my.dev.vanderbilt.edu\/masi\/wp-json\/wp\/v2\/posts\/659\/revisions"}],"predecessor-version":[{"id":759,"href":"https:\/\/my.dev.vanderbilt.edu\/masi\/wp-json\/wp\/v2\/posts\/659\/revisions\/759"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/my.dev.vanderbilt.edu\/masi\/wp-json\/wp\/v2\/media\/660"}],"wp:attachment":[{"href":"https:\/\/my.dev.vanderbilt.edu\/masi\/wp-json\/wp\/v2\/media?parent=659"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/my.dev.vanderbilt.edu\/masi\/wp-json\/wp\/v2\/categories?post=659"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/my.dev.vanderbilt.edu\/masi\/wp-json\/wp\/v2\/tags?post=659"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}![\"Figure](\"https:\/\/my.dev.vanderbilt.edu\/masi\/wp-content\/uploads\/sites\/2304\n2661\/2016\/11\/nihms335127f2.jpg\")
Results from the controlled rater observation simulation. The target intensity image and the corresponding true labels can be seen (A) and (B), respectively. The set of input observations used in this experiment can be seen in (C). The results of the four algorithms, Majority Vote, STAPLE, Binary-COLLATE and Multi-COLLATE can be seen in (D)\u2013(G). Note that the 3 consensus-level implementation of Multi-COLLATE is the only algorithm that gets the correct answer. The 3 energy in each of the consensus levels can be seen in (H)\u2013(J). Note that by distributing the non-consensus voxels into two separate levels, (H) and (I), the estimation process converges to the correct answer.<\/figcaption><\/figure>\n","protected":false},"excerpt":{"rendered":"