2018
DOI: 10.1093/bioinformatics/bty939
|View full text |Cite
|
Sign up to set email alerts
|

Clustering of samples with a tree-shaped dependence structure, with an application to microscopic time lapse imaging

Abstract: Motivation Recent imaging technologies allow for high-throughput tracking of cells as they migrate, divide, express fluorescent markers and change their morphology. The interpretation of these data requires unbiased, efficient statistical methods that model the dynamics of cell phenotypes. Results We introduce treeHFM, a probabilistic model which generalizes the theory of hidden Markov models to tree structured data. While ac… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1

Citation Types

0
5
0

Year Published

2018
2018
2022
2022

Publication Types

Select...
3
1
1

Relationship

1
4

Authors

Journals

citations
Cited by 5 publications
(5 citation statements)
references
References 31 publications
0
5
0
Order By: Relevance
“…Our aim is to introduce dependencies on global tissue context and cell neighborhood and enhance learning results for cell classification from deep convolution neural networks (CNNs). Probabilistic graphical models have successfully been applied to improve cell classification in time-lapse imaging by taking into account the temporal context of a cell (1015). Probabilistic graphical models have also been used successfully in histopathology images for pathology detection and segmentation (16–19), disease and tissue staging (20, 21), and nuclei segmentation (22).…”
Section: Introductionmentioning
confidence: 99%
“…Our aim is to introduce dependencies on global tissue context and cell neighborhood and enhance learning results for cell classification from deep convolution neural networks (CNNs). Probabilistic graphical models have successfully been applied to improve cell classification in time-lapse imaging by taking into account the temporal context of a cell (1015). Probabilistic graphical models have also been used successfully in histopathology images for pathology detection and segmentation (16–19), disease and tissue staging (20, 21), and nuclei segmentation (22).…”
Section: Introductionmentioning
confidence: 99%
“…LEM still leaves room for further improvements that extend its applicability to various problems, some of which may be addressed by using existing techniques of the hidden Markov models. For instance, we may relax the assumption of the independence of the state-switching between the daughter cells (26,45). This generalization may be useful when we include the size of a cell as a state, which naturally correlates between the daughters (46,47).…”
Section: Summary and Discussionmentioning
confidence: 99%
“…First, the inference from a tree should be conducted by appropriately handling the branching relationship among the cells in the tree. This problem has been studied by using the kin-correlation (21,22), an algebraic invariance of the lineage tree (23,24), clustering algorithms (25,26), Monte-Carlo algorithms (27), and model selection (28). See also Supplementary Information (Section S1).…”
Section: Introductionmentioning
confidence: 99%
“…First, the estimation with respect to a tree should be conducted by appropriately handling the branching relationship among the cells in the tree. This problem has been studied by using the kin-correlation (29,30), an algebraic invariance of the lineage tree (31,32), clustering algorithms (33,34), Monte-Carlo algorithms (35), and model selection (36). While this problem seems to be addressed by combining these existing estimation techniques in the machine learning, this naïve anticipations is hampered by the second difficulty.…”
Section: Introductionmentioning
confidence: 99%
“…LEM still leaves room for further improvements that extend its applicability to various problems, 300 some of which may be addressed by using existing techniques of the hidden Markov models. For 301 instance, we may relax the assumption of the independence of the state-switching between the 302 daughter cells (34,54). This generalization may be useful when we include the size of a cell as a 303 state, which naturally correlates between the daughters (25,55).…”
mentioning
confidence: 99%