MétaCan
Menu
← Back to cohort

seededclustR: Using Cluster Analysis of Mitochondria to Improve Cell Segmentation and Automated Quantification of mtDNA in Smooth Muscle Cells.

2020· article· en· W3016573554 on OpenAlexaffabout
Anita Santos, Irvin Ng, Damon Poburko

Bibliographic record

VenueThe FASEB Journal · 2020
Typearticle
Languageen
FieldBiochemistry, Genetics and Molecular Biology
TopicCell Image Analysis Techniques
Canadian institutionsSimon Fraser University
Fundersnot available
KeywordsVoronoi diagramSegmentationCentroidArtificial intelligencePrecision and recallPattern recognition (psychology)Computer scienceIntersection (aeronautics)Biological systemMathematicsComputer visionCombinatoricsChemistryBiologyCartographyGeometry

Abstract

fetched live from OpenAlex

Intro Segmentation of fluorescently labelled cells is essential to pipelines of automated image analysis. Identifying regions of interest (ROIs) around labelled nuclei is relatively easy using tools available in ImageJ/Fiji, but determining the bounds of cells can be much harder. We asked if fluorescently labelled mitochondria could be used to improve the automated generation of cellular ROIs. Method The perimeter coordinates of nuclear ROIs and the centroid coordinates, area, mean intensity, angle, and solidity of ROIs around MitoTracker Orange (MTO)‐labelled mitochondria were generated in ImageJ and exported to R. Using the nuclei as seeds for a recursive k‐Nearest Neighbour search extending from each nucleus, MTO particles were assigned to their nearest‐neighbouring cells. A concave hull was created around MTO particles assigned to each cell using the concaveman package. We determined the sensitivity of this algorithm by comparing the accuracy of the generated concave hulls to the corresponding hand‐drawn ROIs, and ROIs from Voronoi expanded distance maps Results When we compared seededClustR ROIs to those drawn by hand, the Intersection‐Over‐Union (IOU) was 68 ± 8% (mean ± SD), Precision was 88 ± 6% and Recall 95 ± 3%. In contrast, Voronoi expansion of nuclei resulted in the IOU of 46 ± 18%, Precision of 83 ± 17%, and Recall of 91 ± 11%. Optimal IOU and Recall were observed when each MTO particle was compared to its 7–10 nearest neighbours. IOU and Recall were maximized using recursive steps of 40–200 μm, while Precision decreased from 94% to 88% as step size increased from 7–40 μm. Unexpectedly, extending seededClustR to an n‐dimensional kd‐tree calculation that included varying weights of MTO particle mean intensity, area, standard deviation, angle and solidity reduced cell tracing accuracy. Similarly, using Wilks’‐test to identify outliers decreased IOU by 17 percentage points (pp) and Recall by 20 pp, though it increased Precision by 6 pp. Finally, we tested whether if the more accurate cell ROI from seededClustR versus Voronoi ROIs could improve our ability to detect changes in optical measures of mtDNA in A7r5 smooth muscle cells treated with angiotensin II, 5,6‐dideoxycytidine, and H2O2, where results showed promising gains. Conclusion seededClustR is a generalizable algorithm that can use simple XY coordinates of puncta‐like features of cells to generate cellular ROIs around labelled nuclei with measures of accuracy that exceed simple dilation of nuclear ROIs. This approach should facilitate improved precision in automated screens of cellular phenotypes. Support or Funding Information Natural Sciences and Engineering Research Council of Canada

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

How this classification was reachedexpand

Full frame machine prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.

metaresearch head score (Codex)0.007
metaresearch head score (Gemma)0.016
Version: metacan-v3-hybrid-931329e0061cValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Simulation or modeling · Consensus signal: none
GenreCandidate signal: Methods · Consensus signal: Methods
Teacher disagreement score0.009
Threshold uncertainty score0.037

Distilled classifier scores by category (both heads)

CategoryCodexGemma
Metaresearch0.0070.016
Meta-epidemiology (narrow)0.0030.002
Meta-epidemiology (broad)0.0030.004
Bibliometrics0.0040.003
Science and technology studies0.0020.001
Scholarly communication0.0040.003
Open science0.0040.005
Research integrity0.0030.003
Insufficient payload (model declined to judge)0.0060.005

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.013
GPT teacher head0.272
Teacher spread0.260 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.

The models applied no category: nothing in the taxonomy fit this work.
Study designSimulation or modeling
Domainnot available
GenreMethods

How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".

Quick stats

Citations0
Published2020
Admission routes2
Has abstractyes

Explore more

Same venueThe FASEB Journal→Same topicCell Image Analysis Techniques→French-language works237,207→