On the objectivity, reliability, and validity of deep learning enabled bioimage analyses

Abstract

Bioimage analysis of fluorescent labels is widely used in the life sciences. Recent advances in deep learning (DL) allow automating time-consuming manual image analysis processes based on annotated training data. However, manual annotation of fluorescent features with a low signal-to-noise ratio is somewhat subjective. Training DL models on subjective annotations may be instable or yield biased models. In turn, these models may be unable to reliably detect biological effects. An analysis pipeline integrating data annotation, ground truth estimation, and model training can mitigate this risk. To evaluate this integrated process, we compared different DL-based analysis approaches. With data from two model organisms (mice, zebrafish) and five laboratories, we show that ground truth estimation from multiple human annotators helps to establish objectivity in fluorescent feature annotations. Furthermore, ensembles of multiple models trained on the estimated ground truth establish reliability and validity. Our research provides guidelines for reproducible DL-based bioimage analyses.

Data availability

Official repository of our study "On the objectivity, reliability, and validity of deep learning enabled bioimage analyses" can be found at Dryad (https://doi.org/10.5061/dryad.4b8gtht9d). In addition, we also provide all code in our GitHub repository (https://github.com/matjesg/bioimage_analysis).

The following data sets were generated

Article and author information

Author details

  1. Dennis Segebarth

    Institute of Clinical Neurobiology, University Hospital Würzburg, Würzburg, Germany
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3806-9324
  2. Matthias Griebel

    Department of Business and Economics, University of Würzburg, Würzburg, Germany
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1959-0242
  3. Nikolai Stein

    Department of Business and Economics, University of Würzburg, Würzburg, Germany
    Competing interests
    The authors declare that no competing interests exist.
  4. Cora R von Collenberg

    Institute of Clinical Neurobiology, University Hospital Würzburg, Würzburg, Germany
    Competing interests
    The authors declare that no competing interests exist.
  5. Corinna Martin

    Institute of Clinical Neurobiology, University Hospital Würzburg, Würzburg, Germany
    Competing interests
    The authors declare that no competing interests exist.
  6. Dominik Fiedler

    Institute of Physiology, Westfälische Wilhlems-Universität, Münster, Germany
    Competing interests
    The authors declare that no competing interests exist.
  7. Lucas B Comeras

    Department of Pharmacology, University of Innsbruck, Innsbruck, Austria
    Competing interests
    The authors declare that no competing interests exist.
  8. Anupam Sah

    Department of Pharmacology and Toxicology, University of Innsbruck, Innsbruck, Austria
    Competing interests
    The authors declare that no competing interests exist.
  9. Victoria Schoeffler

    Department of Child and Adolescent Psychiatry, University Hospital Würzburg, Würzburg, Germany
    Competing interests
    The authors declare that no competing interests exist.
  10. Teresa Lüffe

    Department of Child and Adolescent Psychiatry, University Hospital Würzburg, Würzburg, Germany
    Competing interests
    The authors declare that no competing interests exist.
  11. Alexander Dürr

    Department of Business and Economics, University of Würzburg, Würzburg, Germany
    Competing interests
    The authors declare that no competing interests exist.
  12. Rohini Gupta

    Institute of Clinical Neurobiology, University Hospital Würzburg, Würzburg, Germany
    Competing interests
    The authors declare that no competing interests exist.
  13. Manju Sasi

    Institute of Clinical Neurobiology, University Hospital Würzburg, Würzburg, Germany
    Competing interests
    The authors declare that no competing interests exist.
  14. Christina Lillesaar

    Department of Child and Adolescent Psychiatry, University Hospital Würzburg, Würzburg, Germany
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5166-2851
  15. Maren D Lange

    Institute of Physiology, Westfälische Wilhlems-Universität, Münster, Germany
    Competing interests
    The authors declare that no competing interests exist.
  16. Ramon O Tasan

    Department of Pharmacology, University of Inssbruck, Innsbruck, Austria
    Competing interests
    The authors declare that no competing interests exist.
  17. Nicolas Singewald

    Department of Pharmacology and Toxicology, University of Innsbruck, Innsbruck, Austria
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0166-3370
  18. Hans-Christian Pape

    Institute of Physiology, Westfälische Wilhlems-Universität, Münster, Germany
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6874-8224
  19. Christoph M Flath

    Department of Business and Economics, University of Würzburg, Würzburg, Germany
    For correspondence
    christoph.flath@uni-wuerzburg.de
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1761-9833
  20. Robert Blum

    Institute of Clinical Neurobiology, University Hospital Würzburg, Würzburg, Germany
    For correspondence
    Blum_R@ukw.de
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5270-3854

Funding

Deutsche Forschungsgemeinschaft (ID 44541416 - TRR58,A10)

  • Robert Blum

Deutsche Forschungsgemeinschaft (ID 44541416 - TRR58,A03)

  • Hans-Christian Pape

Deutsche Forschungsgemeinschaft (ID 44541416 - TRR58,B08)

  • Maren D Lange

Graduate School of Life Sciences Wuerzburg (fellowship)

  • Rohini Gupta
  • Manju Sasi

Austrian Science Fund (P29952 & P25851)

  • Ramon O Tasan

Austrian Science Fund (I2433-B26,DKW-1206,SFB F4410)

  • Nicolas Singewald

Interdisziplinaeres Zentrum fuer Klinische Zusammenarbeit Wuerzburg (N-320)

  • Christina Lillesaar

Deutsche Forschungsgemeinschaft (ID 424778381)

  • Robert Blum

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Ethics

Animal experimentation: All effort was taken to minimize the number of animals used and their suffering.Lab-Wue1: All experiments with C57BL/6J wildtype mice were in accordance with the Guidelines set by the European Union and approved by our institutional Animal Care, the Utilization Committee and the Regierung von Unterfranken, Würzburg, Germany (License number: 55.2-2531.01-95/13). C57BL/6J wildtype mice were bred in the animal facility of the Institute of Clinical Neurobiology, University Hospital of Würzburg, Germany.Lab Mue: Male All animal experiments with male C57Bl/6J mice (Charles River, Sulzfeld, Germany) were carried out in accordance with European regulations on animal experimentation and protocols were approved by the local authorities (Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen).Lab-Inns1: Experiments were performed in adult, male C57Bl/6NCrl mice (Charles River, Sulzfeld, Germany). They were bred in the Department of Pharmacology at the Medical University Innsbruck, Austria. All procedures involving animals and animal care were conducted in accordance with international laws and policies (Directive 2010/63/EU of the European parliament and of the council of 22 September 2010 on the protection of animals used for scientific purposes; Guide for the Care and Use of Laboratory Animals, U.S. National Research Council, 2011) and were approved by the Austrian Ministry of Science.Lab-Inns2: Male 129S1/SvImJ (S1) mice (Charles River, Sulzfeld, Germany) were used for experimentation. The Austrian Animal Experimentation Ethics Board (Bundesministerium für Wissenschaft Forschung und Wirtschaft, Kommission für Tierversuchsangelegenheiten) approved all experimental procedures.

Copyright

© 2020, Segebarth et al.

This article is distributed under the terms of the Creative Commons Attribution License permitting unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 4,603
    views
  • 507
    downloads
  • 30
    citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Dennis Segebarth
  2. Matthias Griebel
  3. Nikolai Stein
  4. Cora R von Collenberg
  5. Corinna Martin
  6. Dominik Fiedler
  7. Lucas B Comeras
  8. Anupam Sah
  9. Victoria Schoeffler
  10. Teresa Lüffe
  11. Alexander Dürr
  12. Rohini Gupta
  13. Manju Sasi
  14. Christina Lillesaar
  15. Maren D Lange
  16. Ramon O Tasan
  17. Nicolas Singewald
  18. Hans-Christian Pape
  19. Christoph M Flath
  20. Robert Blum
(2020)
On the objectivity, reliability, and validity of deep learning enabled bioimage analyses
eLife 9:e59780.
https://doi.org/10.7554/eLife.59780

Share this article

https://doi.org/10.7554/eLife.59780

Further reading

    1. Computational and Systems Biology
    Nobuhisa Umeki, Yoshiyuki Kabashima, Yasushi Sako
    Research Article

    The RAS-MAPK system plays an important role in regulating various cellular processes, including growth, differentiation, apoptosis, and transformation. Dysregulation of this system has been implicated in genetic diseases and cancers affecting diverse tissues. To better understand the regulation of this system, we employed information flow analysis based on transfer entropy (TE) between the activation dynamics of two key elements in cells stimulated with EGF: SOS, a guanine nucleotide exchanger for the small GTPase RAS, and RAF, a RAS effector serine/threonine kinase. TE analysis allows for model-free assessment of the timing, direction, and strength of the information flow regulating the system response. We detected significant amounts of TE in both directions between SOS and RAF, indicating feedback regulation. Importantly, the amount of TE did not simply follow the input dose or the intensity of the causal reaction, demonstrating the uniqueness of TE. TE analysis proposed regulatory networks containing multiple tracks and feedback loops and revealed temporal switching in the reaction pathway primarily responsible for reaction control. This proposal was confirmed by the effects of an MEK inhibitor on TE. Furthermore, TE analysis identified the functional disorder of a SOS mutation associated with Noonan syndrome, a human genetic disease, of which the pathogenic mechanism has not been precisely known yet. TE assessment holds significant promise as a model-free analysis method of reaction networks in molecular pharmacology and pathology.

    1. Computational and Systems Biology
    2. Genetics and Genomics
    Eric V Strobl, Eric Gamazon
    Research Article

    Root causal gene expression levels – or root causal genes for short – correspond to the initial changes to gene expression that generate patient symptoms as a downstream effect. Identifying root causal genes is critical towards developing treatments that modify disease near its onset, but no existing algorithms attempt to identify root causal genes from data. RNA-sequencing (RNA-seq) data introduces challenges such as measurement error, high dimensionality and non-linearity that compromise accurate estimation of root causal effects even with state-of-the-art approaches. We therefore instead leverage Perturb-seq, or high-throughput perturbations with single-cell RNA-seq readout, to learn the causal order between the genes. We then transfer the causal order to bulk RNA-seq and identify root causal genes specific to a given patient for the first time using a novel statistic. Experiments demonstrate large improvements in performance. Applications to macular degeneration and multiple sclerosis also reveal root causal genes that lie on known pathogenic pathways, delineate patient subgroups and implicate a newly defined omnigenic root causal model.