EUVIMEDObjevte EUVIMED
Open access · CC BY 4.0

Analysing cerebrospinal fluid with explainable deep learning: From diagnostics to insights

Leonille Schweizer iD, Philipp Seegerer iD, Hee-Yeong Kim, René Saitenmacher, Amos Muench, Liane Barnick, Anja Osterloh, Carsten Dittmayer, Ruben Jödicke, Debora Pehl, Annekathrin Reinhardt, Klemens Ruprecht and 8 more

Neuropathology and Applied Neurobiology · 2023-02 · 49 (1)

DOI: 10.1111/nan.12866 ↗

At a glance

Journal article in Neuropathology and Applied Neurobiology, published online on 2023-01-17. Written by Leonille Schweizer and 19 co-authors. Institution: Department of Neuropathology, Charité - Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt-Universität zu Berlin, …. Major MEDLINE topics: Deep Learning. Cited 8 times so far (Crossref); 45 references. The full text is freely available (open access, CC BY 4.0). Funded by: Deutsche Forschungsgemeinschaft, Deutschen Konsortium für Translationale Krebsforschung.

AbstraktSubjectsBibliografické údajeCiteRelated studies

Abstract (original language)

Aim

Analysis of cerebrospinal fluid (CSF) is essential for diagnostic workup of patients with neurological diseases and includes differential cell typing. The current gold standard is based on microscopic examination by specialised technicians and neuropathologists, which is time‐consuming, labour‐intensive and subjective.

Methods

We, therefore, developed an image analysis approach based on expert annotations of 123,181 digitised CSF objects from 78 patients corresponding to 15 clinically relevant categories and trained a multiclass convolutional neural network (CNN).

Results

The CNN classified the 15 categories with high accuracy (mean AUC 97.3%). By using explainable artificial intelligence (XAI), we demonstrate that the CNN identified meaningful cellular substructures in CSF cells recapitulating human pattern recognition. Based on the evaluation of 511 cells selected from 12 different CSF samples, we validated the CNN by comparing it with seven board‐certified neuropathologists blinded for clinical information. Inter‐rater agreement between the CNN and the ground truth was non‐inferior (Krippendorff's alpha 0.79) compared with the agreement of seven human raters and the ground truth (mean Krippendorff's alpha 0.72, range 0.56–0.81). The CNN assigned the correct diagnostic label (inflammatory, haemorrhagic or neoplastic) in 10 out of 11 clinical samples, compared with 7–11 out of 11 by human raters.

Conclusions

Our approach provides the basis to overcome current limitations in automated cell classification for routine diagnostics and demonstrates how a visual explanation framework can connect machine decision‐making with cell properties and thus provide a novel versatile and quantitative method for investigating CSF manifestations of various neurological diseases.

Subjects

MeSH terms (MEDLINE)

Deep LearningHumansArtificial IntelligenceNeural Networks, ComputerImage Processing, Computer-Assisted

Author keywords

cell detectioncerebrospinal fluiddeep learningexplainable AIheatmaps

Authors and institutions

  • Department of Neuropathology, Charité - Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt-Universität zu Berlin, Berlin, Germany.
  • German Cancer Consortium (DKTK), Partner Site Berlin, German Cancer Research Center (DKFZ), Heidelberg, Germany.
  • Machine-Learning Group, Department of Software Engineering and Theoretical Computer Science, Technische Universität Berlin, Berlin, Germany.
  • Aignostics GmbH, Berlin, Germany.
  • Systems Medicine of Infectious Disease, Robert Koch Institute, Berlin, Germany.
  • Department of Pathology, Vivantes Hospitals Berlin, Berlin, Germany.

ORCID profiles: Leonille Schweizer, Philipp Seegerer, Werner Stenzel, Annika K Wefers, Ulrich Schüller

Funding

  • Deutsche Forschungsgemeinschaft
  • Deutschen Konsortium für Translationale Krebsforschung

Cite

APA

Schweizer, L., Seegerer, P., Kim, H. Y., Saitenmacher, R., Muench, A., Barnick, L., Osterloh, A., Dittmayer, C., Jödicke, R., Pehl, D., Reinhardt, A., Ruprecht, K., Stenzel, W., Wefers, A. K., Harter, P. N., Schüller, U., Heppner, F. L., Alber, M., Müller, K. R., & Klauschen, F. (2023). Analysing cerebrospinal fluid with explainable deep learning: From diagnostics to insights. Neuropathology and Applied Neurobiology, 49(1). https://doi.org/10.1111/nan.12866

Vancouver

Schweizer L, Seegerer P, Kim HY, Saitenmacher R, Muench A, Barnick L, et al. Analysing cerebrospinal fluid with explainable deep learning: From diagnostics to insights. Neuropathology and Applied Neurobiology. 2023;49(1). doi:10.1111/nan.12866

BibTeX
@article{Schweizer2023,
  author = {Schweizer, Leonille and Seegerer, Philipp and Kim, Hee-Yeong and Saitenmacher, René and Muench, Amos and Barnick, Liane and Osterloh, Anja and Dittmayer, Carsten and Jödicke, Ruben and Pehl, Debora and Reinhardt, Annekathrin and Ruprecht, Klemens and Stenzel, Werner and Wefers, Annika K and Harter, Patrick N and Schüller, Ulrich and Heppner, Frank L and Alber, Maximilian and Müller, Klaus-Robert and Klauschen, Frederick},
  title = {Analysing cerebrospinal fluid with explainable deep learning: From diagnostics to insights},
  journal = {Neuropathology and Applied Neurobiology},
  year = {2023},
  volume = {49},
  number = {1},
  doi = {10.1111/nan.12866}
}
Start for free →