Zur Kurzanzeige

2022-12-15Zeitschriftenartikel
Analysing cerebrospinal fluid with explainable deep learning: From diagnostics to insights
dc.contributor.authorSchweizer, Leonille
dc.contributor.authorSeegerer, Philipp
dc.contributor.authorKim, Hee-yeong
dc.contributor.authorSaitenmacher, René
dc.contributor.authorMuench, Amos
dc.contributor.authorBarnick, Liane
dc.contributor.authorOsterloh, Anja
dc.contributor.authorDittmayer, Carsten
dc.contributor.authorJödicke, Ruben
dc.contributor.authorPehl, Deborah
dc.contributor.authorReinhardt, Annekathrin
dc.contributor.authorRuprecht, Klemens
dc.contributor.authorStenzel, Werner
dc.contributor.authorWefers, Annika K.
dc.contributor.authorHarter, Patrick N.
dc.contributor.authorSchüller, Ulrich
dc.contributor.authorHeppner, Frank L.
dc.contributor.authorAlber, Maximilian
dc.contributor.authorMüler, Klaus-Robert
dc.contributor.authorKlauschen, Frederick
dc.date.accessioned2024-09-03T14:40:07Z
dc.date.available2024-09-03T14:40:07Z
dc.date.issued2022-12-15none
dc.identifier.other10.1111/nan.12866
dc.identifier.urihttp://edoc.rki.de/176904/12078
dc.description.abstractAim 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.eng
dc.language.isoengnone
dc.publisherRobert Koch-Institut
dc.rights(CC BY 3.0 DE) Namensnennung 3.0 Deutschlandger
dc.rights.urihttp://creativecommons.org/licenses/by/3.0/de/
dc.subjectcell detectioneng
dc.subjectcerebrospinal fluideng
dc.subjectdeep learningeng
dc.subjectexplainable AIeng
dc.subjectheatmapseng
dc.subject.ddc610 Medizin und Gesundheitnone
dc.titleAnalysing cerebrospinal fluid with explainable deep learning: From diagnostics to insightsnone
dc.typearticle
dc.identifier.urnurn:nbn:de:0257-176904/12078-3
dc.type.versionpublishedVersionnone
local.edoc.container-titleNeuropathology and Applied Neurobiologynone
local.edoc.container-issn1365-2990none
local.edoc.pages16none
local.edoc.type-nameZeitschriftenartikel
local.edoc.container-typeperiodical
local.edoc.container-type-nameZeitschrift
local.edoc.container-urlhttps://onlinelibrary.wiley.com/journal/13652990none
local.edoc.container-publisher-nameJohn Wiley & Sons, Incnone
local.edoc.container-volume49none
local.edoc.container-issue1none
local.edoc.container-reportyear2022none
dc.description.versionPeer Reviewednone

Zur Kurzanzeige