PuSH - Publication Server of Helmholtz Zentrum München

Lombardo, E.* ; Hess J. ; Kurz, C.* ; Riboldi, M.* ; Marschner, S.* ; Baumeister, P. ; Lauber, K. ; Pflugradt, U. ; Walch, A.K. ; Canis, M. ; Klauschen, F.* ; Zitzelsberger, H. ; Belka, C. ; Landry, G.* ; Unger, K.

DeepClassPathway: Molecular pathway aware classification using explainable deep learning.

Eur. J. Cancer 176, 41-49 (2022)
Postprint DOI PMC
Open Access Green
OBJECTIVE: HPV-associated head and neck cancer is correlated with favorable prognosis; however, its underlying biology is not fully understood. We propose an explainable convolutional neural network (CNN) classifier, DeepClassPathway, that predicts HPV-status and allows patient-specific identification of molecular pathways driving classifier decisions. METHODS: The CNN was trained to classify HPV-status on transcriptome data from 264 (13% HPV-positive) and tested on 85 (25% HPV-positive) head and neck squamous carcinoma patients after transformation into 2D-treemaps representing molecular pathways. Grad-CAM saliency was used to quantify pathways contribution to individual CNN decisions. Model stability was assessed by shuffling pathways within 2D-images. RESULTS: The classification performance of the CNN-ensembles achieved ROC-AUC/PR-AUC of 0.96/0.90 for all treemap variants. Quantification of the averaged pathway saliency heatmaps consistently identified KRAS, spermatogenesis, bile acid metabolism, and inflammation signaling pathways as the four most informative for classifying HPV-positive patients and MYC targets, epithelial-mesenchymal transition, and protein secretion pathways for HPV-negative patients. CONCLUSION: We have developed and applied an explainable CNN classification approach to transcriptome data from an oncology cohort with typical sample size that allows classification while accounting for the importance of molecular pathways in individual-level decisions.
Altmetric
Additional Metrics?
[➜Log in]
Edit extra informations Login
Publication type Article: Journal article
Document type Scientific Article
Corresponding Author
Keywords Cnn ; Deep Learning ; Grad-cam ; Hnscc ; Hpv ; Transcriptome
ISSN (print) / ISBN 0959-8049
e-ISSN 1879-0852
Journal European Journal of Cancer
Quellenangaben Volume: 176, Issue: , Pages: 41-49 Article Number: , Supplement: ,
Publisher Elsevier
Non-patent literature Publications
Reviewing status Peer reviewed
Institute(s) Translational Metabolic Oncology (TMO)
CCG Personalized Radiotherapy in Head and Neck Cancer (KKG-KRT)
Research Unit Analytical Pathology (AAP)
Grants Bundesministerium für Bildung und Forschung