Multimodal deep learning for wound healing prediction
European Journal of Pharmaceutical Sciences · 2026-12 · 227 · 107693
DOI: 10.1016/j.ejps.2026.107693 ↗Kurz gefasst
Fachartikel in der Zeitschrift European Journal of Pharmaceutical Sciences, 2026-12. Verfasst von Tatu Rojalin und 10 weiteren Autor:innen. Beteiligte Einrichtung: Faculty of Pharmacy, Division of Pharmaceutical Biosciences, Drug Research Program, University of Helsinki, University of Finland, P.O. Box 56 (Viiki…. Hauptthemen laut MEDLINE: Wound Healing, Deep Learning, Diabetic Foot. Die Arbeit nennt 38 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Jenny and Antti Wihuri Foundation.
Themen: Wundversorgung
Abstract (Originalsprache)
Chronic wounds, particularly diabetic foot ulcers, heal unpredictably and impose substantial healthcare costs. Deep learning (DL) systems that integrate complementary data streams could support preclinical studies by assessing wound status and therapy efficacy reliably in small-n studies. We developed a three-branch neural network combining convolutional feature extraction for red-green-blue (RGB) images, infrared thermography, and wound-area measurements to classify healing stage and infer treatment response in diabetic mouse wound model. In descriptive ablation analysis, the model integrating RGB and wound-size inputs achieved higher accuracy than the RGB-only configuration (0.75 vs. 0.70), suggesting that explicit size context provided useful complementary information. Thermal images appeared qualitatively informative during early healing but contributed limited additional signal later, likely reflecting subtle inflammatory gradients or angiogenesis in healthy murine skin; this observation motivated a constrained weighted-fusion strategy. In the development-stage comparison, static fusion with tunable fixed weights yielded the highest observed performance (∼0.85 accuracy on the development test subject). Grad-CAM visualizations showed that predictions arose from biologically relevant regions rather than image artifacts. Experiments without subject-level partitioning produced inflated metrics, underscoring the need for grouping by subject to ensure generalizable evaluation. Nested Leave-One-Group-Out cross-validation with KerasTuner-based hyperparameter tuning yielded a mean accuracy of 0.71, estimating performance on unseen subjects. This resulting proof-of-concept model predicts treatment responses for wounds treated with nanofibrillated-cellulose hydrogel. Multimodal DL combining standard and thermal photography with size metrics can stage wound healing and detect divergent treatment trajectories and thus be used in preclinical efficacy evaluation of novel wound treatments. With larger, clinically annotated datasets, this approach could be applied in a clinical setting as well.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Faculty of Pharmacy, Division of Pharmaceutical Biosciences, Drug Research Program, University of Helsinki, University of Finland, P.O. Box 56 (Viikinkaari 5E), 00014, Finland; Max Planck Florida Institute for Neuroscie…
- Faculty of Pharmacy, Division of Pharmaceutical Biosciences, Drug Research Program, University of Helsinki, University of Finland, P.O. Box 56 (Viikinkaari 5E), 00014, Finland.
- Faculty of Pharmacy, Division of Pharmaceutical Biosciences, Drug Research Program, University of Helsinki, University of Finland, P.O. Box 56 (Viikinkaari 5E), 00014, Finland; Department of Energy and Mechanical Engine…
- Department of Energy and Mechanical Engineering, Aalto University, Finland.
- Made Consulting, 20520, Turku, Finland.
- Department of Computing, University of Turku, 20014, Turku, Finland.
Förderung
- Jenny and Antti Wihuri Foundation
Zitieren
APA
Rojalin, T., Kröger, A., Koivunotko, E., Pörhö, B., Laaksonen, T., Partanen, J., Pridgeon, C. S., Madetoja, M., Airola, A., Akmal, J., & Yliperttula, M. (2026). Multimodal deep learning for wound healing prediction. European Journal of Pharmaceutical Sciences, 227, 107693. https://doi.org/10.1016/j.ejps.2026.107693
Vancouver
Rojalin T, Kröger A, Koivunotko E, Pörhö B, Laaksonen T, Partanen J, et al. Multimodal deep learning for wound healing prediction. European Journal of Pharmaceutical Sciences. 2026;227:107693. doi:10.1016/j.ejps.2026.107693
BibTeX
@article{Rojalin2026,
author = {Rojalin, Tatu and Kröger, Aleksi and Koivunotko, Elle and Pörhö, Benjamin and Laaksonen, Timo and Partanen, Jouni and Pridgeon, Chris S and Madetoja, Mari and Airola, Antti and Akmal, Jan and Yliperttula, Marjo},
title = {Multimodal deep learning for wound healing prediction},
journal = {European Journal of Pharmaceutical Sciences},
year = {2026},
volume = {227},
pages = {107693},
doi = {10.1016/j.ejps.2026.107693}
}
EUVIMED