EUVIMEDEUVIMED entdecken
Multizentrische StudieOpen Access · CC BY 4.0

Deep learning for fluorescence confocal microscopy image interpretation in radical prostatectomy

Lixiang Fang iD, Nikhil Mayor iD, Alexander Light iD, Anna Silvanto, Aiman Haider, Chase Ng, Archana Gopalakrishnan, Ranil Johann Boaz, Mariana Bertoncelli Tanaka iD, Bijan Khoubehi, Giles Hellawell, Ricardo Almeida-Magana und 11 weitere

BJU International · 2026

DOI: 10.1111/bju.70273 ↗

Kurz gefasst

Multizentrische Studie in der Zeitschrift BJU International, 2026. Verfasst von Lixiang Fang und 22 weiteren Autor:innen. Beteiligte Einrichtung: Division of Surgery, Department of Surgery and Cancer, Faculty of Medicine, Imperial College London, London, UK. Hauptthemen laut MEDLINE: Prostatectomy, Deep Learning, Prostatic Neoplasms, Image Interpretation, Computer-Assisted. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 21 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Urology Foundation, John Black Charitable Foundation, Research for Patient Benefit Programme.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Objective

To develop and validate a deep learning model for interpretation of fluorescence confocal microscopy (FCM) images for intraoperative surgical margin assessment during radical prostatectomy (RP). Patients and Methods Fluorescence confocal microscopy images from the multicentre Imperial Prostate 8‐Fluorescence Confocal Microscopy for Rapid Evaluation of Surgical Cancer Excision (IP8‐FLUORESCE) study were used to train and test a convolutional neural network model. The modified model incorporated focal loss with label smoothing, dropout regularisation, adaptive class weighting, and weighted sampling to address pronounced class imbalance. Images were pre‐processed by extracting regions of interest at a defined digital zoom level and normalised to 896 × 896 pixels. The reference standard was surgical margin status on conventional histopathology assessed by an expert histopathologist. Diagnostic performance was assessed using sensitivity, specificity, positive and negative predictive value, area under the receiver‐operating‐characteristic curve (AUC), and calibration via Brier scores. External validation was conducted using an independent dataset from the LaserSAFE feasibility trial. Model explainability was evaluated using Gradient‐weighted Class Activation Mapping (Grad‐CAM) and a custom graphical user interface (GUI) was developed to support real‐time deployment.

Results

A total of 275 images (37 tumour and 238 benign from 24 patients) were included for model development and internal testing. On the internal test set ( n = 57), the model achieved a sensitivity of 87.5%, specificity of 97.9%, and an AUC of 0.93, with good calibration (Brier score 0.16). External validation using 46 independent images yielded a sensitivity of 91.3%, specificity of 73.9%, and an AUC of 0.83, with acceptable calibration (Brier score 0.20). Grad‐CAM visualisations aligned with malignant structures on FCM images, and the GUI enabled rapid, interpretable predictions in

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

ProstatectomyDeep LearningProstatic NeoplasmsImage Interpretation, Computer-AssistedHumansMaleMicroscopy, ConfocalConvolutional Neural NetworksMargins of ExcisionMicroscopy, Fluorescence

Stichwörter der Autor:innen

artificial intelligencedeep learningfluorescence confocal microscopyintraoperative margin analysispositive surgical marginsprostate cancerradical prostatectomyrobotic surgery

Autor:innen und Institutionen

  • Division of Surgery, Department of Surgery and Cancer, Faculty of Medicine, Imperial College London, London, UK.
  • Imperial Prostate, Division of Surgery, Department of Surgery and Cancer, Faculty of Medicine, Imperial College London, London, UK.
  • Department of Urology, Imperial College Healthcare National Health Service (NHS) Trust, London, UK.
  • Department of Pathology, University College London Hospitals NHS Foundation Trust, London, UK.
  • Department of Urology, Chelsea and Westminster NHS Foundation Trust, London, UK.
  • Department of Urology, University College London Hospitals NHS Foundation Trust, London, UK.

ORCID-Profile: Lixiang Fang, Nikhil Mayor, Alexander Light, Mariana Bertoncelli Tanaka, Taimur T Shah, Hashim U Ahmed

Förderung

  • Urology Foundation
  • John Black Charitable Foundation
  • Research for Patient Benefit Programme
  • Rosetrees Trust
  • Wellcome/EPSRC Centre for Interventional and Surgical Sciences
  • Wellcome / EPSRC Centre for Interventional and Surgical Sciences

Zitieren

APA

Fang, L., Mayor, N., Light, A., Silvanto, A., Haider, A., Ng, C., Gopalakrishnan, A., Boaz, R. J., Tanaka, M. B., Khoubehi, B., Hellawell, G., Almeida-Magana, R., Mendes, L., Dinneen, E., Shaw, G., Challacombe, B., Cathcart, P., Connor, M. J., Shah, T. T., …, & Winkler, M. (2026). Deep learning for fluorescence confocal microscopy image interpretation in radical prostatectomy. BJU International. https://doi.org/10.1111/bju.70273

Vancouver

Fang L, Mayor N, Light A, Silvanto A, Haider A, Ng C, et al. Deep learning for fluorescence confocal microscopy image interpretation in radical prostatectomy. BJU International. 2026. doi:10.1111/bju.70273

BibTeX
@article{Fang2026,
  author = {Fang, Lixiang and Mayor, Nikhil and Light, Alexander and Silvanto, Anna and Haider, Aiman and Ng, Chase and Gopalakrishnan, Archana and Boaz, Ranil Johann and Tanaka, Mariana Bertoncelli and Khoubehi, Bijan and Hellawell, Giles and Almeida-Magana, Ricardo and Mendes, Larissa and Dinneen, Eoin and Shaw, Greg and Challacombe, Ben and Cathcart, Paul and Connor, Martin J and Shah, Taimur T and Ahmed, Hashim U and Fiorentino, Francesca and Giannarou, Stamatia and Winkler, Mathias},
  title = {Deep learning for fluorescence confocal microscopy image interpretation in radical prostatectomy},
  journal = {BJU International},
  year = {2026},
  doi = {10.1111/bju.70273}
}
Kostenlos starten →