Real-Time-Capable Detection of Glottic, Supraglottic and Hypopharyngeal Lesions Using Artificial Intelligence During Flexible Endoscopy
Echtzeitfähige Erkennung von glottischen, supraglottischen und hypopharyngealen Läsionen mit künstlicher Intelligenz während der flexiblen Endoskopie
Cancers · 13.08.2026 · 18 (16) · 2609
DOI: 10.3390/cancers18162609 ↗Auf einen Blick
Eine DL-Methode zeigte in einem Testset eine gute Detektionsleistung für laryngeale und pharyngeale Läsionen. Die Erkennung frühstadialer T1-Tumoren blieb eine wichtige Limitation.
Fragestellung
Kann künstliche Intelligenz bei der Echtzeit-Erkennung und Klassifikation glottischer, supraglottischer und hypopharyngealer Läsionen während der flexiblen Endoskopie unterstützen?
Methode
DL-Algorithmus für die vereinheitlichte Lokalisation und Klassifikation; Frames aus Endoskopievideos von 2 Zentren und 1 Sekundärzentrum zwischen 2012 und 2024; Test auf unabhängigem Testset nach Training mit 70 % der Daten.
Ergebnisse
Aus 490 Videos wurden 40.059 Frames mit benignen oder malignen Läsionen extrahiert und mit 56.036 glottischen Läsionsframes ergänzt. Im Testset: Präzision 92,1 % (95 %-KI 90,9–93,2), Recall 73,3 % (95 %-KI 69,8–76,6), Sensitivität für Malignität 94,3 % (95 %-KI 91,8–96,5).
Einschränkungen
33 läsionsfreie Videos zeigten eine mittlere frame-level Spezifität von 46,6 % und median 19 False-Positive-Detektionen pro Video. T1-Tumoren waren eine wichtige Limitation.
Schlussfolgerung der Autor:innen
Dies ist die erste Studie zu einem DL-Modell für die Echtzeit-Erkennung und Klassifikation benigner und maligner laryngealer und pharyngealer Läsionen. Die Leistung im Testset war vielversprechend.
Kernaussagen
- Präzision im Testset: 92,1 % (95 %-KI 90,9–93,2)
- Recall im Testset: 73,3 % (95 %-KI 69,8–76,6)
- Für Malignität: Sensitivität 94,3 % (95 %-KI 91,8–96,5)
Deutsche Zusammenfassung: KI-gestützt aus dem Original-Abstract erstellt; alle Zahlen wurden automatisch mit dem Original abgeglichen. Maßgeblich ist der Original-Abstract.
Kurz gefasst
Fachartikel in der Zeitschrift Cancers, online erschienen am 13.08.2026. Verfasst von Nathalie F van Rhee und 12 weiteren Autor:innen. Beteiligte Einrichtung: Department of Otorhinolaryngology and Head and Neck Surgery, University Medical Center Groningen, Hanzeplein 1, 9713 GZ Groningen, The Netherlands. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 32 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Dutch Cancer Society, KWF Dutch Cancer Society and Health Holland.
Abstract (Originalsprache)
Background/Objectives: Supraglottic and hypopharyngeal carcinomas are aggressive malignancies that are often diagnosed at advanced stages. Timely recognition of these malignancies is influenced by many factors, such as endoscope quality and experience. This study evaluated the potential of artificial intelligence (AI) to support real-time detection and classification of such lesions during flexible endoscopy in the outpatient clinic.
Methods
A previously developed deep learning (DL) algorithm was extended from a glottic lesion model to the unified localization and classification of glottic, supraglottic, and hypopharyngeal lesions during flexible endoscopy. Lesion frames were extracted from endoscopy videos obtained at two head and neck oncology centers and one secondary referral center between 2012 and 2024. These frames were annotated and labeled based on histopathological or clinically confirmed reference diagnoses. The primary outcome was the unified model’s performance in detection of lesions per frame. After training (70% of data), the positive predictive value (precision) and sensitivity (recall) of this model were calculated on an independent test set (30% of data), stratified by subsite and tumor (T-) classification. Secondly, the model’s binary classification performance (benign or malignant) was evaluated.
Results
From 490 supraglottic and hypopharyngeal endoscopy videos, 40,059 frames with a benign or malignant lesion were extracted and added to the 56,036 glottic lesion frames in the database, comprising 1336 lesion videos and 123 healthy control videos (total n = 1459). On the test set, the model achieved a detection precision of 92.1% (95% CI: 90.9–93.2) and a recall of 73.3% (95% CI: 69.8–76.6). Detection performance increased with higher T-stage. Among correctly detected lesion instances in a malignancy-dominated test set (65.3% of lesion instances), sensitivity for malignancy detection was 94.3% (95% CI: 91.8–96.5). Combined end-to-end performance for correct malignant lesion detection and classification was estimated at 69.1% (object-level recall 73.3% × classification sensitivity 94.3%). Evaluation of 33 lesion-free videos demonstrated a mean frame-level specificity of 46.6% and a median of 19 false positive detections per video.
Conclusions
This is the first study to report a DL model for real-time endoscopic detection and classification of benign and malignant laryngeal and pharyngeal lesions. The developed model showed promising lesion detection and cancer classification performance in the evaluated test set. T1 tumor detection remains an important limitation, particularly because early-stage detection is a primary aim of AI-assisted endoscopy. Further model testing is required in real-world lesion prevalence settings, where external validation and clinical usability should be investigated.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Otorhinolaryngology and Head and Neck Surgery, University Medical Center Groningen, Hanzeplein 1, 9713 GZ Groningen, The Netherlands.
- Department of Otorhinolaryngology and Head and Neck Surgery, Radboud University Medical Center, Geert Grooteplein Zuid 10, 6525 GA Nijmegen, The Netherlands.
- WSK Medical B.V., Science Park 301, 1098 XH Amsterdam, The Netherlands.
- Department of Otorhinolaryngology, Martini Hospital, Van Swietenplein 1, 9728 NT Groningen, The Netherlands.
- Department of Otorhinolaryngology and Head and Neck Surgery, Rijnstate Hospital, Wagnerlaan 55, 6815 AD Arnhem, The Netherlands.
ORCID-Profile: Nathalie F van Rhee, Celine M L H Wilmes, Inge Wegner, Gyorgy B Halmos, Boudewijn E C Plaat, David J Wellenstein
Förderung
- Dutch Cancer Society
- KWF Dutch Cancer Society and Health Holland
Zitieren
APA
van Rhee, N. F., Wilmes, C. M. L. H., Krijnen, H. K., Hofman, M., Woodburn, J., Schoonbeek, R. C., Wegner, I., Marres, H. A. M., San Giorgi, M. R. M., Halmos, G. B., van den Broek, G. B., Plaat, B. E. C., & Wellenstein, D. J. (2026). Real-Time-Capable Detection of Glottic, Supraglottic and Hypopharyngeal Lesions Using Artificial Intelligence During Flexible Endoscopy. Cancers, 18(16), 2609. https://doi.org/10.3390/cancers18162609
Vancouver
van Rhee NF, Wilmes CMLH, Krijnen HK, Hofman M, Woodburn J, Schoonbeek RC, et al. Real-Time-Capable Detection of Glottic, Supraglottic and Hypopharyngeal Lesions Using Artificial Intelligence During Flexible Endoscopy. Cancers. 2026;18(16):2609. doi:10.3390/cancers18162609
BibTeX
@article{vanRhee2026,
author = {van Rhee, Nathalie F and Wilmes, Celine M L H and Krijnen, Hidde K and Hofman, Mischa and Woodburn, Jonathan and Schoonbeek, Rosanne C and Wegner, Inge and Marres, Henri A M and San Giorgi, Michel R M and Halmos, Gyorgy B and van den Broek, Guido B and Plaat, Boudewijn E C and Wellenstein, David J},
title = {Real-Time-Capable Detection of Glottic, Supraglottic and Hypopharyngeal Lesions Using Artificial Intelligence During Flexible Endoscopy},
journal = {Cancers},
year = {2026},
volume = {18},
number = {16},
pages = {2609},
doi = {10.3390/cancers18162609}
}
EUVIMED