EUVIMEDEUVIMED entdecken
ValidierungsstudieOpen Access

A Multimodal Approach for Deep‐Learning Classification of Vocal Fold Pathologies in Stroboscopy

Sruthi Surapaneni, Rachel B Kutler iD, Sean A Setzen iD, Yeo Eun Kim iD, Peter Yao, Sana H Siddiqui, Michael J Pitman iD, Lucian Sulica iD, Olivier Elemento, Pegah Khosravi, Anaïs Rameau iD

The Laryngoscope · 2026

DOI: 10.1002/lary.70355 ↗

Kurz gefasst

Validierungsstudie in der Zeitschrift The Laryngoscope, 2026. Verfasst von Sruthi Surapaneni und 10 weiteren Autor:innen. Beteiligte Einrichtung: Department of Otolaryngology-Head and Neck Surgery, Weill Cornell Medical College, Sean Parker Institute for the Voice, New York, New York, USA. Hauptthemen laut MEDLINE: Deep Learning, Vocal Cords, Stroboscopy, Vocal Cord Paralysis. Die Arbeit nennt 27 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: NIA NIH HHS, NIH HHS, ODCDC CDC HHS.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Objective

To develop and validate a multimodal deep‐learning classifier trained on stroboscopic image, voice, and clinicodemographic data, differentiating between three different vocal fold (VF) states: healthy (HVF), unilateral paralysis (UVFP), and VF lesions, including benign and malignant pathologies.

Methods

Patients with UVFP ( n = 54), VF lesions ( n = 42), and HVF ( n = 41) were retrospectively identified. Image frames and voice samples were extracted from stroboscopic videos. Clinicodemographic variables were collected from the electronic health record. Patient‐level data was independently divided into training (80%) and testing (20%). Visual features were extracted using a transformer DINOv2 and acoustic features were extracted using Librosa. All three feature modalities were combined using a custom multilayer perceptron. Unimodality models using only image or only voice data were trained for comparison. Accuracy and F 1 scores were used to validate the models.

Results

On a hold‐out test set, the multimodal classifier demonstrated stronger performance (76.9% accuracy) compared to the image classifier (61.5% accuracy) and audio classifier (65.4% accuracy). On an external dataset, the multimodal classifier accuracy dropped to 45%, though still an improvement compared to accuracies of 42% and 31% for the video‐only and audio‐only modalities, respectively.

Conclusions

In this proof‐of‐concept study, we successfully developed a multimodal dataset and classifier for VF pathology, demonstrating the potential of combining stroboscopic frames, voice and text data. The multimodal classifier achieved higher accuracy than the image‐only model and audio‐only models. Future models should validate these findings on larger datasets. Level of Evidence 4

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Deep LearningVocal CordsStroboscopyVocal Cord ParalysisLaryngeal DiseasesHumansRetrospective StudiesMaleFemaleMiddle AgedAdultAgedVideo Recording

Stichwörter der Autor:innen

artificial intelligencemultimodaltransformer

Autor:innen und Institutionen

  • Department of Otolaryngology-Head and Neck Surgery, Weill Cornell Medical College, Sean Parker Institute for the Voice, New York, New York, USA.
  • Michigan State University College of Human Medicine, Grand Rapids, Michigan, USA.
  • The Center for Voice and Swallowing, Department of Otolaryngology-Head and Neck Surgery, Columbia University Irving Medical Center, New York, New York, USA.
  • Department of Otolaryngology-Head and Neck Surgery, St. Luke's University Health Network, Bethlehem, Pennsylvania, USA.
  • Englander Institute for Precision Medicine, Weill Cornell Medicine, New York, New York, USA.
  • Department of Clinical Sciences, College of Medicine, University of Central Florida, Orlando, Florida, USA.

ORCID-Profile: Rachel B Kutler, Sean A Setzen, Yeo Eun Kim, Michael J Pitman, Lucian Sulica, Anaïs Rameau

Förderung

  • NIA NIH HHS
  • NIH HHS
  • ODCDC CDC HHS

Zitieren

APA

Surapaneni, S., Kutler, R. B., Setzen, S. A., Kim, Y. E., Yao, P., Siddiqui, S. H., Pitman, M. J., Sulica, L., Elemento, O., Khosravi, P., & Rameau, A. (2026). A Multimodal Approach for Deep‐Learning Classification of Vocal Fold Pathologies in Stroboscopy. The Laryngoscope. https://doi.org/10.1002/lary.70355

Vancouver

Surapaneni S, Kutler RB, Setzen SA, Kim YE, Yao P, Siddiqui SH, et al. A Multimodal Approach for Deep‐Learning Classification of Vocal Fold Pathologies in Stroboscopy. The Laryngoscope. 2026. doi:10.1002/lary.70355

BibTeX
@article{Surapaneni2026,
  author = {Surapaneni, Sruthi and Kutler, Rachel B and Setzen, Sean A and Kim, Yeo Eun and Yao, Peter and Siddiqui, Sana H and Pitman, Michael J and Sulica, Lucian and Elemento, Olivier and Khosravi, Pegah and Rameau, Anaïs},
  title = {A Multimodal Approach for Deep‐Learning Classification of Vocal Fold Pathologies in Stroboscopy},
  journal = {The Laryngoscope},
  year = {2026},
  doi = {10.1002/lary.70355}
}
Kostenlos starten →