Multimodal Deep Learning for Benign vs Malignant Eyelid Lesion Classification Using Optical Coherence…

AI / informaticsOphthalmic plastic and reconstructive surgery· May 2026

Multimodal Deep Learning for Benign vs Malignant Eyelid Lesion Classification Using Optical Coherence Tomography and Clinical Images.

Weronika Jakubowska, Clément Playout, Renaud Duval, Evan Kalin-Hajdu

OculoplasticsAI / informaticsDiagnosisQ2IF 1.3

Multimodal AI combining optical coherence tomography and photos best distinguishes benign from malignant eyelid lesions

Multimodal Deep Learning for Benign vs Malignant Eyelid Lesion Classification Using Optical Coherence Tomography and Clinical Images.

Weronika Jakubowska … Evan Kalin-Hajdu

Ophthalmic plastic and reconstructive surgery · May 2026 · Q2 in Ophthalmology

Purpose

To evaluate the performance of deep learning models using optical coherence tomography (optical coherence tomography) volumes, clinical photographs, and their multimodal fusion to classify eyelid lesions as benign or malignant, using histopathology as the reference standard.

Methods

Prospective cohort study conducted between January 2023 and January 2025 at a single tertiary academic oculofacial plastic surgery center.

n = 71 lesions

Results

92.1%

combining eye scans and photos correctly told benign from cancerous lesions apart nearly every time

n = 71 lesions

More results

Of 71 lesions, 52% were benign and 48% malignant.

The optical coherence tomography model achieved 73.9% accuracy, 74.4% sensitivity, and an area under the receiver operating characteristic curve of 82.5%.

More results

The photograph model reached 82.3% accuracy, 89.6% sensitivity, and an area under the receiver operating characteristic curve of 91.7%.

“

Conclusion · 1 of 2

This study demonstrates the feasibility of analyzing optical coherence tomography volumes using deep learning, with diagnostic accuracy improved through multimodal integration with clinical photography.

Conclusion · 2 of 2

These results suggest that multimodal artificial intelligence may serve as a scalable, noninvasive, and physician-independent tool to distinguish benign from malignant periocular lesions and streamline patient triage.

Multimodal Deep Learning for Benign vs Malignant Eyelid Lesion Classification Using Optical Coherence Tomography and Clinical Images.

Multimodal Deep Learning for Benign vs Malignant Eyelid Lesion Classification Using Optical Coherence Tomography and Clinical Images.

Related papers