Related Experiment Video
Updated: Jun 16, 2025
10:35
Multimodal Imaging and Spectroscopy Fiber-bundle Microendoscopy Platform for Non-invasive, In Vivo Tissue Analysis
Published on: October 17, 2016
7.8K
Classification of melanocytic lesions using direct illumination multispectral imaging
Elisabeth Victoria Goessinger1, Paul-Gerald Dittrich2,3,4, Philipp Nöcker5
1Department of Dermatology, University Hospital of Basel, Basel, Switzerland.
Scientific Reports
|August 16, 2024
Summary
Direct light coupling in multispectral imaging significantly improves melanoma detection accuracy. This non-invasive technique enhances classification of melanocytic lesions, offering a promising advancement for early skin cancer diagnosis.
Area of Science:
- Dermatology
- Medical Imaging
- Oncology
Background:
- Melanoma incidence and mortality rates are increasing, highlighting the need for effective early detection methods.
- Multispectral imaging (MSI) is a non-invasive technique for skin cancer detection, but current methods using incident light are limited by surface reflectance.
- Deeper light penetration is needed to analyze subsurface tissue characteristics for improved lesion classification.
Purpose of the Study:
- To evaluate the efficacy of multispectral imaging with direct light coupling for enhanced dignity classification of melanocytic lesions.
- To compare the diagnostic performance of direct light coupling MSI against incident light MSI and standard RGB imaging.
- To assess the potential of a novel illumination method for improving skin cancer detection accuracy.
Main Methods:
- A pilot study included 27 suspicious pigmented lesions from 23 patients, including melanomas, dysplastic nevi, and melanocytic nevi.
- Lesions were imaged using a prototype snapshot mosaic multispectral camera with both incident and direct light illumination prior to excision.
- A pre-trained multispectral image analysis model performed dignity classification, with results compared to histopathological findings.
Main Results:
- Direct light coupling MSI achieved a sensitivity of 100% and specificity of 82.4%, outperforming incident light MSI (83.3% sensitivity, 58.8% specificity).
- Convolutional neural network classification using RGB images showed lower performance (83.3% sensitivity, 61.5% specificity) compared to direct light MSI.
- Histopathological examination served as the gold standard for comparison.
Conclusions:
- Direct light coupling multispectral imaging significantly enhances the accuracy of dignity classification for melanocytic lesions.
- This novel illumination technique shows potential for improving multispectral imaging applications in skin cancer detection.
- Further large-scale studies are warranted to validate the findings and the camera prototype.
