Related Experiment Video
Updated: Dec 15, 2025
Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
Published on: July 11, 2025
Computer-aided classification of suspicious pigmented lesions using wide-field images
Judith S Birkenfeld1, Jason M Tucker-Schwartz2, Luis R Soenksen3
1Research Laboratory of Electronics, Massachusetts Institute of Technology, 77 Massachusetts Ave, Cambridge, MA 02139, USA; MIT linQ, Institute for Medical Engineering and Science, Massachusetts Institute of Technology, 77 Massachusetts Ave., Cambridge, MA 02139, USA; Brigham and Women's Hospital - Harvard Medical School, 75 Francis St, Boston, MA 02115, United States; Massachusetts General Hospital - Harvard Medical School, 55 Fruit St, Boston, MA 02114, United States.
This study introduces a computer-aided classification system using wide-field images for rapid identification of suspicious pigmented lesions. The system aids primary care in melanoma screening, improving early detection accuracy.
Area of Science:
- Dermatology
- Medical Imaging
- Artificial Intelligence
Background:
- Melanoma early detection relies on visual examinations, which vary in accuracy.
- Current computer-aided diagnosis tools use controlled single-lesion images, limiting real-world application.
- Wide-field imaging offers a more practical approach for broad skin lesion assessment.
Purpose of the Study:
- To develop and evaluate a computer-aided classifier for rapid identification of suspicious pigmented lesions.
- To support primary care physicians in early melanoma detection.
- To assess the utility of wide-field photography in conjunction with AI for skin lesion screening.
Main Methods:
- Recruited 133 patients with diverse skin lesions.
- Collected wide-field images of lesions under natural illumination using a consumer-grade camera.
- Extracted 1759 pigmented lesions and developed a machine learning classifier to assign a suspiciousness score.
Main Results:
- The system achieved 100% sensitivity for confirmed suspicious pigmented lesions (SPL_A).
- Sensitivity for unconfirmed suspicious pigmented lesions (SPL_B) was 83.2%, and for non-suspicious lesions was 72.1%.
- Overall accuracy was 75.9%, with a suspiciousness score aligned with clinical practice.
Conclusions:
- Wide-field photography and AI classification can effectively distinguish suspicious from non-suspicious pigmented lesions.
- This approach shows potential for assessing lesion severity and supporting population-level skin screenings.
- The system offers a valuable tool for enhancing early melanoma detection in primary care settings.
