Non-contact dermoscopy offers significant advantages over traditional contact methods, including preservation of natural lesion structure, reduced cross-contamination risk, and enhanced capabilities for automated dermatological assessment.
However, the absence of optical coupling introduces challenges, including increased light scattering, lower image contrast, and reduced visibility of fine diagnostic structures. This paper explores the concept of “long-distance dermoscopy”, where imaging occurs from significantly greater distances (25-40 cm), introducing additional resolution constraints. We present a systematic analysis of the spatial resolution required to maintain diagnostic accuracy in long-distance dermoscopy and propose a novel imaging approach incorporating liquid lens technology, focus stacking, and AI-based super-resolution. Our results demonstrate that a resolution of at least 25 μm/pixel is necessary to visualize critical dermoscopic structures.
Quantitative resolution analysis using the USAF 1951 test chart established that our system achieves 22.63 lp/mm (equivalent to 22.1 μm resolution). Comparative evaluation of 156 lesions imaged with both contact dermoscopy and our non-contact system demonstrated that non- contact imaging can achieve comparable diagnostic performance, particularly when enhanced with super-resolution techniques. These findings support the potential of long-distance dermoscopy as a viable imaging modality for automated lesion screening and teledermatology applications, bridging the gap between non-contact imaging and traditional dermoscopy.
Get the full work atiToBoS Whitepaper.