High-resolution imaging system for intelligent total body scanner for early detection of melanoma

Melanoma's increasing global incidence highlights the need for high-throughput, automatable early detection methods like total body scanners integrated with computer-aided diagnosis. This study presents a High-Resolution Imaging Module for non-contact dermoscopy, designed to acquire and process raw image data into high-resolution dermoscopic images using a focus stacking approach. By merging multiple images, each with limited depth of field, into a single hyperfocus image, this method ensures complete focus across all skin topographies. The system utilizes an electrically tunable liquid lens for rapid capture of differently focused images, supported by algorithms for image alignment, focus measurement, and fusion. Deep learning-based super-resolution techniques were applied to further enhance image quality, and an AI classification model was trained for lesion assessment. The developed system achieves a resolution of 27.84 μm and captures focus stacks at 50 FPS, ensuring rapid acquisition and patient comfort. Hyperfocus images produced demonstrate high clarity and detail, significantly improving diagnostic reliability. However, the application of super-resolution techniques necessitates careful consideration to maintain the authenticity of raw data. This work demonstrates the potential of integrating advanced imaging technologies into total body scanners, offering a non-invasive, high-accuracy solution for early melanoma detection in clinical practice.