Skin tumor three-dimensional electrical impedance tomography method and device and storage medium
By utilizing nonlinear techniques of planar electrodes and spatial virtual electrode mapping in skin tumor detection, combined with a deep learning model, the applicability and reconstruction stability issues of traditional 3D-EIT in skin tumor detection across multiple sites are solved, achieving high-precision three-dimensional conductivity distribution imaging, suitable for non-invasive detection and invasion depth assessment of skin tumors.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-02
- Publication Date
- 2026-04-10
AI Technical Summary
Existing 3D-EIT technology for skin tumor detection suffers from several problems, including sensors that are difficult to adapt to the needs of multi-site detection, difficulty in obtaining stable and reliable imaging results from three-dimensional reconstruction, and low measurement stability and reconstruction accuracy.
By using planar electrodes to collect boundary voltage signals on the skin surface, and combining this with a feedforward neural network to achieve a nonlinear mapping from the planar electrode domain to the spatial virtual electrode domain, an equivalent voltage of the spatial virtual electrode is generated. This voltage is then mapped to a two-dimensional voltage feature map through bilinear interpolation, input into a three-dimensional electrical impedance imaging reconstruction network, and a U-Net network with a sliding window Transformer is used for depth matching inversion. This constructs a finite element model of skin tissue and trains it on a simulation dataset.
It achieves enhanced response to perturbations in deep tissue conductivity without changing the measurement boundaries, obtaining stable, non-invasive, and highly accurate three-dimensional conductivity distribution, suitable for non-invasive detection and invasion depth assessment of skin tumors, and improves the robustness of measurement and reconstruction accuracy.
Smart Images

Figure CN121817848A_ABST