Multi-Path Fusion in SFCF-Net for Enhanced Multi-Frequency Electrical Impedance Tomography

Area of Science:

• Biomedical Engineering
• Medical Imaging
• Computational Biology

Background:

• Multi-frequency electrical impedance tomography (mfEIT) reconstructs tissue electrical properties using impedance spectral differences.
• Challenges exist in mfEIT for multi-class lesions due to frequency-dependent background tissue conductivity.

Purpose of the Study:

• To develop an advanced imaging algorithm, the spatial-frequency cross-fusion network (SFCF-Net), to overcome mfEIT limitations.
• To improve the differential imaging of multi-class lesions in mfEIT.

Main Methods:

• Proposed a novel spatial-frequency cross-fusion network (SFCF-Net) algorithm.
• Utilized multi-path structures and hyper-dense connections for spatial and frequency correlation capture.
• Cross-fusion of information for differential imaging of lesion targets.

Main Results:

• SFCF-Net demonstrated superior performance in lesion imaging and category discrimination compared to existing methods (weighted frequency-difference, U-Net, MMV-Net).
• Simulation and physical experiments validated the algorithm's effectiveness.
• Enhanced ability to simultaneously obtain structural and spectral information.

Conclusions:

• The SFCF-Net algorithm significantly improves the accuracy and reliability of mfEIT.
• This advancement opens new possibilities for clinical diagnostics and treatment monitoring.
• Improved mfEIT capabilities for complex tissue characterization.