Multi-frequency symmetry difference electrical impedance tomography with machine learning for human stroke diagnosis.

Area of Science:

• Medical imaging
• Biomedical engineering
• Machine learning in healthcare

Background:

• Electrical impedance tomography (EIT) is a non-invasive imaging technique.
• Detecting stroke etiology, such as hemorrhage or clot, is crucial for timely treatment.
• Existing methods may have limitations in accurately identifying stroke types.

Purpose of the Study:

• To evaluate the efficacy of Multi-frequency Symmetry Difference Electrical Impedance Tomography (MFSD-EIT) for stroke lesion detection.
• To assess the application of machine learning algorithms in conjunction with MFSD-EIT for stroke etiology identification.
• To differentiate between normal, hemorrhage, and clot conditions using MFSD-EIT data.

Main Methods:

• Development of anatomically realistic finite element models of the head based on patient images.
• Generation of EIT data across a frequency range (5 Hz-100 Hz) with simulated lesions (bleed and clot).
• Application of a quantitative symmetry metric and machine learning (SVM classification) to reconstructed conductivity maps.

Main Results:

• MFSD-EIT achieved an average accuracy of 85% in differentiating human bleed from clot using SVM classification.
• An accuracy of 77% was obtained when differentiating normal from stroke conditions in human data.
• The algorithm successfully detected and identified simulated and human lesion data.

Conclusions:

• MFSD-EIT combined with machine learning offers a novel and promising approach for detecting and identifying perturbations in static scenes.
• The technique demonstrates potential for feasible translation to clinical application in human stroke patients.
• This method shows promise for robust lesion detection and etiological identification in challenging conditions.