Resolution Guarantees in Electrical Impedance Tomography
This work addresses the reliability problem in EIT for medical imaging and non-destructive testing by offering theoretical guarantees for anomaly detection.
The paper provides rigorous resolution guarantees for electrical impedance tomography (EIT) in the presence of systematic and random errors, deriving a constructive criterion to determine if a desired resolution is achievable given measurement setup constraints.
Electrical impedance tomography (EIT) uses current-voltage measurements on the surface of an imaging subject to detect conductivity changes or anomalies. EIT is a promising new technique with great potential in medical imaging and non-destructive testing. However, in many applications, EIT suffers from inconsistent reliability due to its enormous sensitivity to modeling and measurement errors. In this work we show that rigorous resolution guarantees are possible within a realistic EIT measurement setting including systematic and random errors. We derive a constructive criterion to decide whether a desired resolution can be achieved in a given measurement setup. Our result covers the detection of anomalies of a known minimal contrast using noisy measurements on a number of electrodes attached to a subject with imprecisely known background conductivity.