Invertibility of the dual energy x-ray data transform

Area of Science:

• Medical Physics
• Image Reconstruction
• X-ray Imaging

Background:

• The Alvarez-Macovski method reconstructs x-ray energy-dependent information using basis functions.
• This method expands the attenuation coefficient into a linear combination of energy-dependent functions.
• The coefficients of these functions encapsulate all energy-dependent information.

Purpose of the Study:

• To investigate factors influencing the invertibility of the L(A) transformation in dual energy x-ray imaging.
• To analyze the dual energy transformation using a two-function basis set and two spectral measurements.

Main Methods:

• Applied a general invertibility theorem requiring Jacobian analysis within the input domain.
• Proved general conditions for invertibility and analyzed noise variance.
• Investigated the relationship between zero Jacobian values and ambiguous data points.

Main Results:

• Invertibility is independent of the chosen valid basis set.
• Non-overlapping spectra (e.g., photon counting detectors) ensure invertibility.
• Spectra from layered detectors approach non-invertibility with increasing object thickness.
• Zero Jacobian values form curves in coefficient space, indicating ambiguous data points.
• Iterative algorithms exhibit convergence errors near zero Jacobian curves.

Conclusions:

• The invertibility of dual energy x-ray systems is fundamentally determined by the presence of zero Jacobian values in the L(A) transformation.
• Understanding these zero Jacobian values is crucial for accurate image reconstruction in dual energy systems.