A nonlinear lag correction algorithm for a-Si flat-panel x-ray detectors

Area of Science:

• Medical Imaging
• Radiological Physics
• Detector Technology

Background:

• Detector lag in a-Si flat-panel (FP) detectors causes shading artifacts in cone-beam computed tomography (CBCT).
• Existing linear, time-invariant (LTI) correction models are sensitive to exposure intensity and IRF measurement techniques, leaving residual artifacts.

Purpose of the Study:

• Develop and evaluate a novel nonlinear, time-invariant (non-LTI) lag correction method.
• Address the exposure dependencies of lag and improve CBCT image quality.

Main Methods:

• Implemented a multiexponential LTI model for lag correction.
• Developed the nonlinear consistent stored charge (NLCSC) method, which estimates stored charge without requiring specific semiconductor parameters.
• Applied correction algorithms to step-response data and CT scans of pelvic and head phantoms using a Varian 4030CB a-Si FP detector.

Main Results:

• NLCSC reduced 1st and 50th frame residual lags to <0.29% and <0.0052%, respectively, compared to LTI corrections (<1.4% and <0.48%).
• NLCSC yielded lower average errors in CT reconstructions (11 HU for pelvic, 3 HU for head) than LTI (14-19 HU, 2-11 HU) and intensity-weighted methods (15 HU, 9 HU).
• NLCSC consistently produced the smallest maximum ROI error and superior artifact removal.

Conclusions:

• The NLCSC lag algorithm effectively corrects for exposure-dependent lag.
• NLCSC provides superior image improvement, particularly for pelvic phantom reconstructions, and comparable results to the best LTI methods for head phantom reconstructions.
• NLCSC better removes blurred ring artifacts compared to LTI corrections.