Enhanced dynamic range x-ray imaging

Abstract

X-ray images can suffer from excess contrast. Often, image exposure is chosen to visually optimize the region of interest, but at the expense of over- and underexposed regions elsewhere in the image. When image values are interpreted quantitatively as projected absorption, both over- and underexposure leads to the loss of quantitative information. We propose to combine multiple exposures into a composite that uses only pixels from those exposures in which they are neither under- nor overexposed. The composite image is created in analogy to visible-light high dynamic range photography. We present the mathematical framework for the recovery of absorbance from such composite images and demonstrate the method with biological and non-biological samples. We also show with an aluminum step-wedge that accurate recovery of step thickness from the absorbance values is possible, thereby highlighting the quantitative nature of the presented method. Due to the higher amount of detail encoded in an enhanced dynamic range x-ray image, we expect that the number of retaken images can be reduced, and patient exposure overall reduced. We also envision that the method can improve dual energy absorptiometry and even computed tomography by reducing the number of low-exposure ("photon-starved") projections.

Keywords: Contrast; Dynamic range; Noise; Saturation; X-ray absorption.