Classical methods for X-ray computed tomography are based on the assumptionthat the X-ray source intensity is known, but in practice, the intensity ismeasured and hence uncertain. Under normal operating conditions, when theexposure time is sufficiently high, this kind of uncertainty typically has anegligible effect on the reconstruction quality. However, in time- ordose-limited applications such as dynamic CT, this uncertainty may cause severeand systematic artifacts known as ring artifacts. By carefully modeling themeasurement process and by taking uncertainties into account, we derive a newconvex model that leads to improved reconstructions despite poor qualitymeasurements. We demonstrate the effectiveness of the methodology based onsimulated and real data sets.
展开▼