Evaluation of performance of various high resolution x-ray imaging detectors
MetadataShow full item record
The basic concepts that are used to describe the performance of an x-ray detector regarding intrinsic image quality are sharpness, contrast, noise and signal-to-noise (SNR) ratio. The assessment of these basic concepts regarding detector's intrinsic performance can be done by applying frequency domain analysis that takes the signal and noise transfer characteristics of the detector into account and defines four intrinsic image quality metrics. These metrics are modulation transfer function (MTF), noise power spectrum (NPS), noise equivalent quanta (NEQ) and detective quantum efficiency (DQE). These metrics are widely used to characterize and compare the intrinsic performance of x-ray imaging detectors but they are not sufficient to indicate the performance of the detector in real clinical situation because they do not include several factors that come into play when the detector becomes a part of total imaging chain. This work presents the performance evaluation of various high-resolution x-ray imaging detectors in different ways by taking clinically relevant factors into account. The effect of system parameters such as focal-spot blur (which is the function of focal-spot size and magnification) and scatter on the intrinsic performance of a high-resolution detector is evaluated by using a set of generalized metrics. In general, these factors degrade the intrinsic performance of the detector. The study shows that the scatter affects low range spatial frequencies more severely compared to mid and high range frequencies which are affected more by the focal-spot blur. This indicates the possibility of a trade-off between the system parameters, based on the frequency spectrum of an object of interest, to improve its detectability. It is also shown that magnification can be used to improve the visualization of smaller details in the image taken with lower resolution detectors but higher magnification values are not possible with high-resolution detectors as they are very sensitive to the focal-spot. A new metric, relative object detectability (ROD), is introduced to quantify the relative performance of two detectors for a specified imaging task. The ROD is defined as the ratio of the integral of detective quantum efficiency (DQE) of a detector weighted by the square of the modulus of the Fourier Transformation (or frequency spectrum) of a simulated object to the integral of the DQE of another detector weighted by the square of the modulus of the Fourier Transformation of the same object. It is shown that the high resolution detector performs better than the low resolution detector regarding detectability of small objects and as the object size decreases the performance of a high resolution detector increases compared to a low resolution detector in a marked and quantifiable way. The relative performance evaluation does not show dramatic change for larger objects. It is also found that the size rather than the specific material have a dramatic effect on the ROD. Further, the metric of generalized relative object detectability (G-ROD) is introduced to include the effect of system parameters on the relative performance evaluation of a high resolution detector compared to a lower resolution detector. The effect of system parameters is taken into account by replacing the DQE with the GDEQ of detectors for a given set of parameters in the formulation of the ROD. The G-ROD shows trend similar to the ROD but it is found to be quantitatively less than the ROD for the same object because the performance of the high resolution detector is relatively degraded more compared to the low resolution detector when the system parameters are taken into account. The decrease in G-ROD is found to be more as the focal-spot blur increases either with the size of focal-spot or with the magnification. The ROD and the G-ROD both use simulated objects; therefore, a metric of generalized measured relative object detectability (GM-ROD) was introduced to make the relative performance evaluation more realistic. The GM-ROD is a fully measured metric because it requires the NPS of the detectors and the images of the object taken with detectors whose relative performance evaluation is to be done. It could be more comprehensive as it calculates the signal-to-noise ratio at each spatial frequency that includes both the real and aliased components of signal and noise before integrating up to the Nyquist of the detectors. Finally, a study is done to investigate the efficacy of an anti-scatter grid when used with a high-resolution detector by determining the image contrast and the CNR and comparing these with the case where a state-of-the-art imaging detector is used which has lower resolution. We observe that the contrast improves when the grid is used with the high resolution detector but the increase in the CNR is not so significant compared to the case of the lower resolution detector. Assuming the quantum noise increases similarly for both detectors when the grid is used (due to reduced photon fluence), it is the substantial increase in the grid's fixed pattern noise which degrades the CNR in the case of the high resolution detector. It may be possible to ameliorate this problem either with a grid of improved design or with additional image processing corrections to minimize the structured grid-line artifacts and to this end a new method to correct for grid-line artifacts was demonstrated.