A study to improve the image quality in low-dose computed tomography (SPECT) using filtration

Background: The output of the X-ray tube used in computed tomography (CT) provides a spectrum of photon energies. Low-energy photons are preferentially absorbed in tissue; the beam spectrum shifts toward the higher energy end as it passes through more tissue, thereby changing its effective attenuation coefficient and producing a variety of artifacts (beam-hardening effects) in images. Filtering of the beam may be used to remove low-energy photon component. The accuracy of attenuation coefficient calculation by bilinear model depends highly upon accuracy of Hounsfield units. Therefore, we have made an attempt to minimize the beam-hardening effects using additional copper filter in the X-ray beam. The quantitative evaluation were made to see the effect of additional filters on resulting CT images.

Materials and methods: This study was performed on dual-head SPECT (HAWKEYE 4, GE Healthcare) with low-dose CT which acquires images at peak voltages of 120/140 kV and a tube current of 2.5 mA. For the evaluation of image quality, we used CT QA Phantom (PHILIPS) having six different density pins of Water, Polyethylene, Nylon (Aculon), Lexan, Acrylic (Perspex) and Teflon. The axial images were acquired using copper filters of various thicknesses ranging from 1 to 5 mm in steps of 1 mm. The copper filter was designed in such a manner that it fits exactly on the collimator cover of CT X-ray tube. Appropriate fixation of the copper filter was ensured before starting the image acquisition. As our intention was only to see the effect of beam hardening on the attenuation map, no SPECT study was performed. First set of images was acquired without putting any filter into the beam. Then, successively, filters of different thicknesses were placed into the beam and calibration of the CT scanner was performed before acquiring the images. The X-ray tube parameters were kept the same as that of unfiltered X-ray beam. All the acquired image sets were displayed using Xeleris 2 (GE Healthcare) on a high-resolution monitor. Moreover, Jaszak's SPECT Phantom after removing the spheres was used to see the different contrast intensities by inserting the different contrast materials of iodine and bismuth in water as background media. Images were analyzed for visibility, spatial resolution and contrast.

Results: Successive improvement in the image quality was noticed when we increased the filter thickness from 1 to 3 mm. The images acquired with 3-mm filter appeared almost with no artifacts and were visibly sharper. Lower energy photons from X-ray beam cause a number of artifacts, especially at bone-tissue interfaces. Additional filtrations removed lower energy photons and improved the image quality. Degradation in the image quality was noticed when we increased the filter thickness further to 4 and 5 mm. This degradation in image quality happened due to reduced photon flux of the resulting X-ray beam, causing high statistical noise. The spatial resolution for image matrix of 512 × 512 was found to be 1.29, 1.07, 0.64 and 0.54 mm for without filter, with 1, 2 and 3 mm filters, respectively. The image quality was further analyzed for signal-to-noise ratio (SNR). It was found to be 1.72, 1.78, 1.98 and 1.99 for open, with 1, 2 and 3 mm filters respectively. This shows that 3-mm filter results in an improvement of 15.7% in SNR.

Conclusion: On the basis of this study, we could conclude that use of 3-mm copper filter in the X-ray beam is optimal for removing the artifacts without causing any significant reduction in the photon flux of the resulting X-ray beam. We also propose that as artifacts have been removed from the images, the value of Hounsfield units will be more accurate and hence the value of attenuation coefficients lead to better contrast and visualization of SPECT images.