Receiver operating characteristic (ROC) analysis of images reconstructed with iterative expectation maximization algorithms

Optimization of iterative reconstruction parameters with attenuation correction, scatter correction and resolution recovery in myocardial perfusion SPECT/CT

OBJECTIVE

The aim of this study was to characterize the optimal reconstruction parameters for ordered-subset expectation maximization (OSEM) with attenuation correction, scatter correction, and depth-dependent resolution recovery (OSEMACSCRR). We assessed the optimal parameters for OSEMACSCRR in an anthropomorphic torso phantom study, and evaluated the validity of the reconstruction parameters in the groups of normal volunteers and patients with abnormal perfusion.

METHODS

Images of the anthropomorphic torso phantom, 9 normal volunteers and 7 patients undergoing myocardial perfusion SPECT were acquired with a SPECT/CT scanner. SPECT data comprised a 64×64 matrix with an acquisition pixel size of 6.6 mm. A normalized mean square error (NMSE) of the phantom image was calculated to determine both optimal OSEM update and a full width at half maximum (FWHM) of Gaussian filter. We validated the myocardial count, contrast and noise characteristic for clinical subjects derived from OSEMACSCRR processing. OSEM with depth-dependent resolution recovery (OSEMRR) and filtered back projection (FBP) were simultaneously performed to compare OSEMACSCRR.

RESULTS

The combination of OSEMACSCRR with 90-120 OSEM updates and Gaussian filter with 13.2-14.85 mm FWHM yielded low NMSE value in the phantom study. When we used OSEMACSCRR with 120 updates and Gaussian filter with 13.2 mm FWHM in the normal volunteers, myocardial contrast showed significantly higher value than that derived from 120 updates and 14.85 mm FWHM. OSEMACSCRR with the combination of 90-120 OSEM updates and 14.85 mm FWHM produced lowest % root mean square (RMS) noise. Regarding the defect contrast of patients with abnormal perfusion, OSEMACSCRR with the combination of 90-120 OSEM updates and 13.2 mm FWHM produced significantly higher value than that derived from 90-120 OSEM updates and 14.85 mm FWHM. OSEMACSCRR was superior to FBP for the % RMS noise (8.52±1.08 vs. 9.55±1.71, p=0.02) and defect contrast (0.368±0.061 vs. 0.327±0.052, p=0.01), respectively.

CONCLUSIONS

Clinically optimized the number of OSEM updates and FWHM of Gaussian filter were (1) 120 updates and 13.2 mm, and (2) 90-120 updates and 14.85 mm on the OSEMACSCRR processing, respectively. Further assessment may be required to determine the optimal iterative reconstruction parameters in a larger patient population.

Performance of a semiconductor SPECT system: comparison with a conventional Anger-type SPECT instrument

Objective

The performance of a new single photon emission computed tomography (SPECT) scanner with a cadmium-zinc-telluride (CZT) solid-state semiconductor detector (Discovery NM 530c; D530c) was evaluated and compared to a conventional Anger-type SPECT with a dual-detector camera (Infinia).

Methods

Three different phantom studies were performed. Full width at half maximum (FWHM) was measured using line sources placed at different locations in a cylindrical phantom. Uniformity was measured using cylindrical phantoms with 3 different diameters (80, 120, and 160 mm). Spatial resolution was evaluated using hot-rod phantoms of various diameters (5, 9, 13, 16, and 20 mm). Three different myocardial phantom studies were also performed, acquiring projection data with and without defects, and evaluating the interference of liver and gallbladder radioactivity. In a clinical study, the D530c employed list-mode raw data acquisition with electrocardiogram (ECG)-gated acquisition over a 10-min period. From the 10-min projection data, 1-, 3-, 5-, 7- and 10-min SPECT images were reconstructed.

Results

The FWHM of the D503c was 1.73–3.48 mm (without water) and 3.88–6.64 mm (with water), whereas the FWHM of the Infinia was 8.17–12.63 mm (without water) and 15.48–16.28 mm (with water). Non-uniformity was larger for the D530c than for the Infinia. Truncation artifacts were also observed with the D530c in a Φ160 mm phantom. The contrast ratio, as defined by myocardial defect/non-defect ratio, was better for the D530c than for the Infinia, and the influence from liver and gallbladder radioactivities was less. Quantitative gated SPECT (QGS) software demonstrated significant differences between data captured over a 10-min period, relative to those acquired over periods of <5 min; there was no difference between ejection fractions calculated using data capture for periods ≥5 min (p < 0.05).

Conclusions

The D530c is superior to the Infinia, with regard to both spatial resolution and sensitivity. In this study, these advantages were confirmed by the myocardial phantom and in a clinical setting, using the QGS software.

Evaluation of the number of SPECT projections in the ordered subsets-expectation maximization image reconstruction method

Filtered back projection (FBP) method, maximum likelihood-expectation maximization(ML-EM) method, and ordered subsets-expectation maximization (OS-EM) method are currently used for reconstruction of SPECT images in clinical studies. In the ML-EM method, images of good quality can be reconstructed even with a small sampling number of projection data, when compared with FBP. Shorter acquisition time and less radionuclide dose are preferable in the clinical setting if image quality is the same. In this study, we attempted to find optimal conditions for reconstruction of OS-EM images with commonly used sampling numbers of 30, 60 and 120 (step angles: 12 degrees, 6 degrees, and 3 degrees, respectively), with acquisition counts/projection of 30, 60, 120 and 240 each. We adjusted the pixel counts of reconstructed images to be constant, by setting combination of sampling number and counts/projection (120 sampling number for 30 counts/projection, 60 for 60, and 30 for 120). Among the 3 acquisition conditions, the small sampling number of 30 had large acquisition counts per direction, resulting in low signal to noise ratio. Under this condition, the resolution was slightly low, but the uniformity of images was high. The combination of OS-EM and smaller sampling projection number may be clinically useful with reduction of the examination time, which is also beneficial to reduce dead time for gamma-camera rotation.

A study on attenuation correction using Tc-99m external TCT source in Tc-99m GSA liver SPECT

PURPOSE

In attenuation correction of ECT images by transmission CT (TCT) with an external 99mTc gamma-ray source, simultaneous TCT/ECT data acquisition is difficult, when the same radionuclide such as 99mTc-tetrofosmin or 99mTc-GSA is used as the tracer. In this case, TCT is usually acquired before administration of the tracer, and ECT is acquired separately after the tracer injection. However, misregistration may occur between the TCT and ECT images, and the repetition of examinations add to the mental and physical stress of the patients. In this study, to eliminate this problem, we evaluated whether attenuation correction of ECT images can be achieved by acquiring TCT and ECT simultaneously, then acquiring ECT alone, and preparing an attenuation map by subtracting the latter from the former using 99mTc-GSA liver ECT.

METHOD

The ECT system used was a three-head gamma camera equipped with one cardiac fan beam collimator and two parallel beam collimators. External gamma-ray source for TCT of 99mTc was 740 MBq, and ECT of 99mTc-GSA was 185 MBq. First, pure TCT data were acquired for the original TCT-map, then, ECT/TCT data were acquired for the subtracted TCT-map, and finally, pure ECT data were acquired. The subtracted attenuation map was produced by subtracting the pure ECT image from the TCT/ECT image, and attenuation correction of the ECT image was done using both this subtracted TCT map and attenuation map from pure TCT. These two attenuation corrected images and non-corrected images were compared. Hot rods phantom, a liver phantom with a defect, and 10 patients were evaluated.

RESULTS

Attenuation corrected ECT values using the subtraction attenuation map showed an error of about 5% underestimation compared with ECT values of the images corrected by original attenuation map at the defect in the liver phantom. A good correlation of y = 22.65 + 1.06x, r = 0.958 was observed also in clinical evaluation.

CONCLUSION

By means of the method proposed in this study, it is possible to perform simultaneous TCT/ECT data acquisition for attenuation correction using Tc-99m external source in Tc-99m GSA liver SPECT. Moreover, it is thought that this method decreases the mental and physical stress of the patients.

Truncation correction of fan beam transmission data for attenuation correction using parallel beam emission data on a 3-detector SPECT system

BACKGROUND

When the simultaneous transmission computed tomography (TCT)/single photon emission CT (SPECT) acquisition protocol is applied to myocardial studies using a 3-detector SPECT, the narrow effective field of view of a fan beam collimator used for TCT acquisition may cause truncation artifacts on TCT images. In this paper, we propose a new method of correcting for the truncation of TCT.

METHODS

The truncated parts of the TCT projection data are corrected using quadratic functions, based on the properties that the integral of non-truncated TCT projection data is constant at any projection angle and the position of the centre of gravity is focused on a fixed point. The usefulness of our method was investigated in phantom and human studies using a 3-detector SPECT equipped with one cardiac fan beam collimator for TCT and two parallel beam collimators for SPECT. We used Tl as a tracer for SPECT and Tc as an external source for TCT.

RESULTS

The phantom and human studies showed that our method can adequately correct for the truncation of TCT data acquired using a fan beam collimator in a 3-detector SPECT, as long as there is no truncation in SPECT data.

CONCLUSION

Our method appears to be useful for improving the SPECT images obtained using simultaneous TCT/SPECT acquisition in a 3-detector SPECT. However, further studies will be necessary to establish the clinical usefulness of this method.

Segmented attenuation correction for myocardial SPECT

PURPOSE

One of the main factors contributing to the accuracy of attenuation correction for SPECT imaging using transmission computed tomography (TCT) with an external gamma-ray source is the radionuclide count. To reduce deterioration of TCT images due to inadequate radionuclide counts, a correction method, segmented attenuation correction (SAC), in which TCT data are transformed into several components (segments) such as water, lungs and spine, providing a satisfactory attenuation correction map with less counts, has been developed. The purpose of this study was to examine the usefulness of SAC for myocardial SPECT with attenuation correction.

METHODS

A myocardial phantom filled with Tc-99m was scanned with a triple headed SPECT system, equipped with one cardiac fan beam collimator for TCT and two parallel hole collimators for ECT. As an external gamma-ray source for TCT, 740 MBq of Tc-99m was also used. Since Tc-99m was also used for ECT, the TCT and ECT data were acquired separately. To make radionuclide counts, the TCT data were acquired in the sequential repetition mode, in which a 3-min-rotation was repeated 7 times followed by a 10-min-rotation 4 times (a total of 61 minutes). The TCT data were reconstructed by adding some of these rotations to make TCT maps with various radionuclide counts. Three types of SAC were used: (a) 1-segment SAC in which the body structure was regarded as water, (b) 2-segment SAC, in which the body structure was regarded as water and lungs, and (c) 3-segment SAC, in which the body structure was regarded as water, lungs and spine. We compared corrected images obtained with non-segmentation methods, and with 1- to 3-segment SACs. We also investigated the influence of radionuclide counts of TCT (3, 6, 9, 12, 15, 18, 21, 31, 41, 51, 61 min acquisition) on the accuracy of the attenuation correction.

RESULTS

Either 1-segment or 2-segment SAC was sufficient to correct the attenuation. When non-segmentation TCT attenuation methods were used, rotations of at least 31 minutes were required to obtain sufficiently large counts for TCT. When the 3-segment SAC was used, the minimal acquisition time for a satisfactory TCT map was 7 min.

CONCLUSION

The 3-segment SAC was effective for attenuation correction, requiring fewer counts (about 1/5 of the value for non-segmentation TCT), or less radiation for TCT.

Attenuation correction of myocardial SPECT images with X-ray CT: effects of registration errors between X-ray CT and SPECT

PURPOSE

Attenuation correction with an X-ray CT image is a new method to correct attenuation on SPECT imaging, but the effect of the registration errors between CT and SPECT images is unclear. In this study, we investigated the effects of the registration errors on myocardial SPECT, analyzing data from a phantom and a human volunteer.

METHODS

Registerion (fusion) of the X-ray CT and SPECT images was done with standard packaged software in three dimensional fashion, by using linked transaxial, coronal and sagittal images. In the phantom study, an X-ray CT image was shifted 1 to 3 pixels on the x, y and z axes, and rotated 6 degrees clockwise. Attenuation correction maps generated from each misaligned X-ray CT image were used to reconstruct misaligned SPECT images of the phantom filled with 201Tl. In a human volunteer, X-ray CT was acquired in different conditions (during inspiration vs. expiration). CT values were transferred to an attenuation constant by using straight lines; an attenuation constant of 0/cm in the air (CT value = -1,000 HU) and that of 0.150/cm in water (CT value = 0 HU). For comparison, attenuation correction with transmission CT (TCT) data and an external gamma-ray source (99mTc) was also applied to reconstruct SPECT images.

RESULTS

Simulated breast attenuation with a breast attachment, and inferior wall attenuation were properly corrected by means of the attenuation correction map generated from X-ray CT. As pixel shift increased, deviation of the SPECT images increased in misaligned images in the phantom study. In the human study, SPECT images were affected by the scan conditions of the X-ray CT.

CONCLUSION

Attenuation correction of myocardial SPECT with an X-ray CT image is a simple and potentially beneficial method for clinical use, but accurate registration of the X-ray CT to SPECT image is essential for satisfactory attenuation correction.