Usefulness of the automatic quantitative estimation tool for cerebral blood flow: clinical assessment of the application software tool AQCEL.

OBJECTIVE: AQCEL enables automatic reconstruction of single-photon emission computed tomogram (SPECT) without image degradation and quantitative analysis of cerebral blood flow (CBF) after the input of simple parameters. We ascertained the usefulness and quality of images obtained by the application software AQCEL in clinical practice. METHODS: Twelve patients underwent brain perfusion SPECT using technetium-99m ethyl cysteinate dimer at rest and after acetazolamide (ACZ) loading. Images reconstructed using AQCEL were compared with those reconstructed using conventional filtered back projection (FBP) method for qualitative estimation. Two experienced nuclear medicine physicians interpreted the image quality using the following visual scores: 0, same; 1, slightly superior; 2, superior. For quantitative estimation, the mean CBF values of the normal hemisphere of the 12 patients using ACZ calculated by the AQCEL method were compared with those calculated by the conventional method. The CBF values of the 24 regions of the 3-dimensional stereotaxic region of interest template (3DSRT) calculated by the AQCEL method at rest and after ACZ loading were compared to those calculated by the conventional method. RESULTS: No significant qualitative difference was observed between the AQCEL and conventional FBP methods in the rest study. The average score by the AQCEL method was 0.25 ± 0.45 and that by the conventional method was 0.17 ± 0.39 (P = 0.34). There was a significant qualitative difference between the AQCEL and conventional methods in the ACZ loading study. The average score for AQCEL was 0.83 ± 0.58 and that for the conventional method was 0.08 ± 0.29 (P = 0.003). During quantitative estimation using ACZ, the mean CBF values of 12 patients calculated by the AQCEL method were 3-8% higher than those calculated by the conventional method. The square of the correlation coefficient between these methods was 0.995. While comparing the 24 3DSRT regions of 12 patients, the squares of the correlation coefficient between AQCEL and conventional methods were 0.973 and 0.986 for the normal and affected sides at rest, respectively, and 0.977 and 0.984 for the normal and affected sides after ACZ loading, respectively. CONCLUSIONS: The quality of images reconstructed using the application software AQCEL were superior to that obtained using conventional method after ACZ loading, and high correlations were shown in quantity at rest and after ACZ loading. This software can be applied to clinical practice and is a useful tool for improvement of reproducibility and throughput.

Semiquantitative measurement of pulmonary hilar gallium-67 uptake using single photon emission computed tomography/computed tomography for the diagnosis of autoimmune pancreatitis.

PURPOSE: We ascertained the difference in the level of pulmonary hilar (PH) gallium-67 (Ga-67) uptake in autoimmune pancreatitis (AIP) patients between the active (before corticosteroid therapy) and remittent (after 4 weeks of corticosteroid therapy) phases using a semiquantitative measurement method based on single photon emission computed tomography (SPECT) with integrated low-dose computed tomography (CT). MATERIALS AND METHODS: Ga-67 planar scans and SPECT were performed in 11 AIP patients before and after 4 weeks of corticosteroid therapy. A region of interest (ROI) was drawn over the bilateral pulmonary hilum and the liver, and average counts of the pulmonary hilum and liver were calculated. The average counts of the pulmonary hilum divided by those of the liver gave the pulmonary hilum/liver ratio (H/L). RESULTS: A significant difference was observed in the H/L between active and remittent phases. H/L in the active phase was 1.03±0.32 and that in the remittent phase was 0.58±0.25 (t-test, P=0.0016). CONCLUSIONS: PH Ga-67 uptake in patients with active AIP was semiquantitatively higher than that in those with remittent AIP. Hence, this may be a useful finding for an early diagnosis, estimating the effectiveness of corticosteroid therapy, and following up patients with this disease.

[Development of the automated Patlak plot method and its verification in clinical examples].

The Patlak plot method is widely used in general clinical practice to calculate mean cerebral blood flow (mCBF) because it does not require arterial blood sampling and the procedure used is relatively simple. However, it has been pointed out that the calculation is affected by operator subjectivity and experience, and reproducibility is poor when the operator changes. The Patlak plot method consists of the following three procedures: (1) a region of interest (ROI) in the normal cerebral hemisphere is chosen to obtain the time-activity curve; (2) an ROI in the aortic arch is determined to obtain the time-activity curve; and (3) graphic analysis of the two time-activity curves is performed, and mCBF is determined from the slope of the graph. In this study, we automated all three of the above procedures, enabling analysis that is not affected by operator subjectivity or experience. Considering the verification result shown below: (1) automation of an ROI setup of a cerebral hemisphere and an aortic arch was possible in all examples; (2) the point of graphic analysis was set in the error range of 1% - 2% of straight line domains; (3) in comparison with the conventional method in 55 cases, the mean of absolute percentage error was 3.1+/-2.5%; (4) in 17 cases that did not accept a right-and-left difference in SPECT, the mean of absolute percentage error of right and left mCBF was 1.3+/-0.9%. It was thought that the processing algorithm was appropriate. Moreover, it was thought that the technique of graphic analysis proposed for this study served as an aid to dissolution of the "ambiguity" of the Patlak plot method.

[Development of an automatic analytical tool for quantitative cerebral blood flow measurement using 99mTc-ECD, and verification of clinical examples].

The following process conventionally has been followed to develop quantitative images of cerebral blood flow: (1) mean cerebral blood flow (mCBF) is calculated by the Patlak plot method; (2) a SPECT slice that includes the basal ganglia is selected; and (3) based on the value of mCBF calculated by the Patlak plot method, the SPECT slice is corrected by the Lassen method and developed into a SPECT image of quantitative regional cerebral blood flow. However, this process is complicated, and the values of rCBF have been reported to fluctuate because selection of the SPECT slice and the ROI setting are in the hands of the operator. We have developed new software that automates this analysis. This software enables automatic processing simply by inputting the value of mCBF in the normal hemisphere. Since there is no need for manual operations such as setting the ROI, reproducibility is improved as well. Regional cerebral blood flow as determined by this software is quite similar to that calculated by the conventional method, so the existing clinical evaluation does not need to be changed. This software is considered to be useful.