| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, St. Hill Hospital, 1462-3 Nishikiwa, Ube, Yamaguchi
755-0151, Japan.
2 Department of Radiology, Yamaguchi University School of Medicine, Yamaguchi
755-8505, Japan.
Received September 29, 2006;
accepted after revision March 2, 2007.
Address correspondence to K. Suga
(sugar{at}sthill-hp.or.jp).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Eighty-seven consecutive patients with chronic obstructive pulmonary disease (COPD) (n = 43), acute pulmonary thromboembolism (PTE) (n = 26), and interstitial lung disease (ILD), (n = 18), underwent respiratory-monitored DIBrH SPECT with a dual-headed SPECT system. Two COPD and four acute PTE patients were excluded because of inappropriate scanning due to DIBrH difficulty. DIBrH SPECT was automatically fused with DIBrH CT. Perfusion defect clarity and heterogeneity and SPECT/CT matching were compared between DIBrH SPECT and non-breath-hold SPECT.
RESULTS. Compared with non-breath-hold SPECT, DIBrH SPECT significantly enhanced defect clarity in acute PTE (p < 0.0001) and perfusion heterogeneity (coefficient of variations [CV] of pixel counts) in COPD and ILD (p < 0.0001). CV in COPD was also better correlated with lung diffusing capacity for carbon monoxide (p < 0.05). DIBrH SPECT also significantly improved SPECT/CT matching (p < 0.0001), with excellent matching of CT lung internal landmarks and pathology with corresponding defects. Fusion images confirmed wedge-shaped defects extending along specific pulmonary arterial branches in acute PTE and heterogeneous defects associated with airway or lung parenchymal abnormalities in COPD and ILD, with perfusion distribution consistent with lung CT attenuation changes.
CONCLUSION. DIBrH SPECT is acceptable for routine application to improve respiratory motion effects and accuracy of SPECT/CT image fusion. Confirmative perfusion-morphologic correlation with reliable fusion images appears useful for clarifying the cause of perfusion defects and abnormal lung CT attenuation.
Keywords: chronic obstructive pulmonary disease • CT • lung • lung diseases • nuclear medicine • pulmonary perfusion • pulmonary thromboembolism • SPECT
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In this study, to resolve these problems of traditional non-breath-hold perfusion SPECT, we developed a respiratory-monitored deepinspiratory breath-hold (DIBrH) SPECT technique using a widely available dual-headed SPECT system. A phantom study was preliminarily conducted to determine the optimal DIBrH SPECT technique and to test the ability of this technique in improving respiratory motion effects. On the basis of the results of the phantom study, we performed DIBrH SPECT in patients with chronic obstructive pulmonary disease (COPD), acute pulmonary thromboembolism (PTE), and interstitial lung disease (ILD) and evaluated the clinical applicability and feasibility of this technique to improve respiratory motion effects and accuracy of SPECT/CT image fusion compared with non-breath-hold SPECT. The feasibility of perfusion-morphologic correlation on DIBrH SPECT/CT fusion images in evaluation of respective diseases was also discussed.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Initially, 360° projection data of the standing phantom were repeatedly obtained with an acquisition time of 20 seconds (simulation of SPECT during 20-second DIBrH in the clinical study) 10 times, using a continuous rotating repeated acquisition (CRRA) mode, in a 128 x 128 matrix (field of view, 422.4 x 422.4 mm; 1 voxel = 3.3 mm3) and with an energy window of 140 keV ± 10%. In this mode, a gantry rotation of 180° around the phantom of each detector in clockwise and counterclockwise directions provided the projection data over a full 360° arc to eliminate the setting time between projection and acquisition of multiple temporal samples of data.
Subsequently, this standing phantom was also scanned using a step-and-shoot mode, with 30 stops over 180° for each detector and a preset time of 16 seconds for each stop of 6°, for a total acquisition time of 8 minutes. This phantom was regularly moving in a head-to-caudal direction with a range of 12 mm at 15 cycles per minute to simulate respiratory lung motion in patients. To control this regular motion, the phantom was placed on a plastic table with four runners. This table was directly connected to the piston arm of the volume-cycled ventilator using hard aluminum arms. SPECT of this moving phantom was performed with a total acquisition time of 8 minutes using the same step-and-shoot mode.
On the image workstation (GCA 9300 A/DI, Toshiba Medical Systems), transaxial, 3.3-mm-thick SPECT images of the standing chest phantom were reconstructed from every 6° projection data set of 3-10 projection data from the continuous rotating repeated acquisition mode using a Butter-worth prefilter (order number, 8; cutoff frequency, 0.49 cycles/cm) and a ramp back-projection filter. The conventional SPECT images of the standing and moving phantom were reconstructed from the every 6° projection data of a step-and-shoot mode, with the same thickness and a Butterworth prefilter of the same order number and cutoff frequency. The lung parenchyma of the phantom was drawn at a 10% threshold of the maximum radioactivity.
Clinical Study
Patient population—A total of 87 consecutive patients
including 43 patients with chronic obstructive pulmonary disease (COPD) (36
patients with pulmonary emphysema, four patients with obstructive
bronchiolitis, two patients with diffuse panbronchiolitis, and one patient
with bronchomalacia), 26 patients with acute pulmonary thromboembolism (PTE),
and 18 patients with interstitial lung disease (ILD) underwent DIBrH perfusion
SPECT during the period between February 2004 and May 2006. Of these patients,
two patients with COPD and four patients with acute PTE were excluded because
of difficulty with DIBrH, although non-breath-hold SPECT was obtained. All
patients with COPD and ILD underwent unenhanced CT within 7 days of the SPECT
examination, and those with acute PTE underwent unenhanced CT and CT pulmonary
angiography after SPECT. Unenhanced CT in patients with acute PTE was
performed to evaluate the presence or absence of other lung lesions causing
dyspnea and hypoxia or pulmonary infarction, and the indication for CT
pulmonary angiography was decided on the basis of the results of the SPECT
study. The diagnosis of each lung disorder was determined by clinical features
and course, laboratory findings, pulmonary function test results,
bronchofiberscopic biopsy, and imaging techniques such as high-resolution CT,
ventilation SPECT, or CT pulmonary angiography and according to the diagnostic
criteria presented in the previously established guidelines
[2,
8,
15]. In all patients with
acute PTE, CT pulmonary angiography showed a single or several filling defects
indicative of intravascular clots in the pulmonary arteries. The procedure of
DIBrH SPECT was approved by the institutional review board of the Yamaguchi
University School of Medicine. After the nature of the procedure had been
fully explained, informed consent was obtained from all patients.
Respiratory-monitored DIBrH perfusion SPECT— DIBrH perfusion SPECT was performed using the same SPECT system as used in the phantom study and a pressure sensor respiratory tracking device (AZ-733, Anzai Sogyo Co.). With each subject in the supine position, the pressure sensor was attached to the subject's thoracic or abdominal walls showing the maximum respiratory movement and connected to a physiologic respiratory tracking unit for monitoring real-time respiratory motion on the time-distance curves. Then each subject was directed to take regular DIBrH for 20 seconds by referring to the time curves on the respiratory tracking monitor. The decision to perform repeated DIBrH SPECT with 20-second DIBrHs was made on the basis of observation of the first 20-second DIBrH with or without nasal oxygen inhalation (1-3 L/min).
In the patients who could not perform the first DIBrH with oxygen
inhalation, only non-breath-hold SPECT was performed. After IV injection of
185 MBq of 99mTc MAA (
800,000 particles), projection data were
acquired only during each DIBrH in a 128 x 128 matrix (field of view,
422.4 x 422.4 mm, 1 voxel = 3.3 mm3) and with an energy
window of 140 keV ± 10% using the CRRA mode, as mentioned earlier. A
gantry rotation of 180° of each detector around the patient's chest during
each DIBrH provided the projection data over a full 360° arc. The
projection data were acquired with sufficient rest for each individual
patient, until eight adequate projection data sets (four clockwise and four
counterclockwise) with almost the same respiratory dimensions could be
obtained. For comparison, non-breath-hold SPECT during rest breathing was
subsequently acquired in a 128 x 128 matrix using a step-and-shoot mode,
with 30 stops over 180° for each detector and a preset time of 16 seconds
for each stop of 6°. The image acquisition time was 8 minutes.
On the image data processor, DIBrH SPECT was reconstructed from every 6° projection of eight selected projection data sets, and non-breath-hold SPECT from every 6° projection of the acquired data. For respective SPECT examinations, a total of 72-128 transaxial sections, 3.3-mm thick, covering the entire lungs were reconstructed using a Butterworth prefilter (order number, 8; cutoff frequency, 0.49 cycles/cm for DIBrH SPECT and 0.15 cycles/cm for non-breath-hold SPECT) and a ramp back-projection filter. The use of the cutoff frequency of 0.49 cycles/cm for DIBrH SPECT was based on the validation in our initial patient group. Lung contour on respective SPECT examinations was drawn at a threshold of 10% of the maximum radioactivity of the lungs in each patient.
Automated image fusion of DIBrH perfusion SPECT and CT—DIBrH CT was performed using a 4-MDCT scanner (Volume Zoom, Siemens-Asahi Medical Ltd.). With the subject in the supine position, 3-mm-thick high-resolution CT images were obtained in the helical mode in a 512 x 512 matrix during DIBrH using 3-mm collimation with a tube rotation time of 1.0 second at 120-135 kVp and 200-230 mA. Transaxial CT images were reconstructed with the lung algorithm.
DICOM data of CT and SPECT were electronically transferred to the teleradiologic workstation (GMS 5500 A/DI, Toshiba Medical Systems) and were reformatted to 168 x 168 matrix images for automated image fusion. Image fusion was performed using a simple rigid-body transformation technique provided by the fully automated 3D multiimage registration software (ART [automatic registration tool], Toshiba Medical Systems) [10, 12]. The automated image fusion was performed by definition of body contour and clustering of the voxels inside the body contour into a set of connected components using a rigid-body transformation technique [12]. The optimal registration parameters (translation for x-, y-, and z-axes and rotation angles for these axes) were automatically and three-dimensionally found by minimizing the variance of the SPECT voxel values within each connected component using the method of coordinate descent. Fully automatic registration was possible even when the slice range and thickness of the object were different between SPECT and CT. This software also allowed manual SPECT/CT matching on three orthogonal (transaxial, coronal, and sagittal) planes. Fusion images throughout the lungs were displayed together with the corresponding SPECT and CT images on the three orthogonal planes. To visually distinguish CT from SPECT information on fusion images, CT images were displayed in gray-scale and SPECT images in color.
Image Interpretation and Data Analysis
Phantom study—Two experienced observers visually compared
the perfusion defect clarity of the simulated nodules among SPECT images of
the standing phantom reconstructed from 1-13 continuous rotating repeated
acquisition mode projection data sets and reconstructed from the projection
data obtained using a step-and-shoot mode with a total acquisition time of 8
minutes and of the moving phantom obtained using a step-and-shoot mode with a
total acquisition time of 8 minutes. The result was based on a consensus of
these observers. The inhomogeneity of radiotracer distribution in the
simulated lung parenchyma was quantitatively compared among those SPECT images
mentioned earlier by estimating the coefficient of variation (CV = SD/mean
counts) values of pixel counts. For this assessment, the regions of interest
(ROIs) were manually placed on the same location of the lung parenchyma at
five slices on each SPECT image. The size of these ROIs ranged from 1,194 to
1,562 pixels (mean, 1,247 ± 232 pixels).
Clinical study—To compare the detectability of perfusion defects between DIBrH and non-breath-hold SPECT examinations, the number of perfusion defects in patients with PTE was counted independently by the two observers. These observers randomly reviewed SPECT images of all cases on the display. The confidence level for each defect was scored (no defect = 0, possible defect = 1, definitive defect = 2) to evaluate the interobserver agreement. A perfusion defect with the confidence level score of 1 or 2 for both reviewers was defined as a positive defect, and a defect with the confidence level of 0 for either observer was judged as a negative defect.
For quantitative analysis, the maximum difference in respiratory dimension among the eight continuous rotating repeated acquisition mode projection data sets used for reconstructing DIBrH SPECT was calculated. The total lung radioactivity counts in all patients, perfusion defect clarity in patients with acute PTE, and perfusion heterogeneity in patients with COPD and ILD were compared between DIBrH and non-breath-hold SPECT. Perfusion defect clarity in acute PTE was compared by estimating the count ratio of defects against the adjacent normal areas (D/N ratio) in 54 relatively small defects at various locations. For this assessment, ROIs were manually placed over each defect and the adjacent normal areas at the same locations between respective SPECT images by consensus of two experienced observers, and the D/N ratio was calculated by using the mean counts per pixel. The ROI size ranged from 8 to 94 pixels (mean, 34 ± 23 pixels). Perfusion heterogeneity in COPD and ILD was compared by estimating CV values in the entire lungs. The CV values estimated on DIBrH and non-breath-hold SPECT in patients with COPD were correlated with the diffusing capacity of the lungs for carbon monoxide/alveolar volume ratios (DLCO/VA) as an indicator of alveolar dysfunction. (DLCO/VA) was measured by the single-breath method (Pulmorecorder, Model R1551S, Anima), according to the recommendation of the American Thoracic Society [15].
Two experienced observers made a consensus comparison of SPECT/CT matching accuracy between the automated fusion images created by DIBrH and non-breath-hold SPECT examinations by assessing boundary matching of the lungs, internal landmarks (the large hilar vessels and bronchi, mediastinal structures, and diaphragm), and visible CT lung abnormalities with corresponding perfusion defects. After confirming accurate SPECT/CT matching, these observers also made a consensus interpretation of perfusion defect-morphologic correlation on DIBrH SPECT/CT fusion images. Because the reformatted 168 x 168 matrix CT images on fusion images still kept sufficient resolution to identify pulmonary vessels at the subsegmental level, comprehensive correlation with lung morphology on unreformated 512 x 512 matrix CT images could be performed.
The other two observers quantitatively compared SPECT/CT mismatch distance
between the lung base and diaphragm contours on these fusion images. For this
assessment, fusion images were displayed on three orthogonal planes, and these
observers independently and manually reperformed the best matching of the lung
base and diaphragm contours. From the translated distances of
X, Y, and Z (mm) relative to
the x-, y-, and z-axes for this best matching, 3D mismatch
distance was calculated by the formula of (X2 +
Y2 + Z2)0.5.
The values were averaged between the two measurements by these observers and
were compared between the respective fusion images.
Statistical Analysis
Values were expressed as mean ± SD, and a paired or unpaired
Student's t test was applied to compare the differences in the data
between DIBrH and non-breath-hold SPECT images. Interobserver agreement in
detection of perfusion defects of patients with acute PTE on DIBrH SPECT was
assessed by calculating the kappa value using the confidence level scores of
two observers. To evaluate the correlation between CV values and
(DLCO/VA) in patients with COPD, a linear regression
analysis was performed using commercially available software (StatView 4.02 SE
+ Graphics, Abacus Concepts), and a p value of less than 0.05 was
considered significant for each correlation coefficient (R). The
significant difference of R values of the multiple regression lines
was evaluated by Neyaman-Pearson test with Fischer's Z transformation scores,
where the Z value of 1.96 was used for 95% confidence.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
Clinical Study
In all 81 patients, eight adequate continuous rotating repeated acquisition
mode projection data sets were obtained with nine to 14 repeated DIBrHs, for a
total examination time of 8.7 ± 1.8 minutes (range, 6-14 minutes),
although oxygen inhalation was given in eight patients. The maximum difference
in respiratory dimension among these projection data sets was minimal at 2.2
± 0.5 mm (range, 1.1-4.3 mm).
Total lung radioactivity counts on DIBrH SPECT were decreased to approximately 29% of those on non-breath-hold SPECT (4.2 x 106 ± 0.64 counts vs 14.7 x 106 ±0.63 counts). However, DIBrH SPECT often enhanced perfusion defect clarity in patients with acute PTE compared with non-breath-hold SPECT, with frequent expansion of defects (Figs. 2A, 2B, 3A, 3B, 3C, 4A, 4B, 5A, 5B). This SPECT, in addition, detected 32 (31%) defects compared with non-breath-hold SPECT (135 vs 103) in these patients, with a good interobserver agreement (kappa value, 0.79). D/N ratios at 54 small defects were significantly less on DIBrH SPECT (0.29 ± 0.18 vs 0.35 ± 0.24; p < 0.0001).
|
|
|
|
|
|
|
|
|
|
On unreformatted CT, all patients with COPD had low CT lung attenuation areas representative of emphysematous changes or air trapping, with or without bullae or bronchial wall thickening. All patients with ILD had high CT attenuation areas representative of interstitial thickening. Twelve of the 22 patients with acute PTE had low CT attenuations or mixed low and high (mosaic) attenuations in embolized lungs. DIBrH SPECT/CT fusion images showed an excellent matching of these visible lung abnormalities and corresponding defects, with excellent boundary matching of the lungs and internal landmarks with their defects (Figs. 2A, 2B, 3A, 3B, 3C, 4A, 4B, 5A, 5B). These fusion images could confirm wedged-shaped defects extending along the specific pulmonary arterial branches in patients with acute PTE (Fig. 4A, 4B) and heterogeneous defects associated with the airway or lung parenchymal morphologic abnormalities described earlier in patients with COPD and ILD (Figs. 2A, 2B, 3A, 3B, 3C, 5A, 5B). These images also could accurately confirm peripheral or subpleural defects according to the outer boundary of the lungs delineated by coregistered CT (Figs. 2A, 2B, 3A, 3B, 3C, 5A, 5B). The heterogeneous defects in normal lung areas on CT were also occasionally confirmed in patients with COPD and ILD (Figs. 2A, 2B, 5A, 5B). In three patients with COPD and two patients with acute PTE, only focal relatively high CT attenuation areas within widely affected lungs mimicked ground-glass opacities associated with focal pneumonia or alveolar edema. Fusion images could confirm that these areas were normal with preserved perfusion (Fig. 3A, 3B, 3C).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The enhancement of perfusion defect clarity and heterogeneity on DIBrH SPECT in our patients can be explained by significant reduction of respiratory motion effects and by expansion of defects due to lung volume increases at DIBrH [9-11]. On conventional non-breath-hold SPECT, the lung motion during every respiratory cycle and the cyclically changing lung volume inherently degrade image sharpness of ill-defined defects [11]. Respiratory lung volume change alters radioactivity per unit of lung volume and perfusion defect size, although 99mTc MAA distribution is fixed and the size of defects depends on the respiratory conditions existing during injection [11].
Due to lung volume increases at DIBrH, various degrees of expansion of perfusion defects should occur in patients with acute PTE, as shown in the previous PTE animal model study on planar perfusion [16]. The radioactivity per unit of lung volume should not be significantly changed in diseased lung areas of patients with COPD and ILD due to air trapping or low breathing compliance, although it is significantly increased in the surrounding normal lungs due to significant reduction of alveolar air space enhancing perfusion defect clarity. The superiority of perfusion SPECT compared with CT in detecting mild lesions of COPD and ILD has been previously reported [5, 8, 17-21]. DIBrH SPECT should further facilitate this advantage of perfusion SPECT.
DIBrH SPECT provided reliable automated fusion images with separately performed DIBrH CT without additional CT and without the use of multiple radioactive external landmarks or transmission images obtained using an external radionuclide [22, 23]. The use of the same respiratory phase of SPECT and CT for image fusion can yield automated reliable fusion images without use of other complex nonrigid deformation algorithms [13, 24]. The mean 3D misregistration of 2.6 ± 1.2 mm between the lung base and diaphragm contours on the present fusion images is smaller than the previously reported data in SPECT/CT and PET/CT fusion image studies [11, 13, 24-29].
Coregistered CT on these reliable fusion images compensated low-spatial-resolution SPECT and allowed accurate localization of perfusion defects on underlying detailed CT lung anatomy. As a result, segmental or heterogeneous nonsegmental defects, defects consistent with abnormal lung CT morphology or attenuation changes, could be confirmed in our patients. In acute PTE, wedge-shaped defects should appear along embolized pulmonary arterial branches, and these defects occasionally result in low CT lung attenuation due to direct reduction in the diameter and volume of peripheral vessels distal to vascular obstruction [30]. Previous angiographic studies showed focal low-attenuation areas in patients with acute PTE, with reduction of diameters and volume of the embolized pulmonary arteries [31, 32]. In this disease, objective localization of perfusion defects with fusion images is of value, because CT pulmonary angiography cannot reliably identify actually affected lung areas, regardless of excellent detectability of intravascular clots [2, 30].
In COPD, heterogeneous alveolar destruction or alveolar inflation or hypoxic vasoconstriction associated with heterogeneous air trapping should result in perfusion defects consistent with heterogeneous low CT attenuation [33, 34]. In ILD, alveolar interstitial tissues are usually affected heterogeneously with spared lung areas, and perfusion should be reduced at high CT attenuation areas representative of alveolar wall thickening or cellular infiltration [35]. As seen in several of our patients with acute PTE and COPD, focally spared normal lung areas within widely affected lungs might be misinterpreted as lesions. Accurate perfusion-morphologic correlation with fusion images appears useful for preventing such misinterpretation [10]. The present fusion images also occasionally confirmed perfusion defects even in morphologically normal lung areas in our patients with COPD and ILD. Detection of such areas with fusion images is of value because these defects reflect early stage lesions of COPD and ILD [18-21]. Previous studies showed the superiority of perfusion scintigraphy to CT in detecting mild lesions of these diseases [5, 8, 10, 17-21].
In conclusion, the present DIBrH SPECT technique appears acceptable for routine practice. By improving respiratory motion effects, DIBrH SPECT facilitates detectability of perfusion defects and provides reliable automated SPECT/CT fusion images. Confirmative perfusion-morphologic correlation with these fusion images appears to have incremental value for clarifying the cause of perfusion defects or abnormal CT appearance and may enhance diagnostic accuracy in various lung diseases. Although hybrid SPECT/CT systems will facilitate the use of lung fusion images, the problem of adverse respiratory motion effects should remain. The present technique will be also useful for resolving this problem. However, for patients who have difficulty with repeated DIBrH, a shorter projection data acquisition with optimization of administration dose of the radiotracer, numbers of projection data sets, or detector rotation speed is warranted.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
