| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Diagnostic Imaging and Nuclear Medicine, Kyoto University
Graduate School of Medicine, 54 Shogoinkawahara-cho, Sakyo-Ku, Kyoto 606-8507,
Japan.
2 Department of Image-Based Medicine, Institute of Biomedical Research and
Innovation, Kobe, Japan.
Received April 26, 2005;
accepted after revision December 16, 2005.
Address correspondence to Y. Nakamoto.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Ethics committee approval and informed consent were obtained. Sixty-three patients with suspected recurrent colorectal cancer underwent whole-body 18F-FDG PET followed by diagnostic CT. The acquired PET and CT images were merged on a workstation on a pixel-to-pixel basis. CT, PET, PET + CT, and fused images were evaluated separately in terms of the presence or absence of recurrence, new metastases, or both using a 5-point grading scale (0 = definitely negative, 1 = probably negative, 2 = equivocal, 3 = probably positive, and 4 = definitely positive). Lesions determined to be grade 3 or 4 were considered positive, and diagnostic accuracy and certainty were evaluated with statistical analysis using the chi-square test for independence.
RESULTS. Of 119 pathologically or clinically confirmed lesions in 36 patients, evaluation of CT, PET, PET + CT, and fused images resulted in the detection of 75 (63%), 84 (71%), 91 (76%), and 111 (93%) lesions, respectively (p < 0.01) with the number of grade 4 lesions detected being 59 (50%), 72 (61%), 84 (71%), and 108 (91%), respectively (p < 0.01). Overall, the diagnostic accuracy of CT, PET, PET + CT, and fused images according to patient were 78%, 79%, 84%, and 92%, respectively (p = 0.13).
CONCLUSION. Interpreting fused images provided more accurate diagnoses than interpreting CT, PET, or PET + CT images. This method of manually fusing separately obtained PET and CT images increased the diagnostic certainty for detecting colorectal cancer recurrence and decreased the number of equivocal cases.
Keywords: colorectal cancer • colorectal cancer recurrence • CT • FDG PET • fusion imaging • oncologic imaging • PET • PET/CT • software-based fusion imaging techniques
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Combined PET/CT scanners were developed to address this issue [9, 10], and reports examining the clinical usefulness of that new technique in cases of suspected recurrent colorectal cancer are promising [11-13]. The PET/CT device captures PET and CT image data sets at a single examination without the need to reposition the patient. Visualization of metabolic abnormalities using PET superimposed on high-spatial-resolution CT allows the highly precise localization of those abnormalities. In addition, fused images obtained using a PET/CT scanner allow better identification of areas for biopsy.
Although an increasing number of institutes have installed combined PET/CT scanners rather than dedicated PET scanners, not all institutes have access to this new technology. Many dedicated PET scanners have been replaced by in-line PET/CT systems recently, whereas some institutes cannot install a PET/CT scanner because of economic considerations. In addition, many hospitals already have a dedicated PET camera and MDCT scanner. If a simple manual fusion technique for the whole body based on high-quality images was effective, such a technique would have a major impact on clinical PET practice.
The purpose of this retrospective study was to investigate the clinical usefulness of a simple manual fusion technique for separately acquired CT and PET images obtained using a vacuum cushion for positioning in both studies and to compare the diagnostic performance of the manually fused images with CT images alone, PET images alone, and side-by-side review of separately obtained PET and CT images (hereafter referred to as "PET + CT").
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Of the 63 patients, 45 were suspected of having recurrence due to foci revealed on other techniques, such as CT (n = 31) or MRI (n = 8), or to the recognition of increased levels of tumor markers, such as carcinoembryonic antigen (n = 30) or cancer antigen (CA) 19-9 (n = 1). These patients were called group A. Group B consisted of the remaining 18 patients who had been considered stable without recurrence, but a referring physician or the patient (or both) had requested PET and CT scans during the follow-up period. This study had institutional review board approval, and all patients gave written informed consent before commencement of the study.
PET and CT
After fasting at least 4 hours, patients received 111-148 MBq (3-4 mCi) of
18F-FDG synthesized by the Merrifield method
[14]. The plasma glucose level
was monitored just before IV administration of 18F-FDG; it ranged
from 75 to 164 mg/dL, with an average of 101 mg/dL. Fifty minutes after
injection, patients were urged to void. They were then positioned in a large
(200 x 60 x 5 cm) vacuum cushion (ESFORM, Engineering System),
which has been widely used for positioning patients for radiation therapy on
the table of a PET scanner with their arms over their head
[15]. When air is drawn from
the cushion, it becomes a rigid cradle for the anatomic area to be
immobilized. After the patient was positioned, a felt-tip pen was used to mark
a median line on the skin from the navel to the sternum, bilateral horizontal
longitudinal lines, and a transverse line indicating the start of scanning on
the upper thigh.
PET studies were performed using either an ECAT Exact 47 or an ECAT Exact HR+ PET camera (both cameras, Siemens Medical Solutions). These devices simultaneously acquire 47 planes over a 16.2-cm (ECAT Exact 47) axial field of view or 63 planes over a 15.5-cm (ECAT Exact HR+) axial field of view. After the patient was positioned as described, a static emission scan was obtained in the 3D mode with 2-3 minutes of acquisition at each table position to cover the area from the upper thigh to the meatus of the ear. Then a transmission scan using germanium-68 (68Ge)-gallium-68 (68Ga) rod sources was obtained over the same area for 2 minutes per table position. Attenuation-corrected images were made using an ordered subset expectation maximization iterative reconstruction algorithm (four iterations, eight subsets).
After PET scanning was completed, patients were urged to void again to get a consistent bladder shape and reduce radiation exposure. Patients were then moved to a CT room where they were repositioned in the same molded vacuum cushion as that used for PET; the median and bilateral surface lines that were marked on the patient before PET were used as a reference for positioning (Fig. 1). The CT device was an MDCT scanner (Aquilion, Toshiba Medical Systems) with four detectors. The technical parameters used for CT were as follows: 120-kV peak energy, 200- to 450-mA tube electric current with automated radiation exposure control, helical pitch of 5.5 (high-speed mode), 3-mm collimation, and 5-mm reconstruction thickness.
|
Image Processing
Both CT and PET data sets were transferred to a workstation (Sun Ultra 60,
Sun Microsystems). PET images, enlarged by multiplying a zooming factor to fit
the field of view of the CT images (50 cm), were interpolated and resliced
using a matrix size of 512 x 512 and a 5-mm interval using a software
package (Dr. View, Asahikasei-Joho Systems). The slice showing the lower
margin of the urinary bladder was then determined on both the CT and the
modified PET images, and the PET images shifted craniocaudad to match the CT
images. Last, the two sets of images were merged on a pixel-to-pixel basis for
the whole body. These processes were performed by one of the observers, and
the computing time was approximately 5 minutes for each patient.
Image Evaluation
CT images were reviewed by two board-certified radiologists who had 13
(observer 1) or 14 (observer 2) years of experience and were blinded to other
examination results and clinical data. The location and certainty of abnormal
findings were determined by consensus of the two observers. In evaluating the
lymph nodes on CT, a size criterion was used to determine whether lymph nodes
were abnormal: They were considered abnormal if larger than 1 cm. Certainty
was assessed using a 5-point grading scale: 0, definitely negative; 1,
probably negative; 2, equivocal; 3, probably positive; and 4, definitely
positive.
After the CT images had been interpreted, the fused images were synthesized and reviewed by two board-certified radiologists and nuclear medicine physicians (observer 1, 10 years of experience with PET; observer 3, 12 years of experience with CT and 11 years of experience with PET) using the same grading scale. When interpreting the fusion images, the observers were also allowed access to the unfused images. The PET images were interpreted by two other board-certified nuclear medicine physicians (observer 4, 11 years of experience with PET; observer 5, 10 years of experience with PET) who evaluated for regions of focally increased radiotracer uptake but did not have access to additional information including CT. The location and diagnostic certainty of abnormal uptake were determined using the same grading scale described earlier. Semiquantitative analysis, including standardized uptake values, was not applied in this study. After interpreting the PET images, the same two observers, who are board-certified radiologists (observer 4, 15 years of experience; observer 5, 22 years of experience), placed the PET and CT images side-by-side (PET + CT) and interpreted the images, arriving at a consensus for location and grading scores. Image interpretation was performed using a workstation that gave proper intensity for PET and optimal density level for CT. Lesions determined to be grade 3 or 4 were considered positive. Based on the final diagnoses, diagnostic performances, such as sensitivity, specificity, accuracy, and certainty, were compared among the methods. In addition, the sensitivity of each method was evaluated according to lesion location.
For the lesion-based analysis, all findings of suspected metastasis or recurrence by at least one method were assessed as true-positive, false-positive, true-negative, or false-negative for each method based on the final diagnoses. When metastatic or recurrent foci appeared during the follow-up period despite all four methods showing no suspect lesions or morphologic changes, such cases were counted as false-negative lesions.
For the patient-based analysis, we regarded a case as a true-positive finding when at least one lesion that was assessed as grade 3 or 4 using that method was confirmed as a true-positive. When suspected lesions but not the true-positive lesion or lesions were identified, we considered that case as a false-positive finding. If no lesions were identified using any of the methods, the case was considered a true- or false-negative finding according to the outcome of the follow-up period.
For the side-by-side PET + CT interpretation and fused image interpretation, additional information and clinical impact were also considered. Additional information was defined as any additional findings seen on PET + CT or fusion imaging compared with previous findings, such as the detection of lesions not observed on conventional imaging or the visualization of lesion characteristics that had been inconclusive on conventional imaging. Potential new findings that were incorrect were not considered as additional information. Clinical impact was defined as cases in which the therapeutic management of a patient was influenced by the additional information acquired.
Standard of Reference
Histopathologic examination or clinical follow-up was used as the standard
of reference. Surgery was performed in 17 patients and biopsy, in one patient.
For these patients, histopathologic evaluation was conducted. In the remaining
45 patients, final diagnoses were determined on the basis of clinical
follow-up. Two patients died 2 months after examination due to recurrent
disease, and examination findings were clinically determined. Excluding those
two patients, the follow-up period ranged from 6 to 14 months, with an average
of 7.6 months. All available information, including CT scans of all patients
and follow-up PET scans of 14 patients, was used in determining the final
diagnosis. In addition, findings on other conventional imaging techniques,
such as MRI and sonography, were also considered.
Statistical Analysis
Diagnostic certainty, location-based data, and the presence or absence of
additional information and clinical impact were evaluated using the chisquare
test for independence. A p value of less than 0.05 was considered to
indicate a statistically significant difference.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Diagnostic capability on a per-lesion basis and on a per-patient basis is summarized in Tables 2 and 3, respectively. There was a significant difference among the detectability of true-positive lesions on CT (n = 75, 63%), PET (n = 84, 71%), PET + CT (n = 91, 76%), and fused (n = 111, 93%) images (p < 0.01). On a per-patient basis, the overall diagnostic accuracy of CT, PET, PET + CT, and fused images was 78%, 79%, 84%, and 92%, respectively, but the differences in accuracy were not statistically significant. Forty-two lesions equivocal on CT or not interpreted as positive were accurately diagnosed using the fused images, as shown in Figures 2A, 2B, and 2C.
|
|
|
|
|
To evaluate diagnostic certainty, the grading for true-positive lesions was examined. As shown in Table 4, of all the true-positive lesions, 59 (50%) were interpreted as grade 4 lesions on CT, 72 (61%) on PET, 84 (71%) on PET + CT, and 108 (91%) on fused images. The increase in lesions diagnosed as positive after interpretation of the fused images represented a statistically significant increase in diagnostic certainty (p < 0.01).
|
Location-based results are shown in Table 5. PET, PET + CT, and fused images were more accurate than CT for detecting metastases to bone, peritoneum, and lymph nodes (Figs. 3A, 3B, and 3C), whereas CT was slightly better than the other methods for detecting hepatic and pulmonary metastases (Figs. 4A, 4B, and 4C). Compared with side-by-side PET + CT, the fused images were especially helpful in detecting peritoneal metastases (13 lesions detected on side-by-side PET + CT images vs 20 lesions on fused images) and local recurrence (six vs 12 lesions, respectively) (Figs. 5A, 5B, and 5C). However, differences according to location were not statistically significant.
|
|
|
|
|
|
|
|
|
|
The number of cases that showed additional information and clinical impact are summarized in Table 6. In group B, additional information and clinical impact were obtained in six patients by interpreting PET + CT images or fused images. For group A patients, interpretation of fused images gave additional information in four more patients and that information had a clinical impact in three more patients than interpretation of PET + CT images. These three patients had local recurrence, which was not interpreted as positive on review of side-by-side images, but accurate diagnoses were obtained on fusion imaging. However, these results were not statistically significant.
|
Overall, fusion imaging gave additional information in 28 patients (44%) and had clinical impact in 25 patients (40%), of whom 10 patients underwent surgery, four patients had radiation therapy, six patients underwent chemotherapy, two patients received ablation therapy for their liver metastases, and scheduled treatment was cancelled in three patients.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Although the PET/CT system is helpful in a variety of clinical settings, the actual advantage of the PET/CT system is derived from the fused images provided by integrated scanners. However, fused images can be constructed not only by a combined PET/CT unit, but also by software alone. If comparable diagnostic accuracy could be obtained using software-based manual image-fusion techniques, this would represent a great benefit because not all institutes can afford PET/CT devices. Therefore, in the present study, we attempted to derive clinically acceptable fused images using software rather than a PET/CT scanner [15]. Using our method, we were able to derive diagnostic performance comparable with that obtained using combined PET/CT scanners.
Manual image-fusion techniques using software-based approaches have been used since the 1990s [18], with the head and neck considered a suitable area for these methods [19-21]. Compared with acquisition of head and neck images, accurate image acquisition of other areas was thought to be too challenging because of the variability of patient positioning and the involuntary movement of internal organs caused by respiration and peristalsis. With the rapid development of PET technology over the past 10 years, whole-body scans have become widely available and image quality has improved dramatically. Conventional imaging techniques, such as CT, and computer software and hardware have also made remarkable progress. Consequently, whole-body imaging using both PET and CT has now become possible and has resulted in increased diagnostic accuracy and clinical impact.
The interpretation of fused images has a number of advantages. For example, using PET alone, differentiating between pathologic and physiologic 18F-FDG accumulation is often difficult. In such cases, areas of uptake can be easily related to morphologic information obtained by CT, which then reduces false-positive or false-negative findings. Identifying some abnormalities using PET alone can also be difficult due to low signal uptake. The use of fused images allows the detection of metastases based on only slight morphologic changes on the CT images. Thus, the use of fused images has been suggested to allow higher diagnostic certainty and accuracy [11, 12].
Reinartz et al. [22] reported that integrated PET/CT devices provided additional information in approximately 6-7% of all lesions. However, in their report, abnormal CT findings in the absence of increased 18F-FDG uptake were excluded. Such findings are sometimes accurately diagnosed as positive only by interpreting fused images because lesions that result in only slight morphologic changes on CT and faint 18F-FDG uptake on PET can easily be missed on side-by-side image review. In our series, the interpretation of fused images provided higher diagnostic accuracy compared with side-by-side review than that reported by Reinartz et al.
Kim et al. [23] have compared diagnostic accuracy between software-based fusion imaging and integrated PET/CT, and those researchers reported that in-line PET/CT systems tended to show superior performance. However, such findings may be dependent on how the patients were positioned and on what kind of software was used for image handling. When our technique of patient positioning and coregistration for image fusion was applied, the diagnostic accuracy appeared to be comparable with the results obtained using an in-line PET/CT device.
For our fusion method, no specific algorithm for fusion was used, with image registration conducted by parallel shifts. The software package we used was commercially available and cost approximately US$10,000, including the latest Pentium 4 (Intel)-based computer. Therefore, one advantage of our technique compared with the installation of a combined PET/CT device is cost-effectiveness. Positioning for PET and repositioning for CT took 2-3 minutes, and computing time for image registration was less than 5 minutes once the CT and PET images were down-loaded from a server. The average displacement of pathologic uptake was approximately 1 cm when we evaluated it in establishing this method [15], whereas Cohade et al. [24] reported that it was approximately 8 mm in PET/CT in evaluating pulmonary nodules. We consider our fusion technique to be acceptable in the clinical setting.
During image analysis, we noted differences in diagnostic performance between the techniques based on location. For instance, small pulmonary nodules and liver metastases were slightly better detected on CT than PET, whereas metastases to the peritoneum, lymph nodes, and bone were more easily detected on PET. Our analysis also showed that some liver or lung metastases were misdiagnosed as negative even using PET and CT fusion images. These misdiagnoses appeared to be due to lesions being identified on CT, with a grading score of 3, but being considered negative because of absent or faint 18F-FDG uptake on PET, with a grading score of 2 or less. Based on CT findings alone, many small nodules in the lung would likely be interpreted as multiple lung metastases. Therefore, the lung should be carefully assessed using an optimal window level setting for the lung field on CT. Given solitary nodules that are 18F-FDG-dim, accurately diagnosing lung metastasis may be difficult. Likewise, some small liver metastases were not accurately diagnosed on both PET + CT images and fused images because of low accumulation of 18F-FDG in the metastatic tumor, even though the metastases would have been suspected by referring to CT findings alone. It may also be necessary to carefully check for abnormalities indicating metastatic foci on liver CT.
Local recurrence was accurately diagnosed in only half of the patients using side-by-side review, whereas interpreting fused images enabled us to reach true-positive diagnoses for all of the patients in our series. Of 12 cases of local recurrence, three lesions were regarded as equivocal because of their faint uptake of 18F-FDG, and the three other lesions were not identified using CT or PET because of indeterminate findings. Therefore, even if the uptake is faint on PET or the morphologic finding is not clear on CT, it may be possible to reach a true-positive diagnosis of local recurrence by combining morphologic findings with the 18F-FDG uptake on fused images. In contrast, when fused images were not available, such recurrent foci might be missed because we are not able to see the lesions on each image.
In our study, CT was performed in the venous phase without breath-holding. CT images are often not of sufficient quality to evaluate the lung and liver. Although CT should be performed in the end-expiratory phase to get the best registration of the diaphragm [25], we adopted free breathing because breath-holding for approximately 30 seconds was painful for some elderly patients, and it has been reported that free breathing can be used during CT in PET/CT [26]. Consequently, compared with conventional diagnostic CT images obtained during the end-inspiratory phase, our CT images used for fusion were a little noisy and blurred because of respiratory movement. Nonetheless, the quality of the images in the present study was still good enough to allow proper evaluation of whether pulmonary nodules were present in the lung field, indicating pulmonary metastasis. The results would also be adequate for use in disease staging or restaging, even though it may be difficult to accurately evaluate diffuse pulmonary disease.
The timing of the CT scanning is also open to discussion. Images showing the intrapelvic region must be reviewed carefully in patients with suspected colorectal cancer. To more easily evaluate metastases to the iliac nodes, we began CT 90 seconds after injection of the IV contrast material. However, this timing may be suboptimal for evaluation of the liver. Because liver metastases from colorectal cancer are common, further evaluation is required to determine the optimal CT protocol.
As shown by Slomka [27], software-based image fusion has some advantages, especially in terms of cost-effectiveness and the possibility of versatile image registration—for example, CT and SPECT or MRI and PET. Indeed, with more effort, it may be possible to get higher diagnostic accuracy without combined PET/CT scanning by obtaining convincing fused PET and CT images, as shown in the present study. However, our method is not a replacement for integrated PET/CT devices. Compared with integrated PET/CT, our PET and CT image fusion method takes significantly longer to perform. For patients with severe pain, the advantage of integrated PET/CT in being able to obtain clear fusion images in a short time is especially important. Nonetheless, our fusion imaging technique is of clear benefit to patients with suspected recurrent colorectal cancer. Although integrated PET/CT devices should be used when available, our technique represents an effective alternative.
There were some limitations in the design of this study. Colorectal cancer often spreads to the peritoneum, with small disseminated foci common. Because imaging techniques often miss these tiny lesions [28], the lesion-based sensitivity might be overestimated, even though we followed up patients for at least 6 months after examination. In addition, when multiple metastases were suspected, patients often receive chemotherapy without histopathologic confirmation. Although we assumed all malignant lesions were related to the primary cancer, there was no proof that all the lesions were pathologically true metastases. CT was performed 80-90 seconds after the administration of 100 mL of contrast material in this study to get clear contrast from the thoracic to pelvic regions. This compromised timing may not be optimal for evaluating the liver. Dual-phase CT—that is, early CT for the thoracic to upper abdominal regions and delayed CT for the abdominal to pelvic regions—might be helpful for obtaining higher diagnostic accuracy. Further investigations to determine optimal dose and timing in CT are necessary, although the use of 100 mL of 300 mg I/mL contrast material may be sufficient at least for fusion purposes [29].
In conclusion, our data indicated that the most accurate image-based diagnosis was obtained by interpreting fused PET and CT images. This technique increased the certainty of the diagnosis and yielded positive clinical impact in approximately a third of the cases. When a combined PET/CT scanner is not available, our manual fusion technique may be an effective alternative. However, because it is possible to miss liver and lung metastases even when using the fusion technique, CT images should be reviewed carefully with an optimal window level setting for the lung field, and liver results should also be evaluated carefully even when PET shows no significant metabolic activity.
Acknowledgments
We greatly appreciate Keiichi Matsumoto, Eiri Minota, and Keiji Shimizu for
their excellent technical support in scanning and Julia Buchanan for her
editorial assistance.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  Y. Nakamoto, K. Togashi, T. Kaneta, H. Fukuda, K. Nakajima, K. Kitajima, K. Murakami, H. Fujii, M. Satake, U. Tateishi, et al. Clinical Value of Whole-body FDG-PET for Recurrent Gastric Cancer: A Multicenter Study Jpn. J. Clin. Oncol., May 1, 2009; 39(5): 297 - 302. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. L. Bridges Software Fusion: An Option Never Fully Explored J. Nucl. Med., May 1, 2009; 50(5): 834 - 836. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
