There have still been many institutions implementing only dedicated SPECT. However, its inability to perform quantitative analysis reduces the value of bone SPECT, which may account for increased high-quality evidence of bone planar scintigraphy instead of bone SPECT. Recently, quantitative bone SPECT/CT was reported to provide more accurate assessment of bone metastasis than planar scintigraphy [
19], suggesting that bone SPECT with B-SAC could have better performance than planar scintigraphy when assessing the metastatic tumor burden. Furthermore, B-SAC may allow to estimate the delivered dose of therapeutic radionuclides such as Ra-223 to metastatic lesions on a per-lesion basis because of an easier evaluation of individual bone metastasis than planar scintigraphy. Therefore, an achievement of quantitation with dedicated SPECT would expand its clinical usability considering the fact that cancer is now a common disease.
Importance of attenuation correction for quantitation
We found that SUVs in all of the five evaluated images (B-SACN (+)SC(+)RR(+), B-SACM (+)SC(+)RR(+), UAC(+)SC(+)RR(+), AC(−)SC(+)RR(+), and NC) had substantial correlations with those in CTAC(+)SC(+)RR(+) images. However, B-SAC(+)SC(+)RR(+) revealed the best correlation with CTAC(+)SC(+)RR(+), and the absolute difference between B-SAC(+)SC(+)RR(+) and CTAC(+)SC(+)RR(+) images was much smaller than those between the others and CTAC(+)SC(+)RR(+) images. Even uniform attenuation correction, which is widely applied to quantitative brain SPECT, was found to be insufficient as a method of attenuation correction for bone SPECT. These results clearly suggested that an attenuation correction of bone as well as soft tissues plays an important role in the quantitation of bone uptake.
Segmentation according to pixel value in bone SPECT
A major advantage of using bone SPECT emission data for attenuation correction is that it is feasible to semi-automatically delineate bone structures which exhibit the highly accumulated areas on SPECT images. Photon attenuation is dependent on transmission medium (usually defined as a function of attenuation coefficient) inside the body through which photon travels into the detector. If the human body is roughly divided into three components according to photon attenuation (i.e., bone, soft tissue/water, and air), bone has the greatest impact on attenuation correction whereas the attenuation effect of air is negligible. Although the other tissues than bone or air have different attenuation coefficient values (in this study the coefficient for them was constant), the difference would be much smaller than the difference between bone and soft tissue/water, or than the difference between soft tissue/water and air. Based on this assumption, the fixed values replacing areas A (300 HU or 600 HU) and B (50 HU) for B-SAC were applied to every case in spite of the non-uniform human structure. In addition, metastatic bones had different CT values. Therefore, SUVs obtained with the current version of B-SAC seem difficult to be directly compared with those obtained with CTAC.
In the present study, the cutoff values for the segmentation on original SPECT images were based on the lung counts because of the following reasons. First, the contour of the lung is roughly identified on SPECT images because of its very low background uptake and surrounding uptake in the rib, liver, and mediastinum (Fig.
2). Second, SPECT count density in the lung seems constant depending on the VOI position within the lung (e.g., peripheral or deep side of the lung) even on non-attenuation-corrected SPECT images because of minimal attenuation effect within the lung. However, it remains uncertain whether the proposed method is most suitable for B-SAC. In addition, the threshold settings for discriminating areas A, B, and C were only based on our visual inspection of the bone SPECT images in the 55 patients with prostate cancer with (
n = 15) and without (
n = 40) bone metastasis, which is one of the weaknesses of this study.
In B-SAC, the urinary system, known as high uptake in bone SPECT, is recognized as one of the bone tissues. Therefore, the kidney, urinary tract, and bladder show as high SPECT values as the bone even after B-SAC. However, as far as we evaluated the cases in the present study, we have observed no measurement error due to high accumulation in non-bone tissues including the urinary system. However, it is possible to cause the error especially when extraosseous accumulation is very close to any bone structure.
Moreover, the tissue segmentation with the current B-SAC method is susceptible to image noise because of the lack of the contouring process. Therefore, image smoothing and/or longer scan duration seems important when performing the segmentation. However, considerable image filtering can lower SPECT values of both normal and metastatic bones, resulting in the potential failure of bone segmentation. It is therefore necessary to perform further research about the optimization of smoothing and/or scan duration to improve the accuracy of the B-SAC.
Despite these limitations, the results of this preliminary study suggest the potential for B-SAC to improve the quantitation of bone metastases in bone SPECT when X-ray CT or transmission CT data are not available.
Possible advantage of B-SAC over CTAC
B-SAC may contribute to the reduction of radiation dose compared to CTAC if quantitative bone uptake is a key to clinical assessments including differentiation of benign and malignant bone lesions, evaluation of active inflammatory bone diseases, and treatment evaluation of a variety of bone diseases. In pediatric patients, sports injuries, scoliosis, trauma, and bone tumors causing back pain are often evaluated with bone SPECT/CT or fluoride-18 PET/CT [
20], although the radiation exposure from low-dose CT would be a matter of debate [
21]. Indeed, MR-based segmented attenuation correction to pediatric PET/MR imaging has been specifically developed for generating attenuation maps without CT [
11].
Recently, a randomized controlled clinical trial succeeded in demonstrating that BSI in planar scintigraphy was a good prognostic indicator in patients with prostate cancer [
5]. Another large-scale clinical trial also showed that the volume of bone metastases affected the benefit of radiotherapy to the primary prostate cancer [
22]. Thus, the measurement of metastatic tumor burden is crucial for the management of prostate cancer patients, and BSI has been used as one of the biomarkers for this purpose. However, since the automatic extraction of metastatic lesions for calculating BSI is based on the image contrast between normal and metastatic lesions, it can cause an underestimation of metastatic tumor burden especially in case of diffuse bone metastases as a result of the reduced ability to differentiate between normal and pathologic uptake areas from anteroposterior planar images [
19]. On the other hand, TBU is a promising alternative to BSI because of both the improved image contrast by SPECT and the objective assessment of measuring lesional SUVs. We found that TBUs observed with B-SAC were slightly and constantly (17%) higher than those observed with CTAC when the same threshold was used. These results suggested that B-SAC is probably comparable to CTAC when comparing TBUs before and after treatment. It may further facilitate the comparison of SUVs among different dedicated SPECT scanners or between dedicated SPECT and SPECT/CT if the reconstruction condition and the threshold for tumor delineation are properly adjusted. However, at present, an appropriate application of B-SAC requires full validation through significant clinical data from many institutions.