ORIGINAL 
Annals of Nuclear Medicine Vol. 3, No. 3, 119-124, 1989 
Usefulness of pinhole collimator in differential diagnosis of metastatic disease and degenerative joint disease in the vertebrae ; Evaluation by receiver operating characteristics (ROC) analysis 
Shigeru KOSUDA, Syunji KAWAHARA, Akihiko ISHIBASHI, Kohei TAMURA, Yasushi TSUKATANI, Hiroshi FUJII, Atsushi KUBO and Shozo HASHIMOTO 
Department of Radiology, Okura National Hospital and Keio University Hospital, Tokyo 
In order to evaluate the diagnostic efficacy of pinhole collimator (PHC) imaging combined with an X-ray for vertebral metastasis, our prospective study has employed receiver operating characteristics (ROC) analysis in 21 patients, 11 with osseous metastasis and 15 with degenerative joint disease in the lumbar vertebrae. 
PHC imaging provided better anatomic information on the extent of 99mTc-MDP accumulation. PHC vertebral scintigraphy had a considerable impact on the decision-making process, although with variations and not very satisfactory results among the physicians with little experience. Our study suggests that PHC imaging and X-ray film are useful in differentiating between osseous metastasis and degenerative joint disease in the vertebra. 
Key words: pinhole collimator, bone scintigraphy, Receiver operating characteristics analysis, Bone metastasis, Degenerative joint disease 
INTRODUCTION 
IN AGED PATIENTS with suspected vertebral metastasis, it is often difficult to differentiate early osseous metastasis from degenerative joint disease in the vertebrae on bone scan. It is well known that a pinhole collimator (PHC) provides clearer definition of small areas and can delineate lesions otherwise barely visualized with a parallel hole collimator. 
The aim of our study is to evaluate the diagnostic efficacy of PHC imaging combined with X-raying for vertebral metastasis. The method used in our prospective study was receiver operating characteristics (ROC) analysis, to determine to what extent the usage of PHC in vertebral bone scan is able to improve diagnostic accuracy. 

Received March 14, 1989; revision accepted June 7, 1989. 
For reprints contact : Shigeru Kosuda, M.D., Department of Radiology, Tokyo Metropolitan Komagome Hospital, 3-18-22, Honkomagome, Bunkyo-ku, Tokyo 113, JAPAN. 

MATERIALS AND METHODS 
Bone scans with 99mTc-MDP were performed in 21 patients, 11 with osseous metastasis in the lumbar spine from various primary lesions (5 males, 6 females, mean age 64 yr) and 15 with degenerative joint disease in the lumbar vertebrae, that is, spondylosis defermans (6 males, 9 females, mean age 69 yr). In each case, the final diagnosis was established by biopsy or autopsy, and by reviewing each patient's medical record, including subsequent clinical course over a 2-yr period of time, and correlation with X-CT or other radiological studies. 
Posterior dual images were obtained at 3 hours after i.v. injection of 20 mCi (740 MBq) of 99mTc-MDP in all of the patients, with a low energy all-purpose parallel hole collimator (LEAP) and PHC. The LEAP has 42,000 holes and the PHC has a 6-mm aperture, with a 10 cm focal length. PHC imaging was made at a source-aperture distance of 5 to 10 cm and 500 Kcount were obtained. The gamma camera used was Toshiba GCA 401 model. 
The blinded observers, 1 junior resident (3 months' nuclear medicine experience), 1 senior resident (1 year's nuclear medicine experience), 1 nuclear medicine specialist, separately examined the films for each of the three different scitigraphic detection situations, that is, first, the spot film taken by LEAP only, second, the combination of spot films taken by LEAP and PHC, third, X-ray films in addition to the combination. The impact of PHC imaging and X-ray films on the physician's diagnostic confidence was evaluated by ROC analysis. The confidence ratings used to produce a ROC curve are shown in Table 1. 
Further, we assessed accumulation patterns in the vertebra obtained by PHC scintigraphy. The PHC findings were characterized in terms of localization, appearance and homogeneity of abnormal radionuclide accumulation. 
RESULTS 
The ROC curves obtained by three interpreters, a junior resident (Doctor A), a senior resident (Doctor B) and a specialist in nuclear medicine (Doctor C), are shown in Figs. 1, 2, 3, respectively. The area of each curve is shown on the top of each figure. 
The junior and senior residents' detection performances were relatively poor, and PHC imaging did not greatly affect the differential diagnosis between osseous metastasis and degenerative joint disease in the vertebra. It is difficult to evaluate the X-ray film's impact on the detection performances because the ROC curves cross. On the other hand, in the case of the specialist in nuclear medicine, the PHC information apparently makes a better detection performance (area under the ROC curve ; 0.83) possible, compared with the LEAP information only (area under the ROC curve ; 0.69). The information obtained from the X-ray makes possible an even better detection performance (area ; 0.98). 
Table 2 shows accumulation patterns for the vertebra, obtained from PHC scintigraphy. End-zone accumulation in the two opposing end-plates and board-like accumulation were relatively characteristic patterns for degenerative joint disease in the vertebrae. PHC imaging provided superior anatomic information of the extent of 99mTc-MDP accumulation (Figs. 4, 5, 6). 
DISCUSSION 
Bone scintigraphy is at present the most common imaging procedure used in nuclear medicine. The major advantage of the bone scan is its extreme sensitivity. Studies comparing bone scans with skeletal surveys find that about one-third to one-half of lesions noted on the scan are not seen in the conventional X-ray surveys ; the false-negative rate for bone scan is about 2%.1-5 Despite the spectacular growth of bone scanning, one major problem in osseous metastasis survey is that there is a considerable variation in the reported incidence of patients with primary malignancies.6-10 Especially a solitary lesion detected on the bone scan of a patient being assessed for osseous metastasis presents a major challenge to the diagnostic acumen of the physician. Miscellaneous benign bone lesions can frequently account for these focal areas of uptake. Every aged patient has more or less degenerative joint diseases in the vertebrae, so that early vertebral metastasis is often difficult to discriminate from degenerative joint disease by means of X-rays only. Several investigators reported that approximately 30 to 65% of solitary bone scan lesions in patients with known primary malignancies were found to be osseous metastasis.11,12 O'Connell13 reported that bone scan read as "equivocal" reached 22% of the total number interpreted. Also, interpreter variation can alter results. 
In these circumstances, these variations are probably due, for the most part, to difference in the scanning technique and recognition of patterns such as can be seen in degenerative joint disease. We are convinced that high quality image and precise and careful interpretation of bone scans are needed in differential diagnosis. 
PHC generally provides better resolution and greater magnification, although it has drawbacks such as lack of sensitivity, distortion of image and foreshortening. It is well known that PHC is particularly useful for the examination of the hips, the hands and wrists or a localized area. Bahk et al.14 published an excellent paper on the usefulness of PHC imaging of the spine. However, they did not implement quantitative assessment of PHC imaging. We consider that the usefulness of PHC imaging should be evaluated quantitatively by some detection theory or in a meaningful way. 
PHC imaging and X-ray film information did not have a significant impact on the decision-making process in the junior and senior residents. The fact that the area under the ROC curve of LEAP was less than 0.5 in the case of the junior resident is probably because he did not have a thorough understanding of the recognition patterns in bone scan and roentgenogram of both diseases. The detection of the senior resident was a little better than that of the junior resident, but PHC imaging and X-ray film information did not have a great impact on the decision-making process, either. The reason why there were crossings of the ROC curves seems to be that lesions with predominant uptake which read as osseous metastasis on the bone scans could not be clearly identified as an abnormal finding on the X-ray films. 
On the other hand, the detection performance of the nuclear medicine specialist was much better and PHC and X-ray film information apparently had a great impact on the decision-making process, and were capable of improving diagnostic accuracy. PHC and X-ray film information contributed, not to the increase in the true positive rate, but to the decrease in the false positive rate in the case of the nuclear medical specialist, whereas it contributed to a little increase in true positive rate in the residents. The results reveal that the use of PHC results in a noticeable improvement in anatomic and spacial resolution. In reviewing the results obtained by the junior and senior residents, we believe that it is very important to have a full knowledge of recognition patterns of diseases in the bone scan, based on considerable experience in nuclear medicine. For the specialist, the X-ray findings were also useful for differentiation, although it is often difficult to differentiate osseous metastasis from degenerative joint disease in patients with both early osseous metastasis and mild degenerative joint disease in the same vertebra. 
Another scintigraphic detection situation, that is, the combination of spot films taken by LEAP and X-ray was not evaluated in our study. In view of our ROC curve results (Figs. 1, 2, 3), we believe that X-ray film information has some degree of impact on the decision-making process, not for a resident but for a specialist. 
In conclusion, PHC imaging provided better anatomic information on the extent of 99mTc-MDP accumulation. PHC vertebral scintigraphy had a considerable impact on the decision-making process, although with variations and not very good results by physicians with little experience. Our study suggests that PHC imaging and X-rays are useful for differentiating between osseous metastasis and degenerative joint disease in the vertebra. 
REFERENCES 
1. McNeil BJ : Rationale for the use of bone scans in selected metastatic and primary bone tumors. Semin Nucl Med 8 : 336-345, 1978 
2. Belliveau RE, Spencer RP: Incidence and sites of bone lesions detected by 99mTc polyphosphate scans in patients with tumors. Cancer 36: 359-363, 1975 
3. Kirkman S, Henk JM: The value of bone scanning in the staging of breast cancer. Clin Radiol 30: 11-14. 1979 
4. McKillop JH, Blumgart LH, Wood CB, et al: The prognosis and therapeutic implications of the positive radionuclide bone scan in clinically early breast cancer. Br J Surg 65 : 649-652, 1978 
5. Silberstein EB, Saenger E, Tofe AJ, et al : Imaging of bone metastases with 99mTc-Sn-EHDP (diphosphonate), F-18, and skeletal radiography. Radiology 107 : 551-555, 1973 
6. Nolan NG, Koppikar MM, Kotlyarov EV: Role of bone scanning in carcinoma of the breast. Ann Clin Lab Sci 10: 105-110, 1980 
7. Hahn P, Vikterlof KJ, Rydman H, et al : The value of whole body bone scan in the preoperative assessment in carcinoma of the breast. Eur J Nucl Med 4: 207-210, 1979 
8. Front D, Schneck SO, Frankel A, et al: Bone metastases and bone pain in breast cancer. Are they closely associated? JAMA 242: 1747-1748, 1979 
9. Clark DG, Painter RW, Sziklas JJ : Indication for bone scans in preoperative evaluation of breast cancer. Am J Surg 135 : 667-670, 1980 
10. Baker RR: Preoperative assessment of the patient with breast cancer. Surg Clin North Am 58 : 681-691, 1978 
11. Corcoran RJ, Thrall JH, Kyle RW, et al : Solitary abnormalities in bone scans of patient with extraosseous malignancies. Radiology 121 : 663-667, 1976 
12. Coliins JD, Bassett L, Main GD, et al : Percutaneous biopsy following positive bone scans. Radiology 132: 439-442, 1979 
13. O'Connell JM, Wahner HW, Ahmann DL, et al: Value of preoperative radionuclide bone scan in suspected primary breast carcinoma. Mayo Clin Proc 53: 221-226, 1978 
14. Bahk YW, Kim OH, Chung SK: Pinhole collimator scintigraphy in differential diagnosis of metastasis, fracture, and infections of the spine. J Nucl Med 28 : 447-451, 1987