MRI of Bone Tumors: Fast STIR Imaging as a Substitute for T1-Weighted Contrast-Enhanced Fat-Suppressed Spin-Echo Imaging

Transcription

1 JOURNAL OF MAGNETIC RESONANCE IMAGING 19: (2004) Original Research MRI of Bone Tumors: Fast STIR Imaging as a Substitute for T1-Weighted Contrast-Enhanced Fat-Suppressed Spin-Echo Imaging Osamu Tokuda, MD, Noriko Hayashi, MD, and Naofumi Matsunaga, MD* Purpose: To compare the usefulness of short inversion recovery (STIR) and T1-weighted, contrast-enhanced, fatsuppressed (T1W-CEFS) sequences for the evaluation of bone tumors. Materials and Methods: Eighteen patients with 19 bone tumors who underwent both STIR and T1W-CEFS imaging were evaluated. The tumors were categorized in pairs as follows: bone marrow and soft-tissue components, benign and malignant tumors, and tumors with and without mineralization. The signal difference-to-noise ratio (SDNR), signal-to-noise ratio (SNR), and tumor volume were calculated in each group. An additional qualitative analysis was performed by means of the ratings of imaging contrast. Results: The mean SDNRs of all bone marrow components and bone marrow components without mineralization were significantly higher on fast STIR images than on T1W-CEFS images (P 0.05). There was no significant difference in the mean SDNR and SNR of the other group (surrounding soft tissue components, bone marrow components with mineralization, benign and malignant lesions) between fast STIR images and T1W-CEFS images. The mean volume of the tumors was significantly higher with STIR than with the T1W-CEFS sequence (P 0.05). Conclusion: The STIR sequence should be used instead of T1W-CEFS imaging for the evaluation of bone tumors. Key Words: STIR; contrast-enhanced; bone tumor; SDNR; SNR J. Magn. Reson. Imaging 2004;19: Wiley-Liss, Inc. PREVIOUS STUDIES HAVE SHOWN that magnetic resonance imaging (MRI) is useful for evaluating musculoskeletal lesions, including bone and soft-tissue tumors (1 12), cartilaginous injuries (13), and bone marrow disorders (14,15). However, a consensus has not yet been reached as to which MRI technique is best for evaluating lesions. The most promising sequences are those that use some form of fat suppression. Two of the most widely available methods are short inversion recovery (STIR) and fat saturation (15). The STIR technique uses an alternative MRI sequence that suppresses signals from fat and the additive effects of T1 and T2 mechanisms on tissue brightening (3). STIR imaging is commonly used to detect bone marrow lesions because it is a sensitive technique for detecting tumor, edema, and infection in bone marrow (15). Gadolinium diethylenetriamine-pentaacetic acid (DTPA) produces increased signal intensity (SI) in T1- weighted (T1W) spin-echo (SE) images by reducing the T1-relaxation time (4). It has been demonstrated that zones that show a marked increase in SI are correlated with highly vascularized regions, whereas zones with low or no increase in SI are often correlated with necrotic areas (1,4,9,12). T1W contrast-enhanced fatsuppressed (T1W-CEFS) imaging, which combines the use of gadolinium-dtpa and a fat saturation technique, is a superior method for evaluating adrenal masses (16), and neoplastic and inflammatory disease of the spine (17,18), kidney (19), head, and neck (20,21). T1W-CEFS imaging is also a very useful technique for musculoskeletal disorders (1,12,22). However, to our knowledge, the value of T1W-CEFS images in comparison with STIR images for depicting bone tumors has not been formally evaluated. The utility of STIR and T1W-CEFS images for evaluating bone and surrounding soft-tissue tumors was investigated in the present study. Specifically, the signaldifference-to-noise ratio (SDNR) (14,23) and signal-tonoise ratio (SNR) obtained on STIR images were compared with those obtained on T1W-CEFS images. Additionally, the volumes of abnormality detected on STIR images were compared with those found on T1W- CEFS images. A qualitative analysis was performed by two musculoskeletal radiologists. Department of Radiology, Yamaguchi University School of Medicine, Yamaguchi, Japan. *Address reprint requests to: N.M., Department of Radiology, Yamaguchi University School of Medicine, Minamikogushi, Ube, Yamaguchi , Japan. skier@yamaguchi-u.ac.jp Received April 22, 2003; Accepted February 15, DOI /jmri Published online in Wiley InterScience ( MATERIALS AND METHODS Patient Population Eighteen patients (eight males and 10 females, years old, average age 44.9 years) who underwent MR examinations, including fast STIR and T1W-CEFS im Wiley-Liss, Inc. 475

2 476 Tokuda et al. Figure 1. Fibrous dysplasia of the proximal left fibula in a 13-year-old female. a: Coronal fast STIR image (TR/TE/TI/ETL 6100/60/100/10 msec) shows a well-defined decreased signal mass (arrow) with a surrounding ill-defined increased signal. b: Coronal T1W-CEFS image (TR/TE 800/12 msec) shows a well-defined inhomogeneous marginal enhancement (curved arrow) with a surrounding ill-defined strong enhancement. aging, were prospectively evaluated. Histologic diagnoses revealed a total of 19 bone tumors (five benign tumors and 14 malignant tumors). The benign tumors consisted of enchondroma (N 1), non-ossifying fibroma (N 1), solitary bone cyst (N 2), and fibrous dysplasia (N 1) (Fig. 1). The malignant tumors consisted of leukemia (N 1), adamantinoma (N 1), osteosarcoma (femur (N 2) (Fig. 2) and radius (N 1)), metastatic bone tumor of the hepatocellular carcinoma (femur (N 1), ischium (N 1) and scapula (N 1)), renal cell carcinoma (humerus (N 1)), thyroid cancer (femur (N 1)), gingival carcinoma (femur (N 1)), rectal cancer (pubis (N 1)), and lung cancer (femur (N 1) and humerus (N 1)). Of the 19 tumors, seven (fibrous dysplasia (N 1) (Fig. 1), adamantinoma (N 1), enchondroma (N 1), osteosarcoma (N 3) (Fig. 2), and metastatic bone tumor of lung cancer (humerus (N 1)) had mineralization in the bone marrow compo- Figure 2. Osteosarcoma of the distal right femur in a 14-year-old female. a: Sagittal fast STIR image (TR/TE/TI/ETL 6100/60/100/10 msec) shows an ill-defined, inhomogeneous, slightly increased signal mass, with involvement of soft tissue. Extremely decreased lesions (thin arrow), corresponding to mineralization, are found in the bone marrow of the femur. Peritumoral brightening (thick arrow) is seen adjacent to the soft-tissue tumor. b: Sagittal T1W-CEFS image (TR/TE 800/12) shows an inhomogeneous, intensely enhancing tumor, with involvement of soft tissue (curved arrow). The lesions of peritumoral brightening on the STIR image are mostly enhanced following gadolinium administration.

3 MRI of Bone Tumors 477 nent, and 12 had no mineralization. Mineralization included both ossification and calcification. In addition, 10 surrounding soft-tissue components (leukemia (N 1), osteosarcoma (N 3) (Fig. 2), metastatic tumors of gingival carcinoma (N 1), hepatocellular carcinoma (N 2), lung cancer (N 2), and rectal cancer (N 1)) were included in the present study. In one case, a diagnosis of non-ossifying fibroma was made because plain radiography revealed an eccentric location, geographic bone destruction, radiolucent lesion with lobulated trabeculation, and cortical expansion of the tibia, and no increase in the size of the tumor was found on 6-month follow-up plain radiography and CT. In two cases, a diagnosis of solitary bone cyst was made because plain radiography showed a centrally located, well-marginated, radiolucent metaphyseal lesion with cortical thinning; T1W MRI revealed a homogeneous decreased SI; T2-weighted (T2W) MRI revealed a homogeneous, extremely increased SI at the site of the lesions; and there was no increase in size on the 6-month follow-up MR images. One case of enchondroma, fibrous dysplasia, and adamantinoma; three cases of osteosarcoma; and one case of metastatic humeral tumor of lung cancer (which resulted in pathological fracture) were pathologically confirmed at operation. In the case of metastatic humeral tumor from lung cancer, calcific stippling, destruction of the cortical bone, and soft-tissue component was found on CT and MR images. The humeral lesion was a collision metastasis of lung cancer (adenocarcinoma) to an enchondroma (24). One case of leukemia was diagnosed on the basis of the decreased size of the soft-tissue component and improved abnormal SI in the bone marrow on 3-month follow-up MR images, obtained after chemotherapy. Diagnoses of metastatic bone tumor in one case of thyroid cancer, and three cases of hepatocellular carcinoma were made on the basis of the findings of multiorgan metastasis, and were confirmed pathologically at other metastatic sites (supraclavicular lymph node (one case of thyroid cancer) and rib (two cases of hepatocellular carcinoma)). A diagnosis of a metastatic femoral tumor of lung cancer was made on the basis of the findings of multiple lung metastases on follow-up MR images (performed two months after the initial examination). Diagnoses of the other metastatic bone tumors were made on the basis of the apparently increased size of the tumors (both primary and metastatic lesions) on follow-up CT and MR images (performed 1 6 months after the initial examination). MRI MR images were obtained with the use of a 1.5 T superconducting MR system (Signa Horizon, GE Medical Systems, Milwaukee, WI). In general, T1W SE (repetition time (TR) msec, effective echo time (TE) msec) and T2W fast SE (TR/TE /96 msec) imaging was performed in two planes. In addition, a fast SE STIR sequence (TR/TE / msec, inversion time 150 msec, echo train length (ETL) 10, field of view (FOV) mm, matrix size , section thickness 3 5 mm, and intersection gap mm) was performed in the coronal plane. There were four excitations, and the acquisition time for the fast STIR sequence varied between 5 minutes 12 seconds and 5 minutes 42 seconds. All patients agreed to receive an additional intravenous injection of contrast material. A T1W SE sequence (TR/TE /10 12) with fat suppression was performed after intravenous administration of 0.1 mmol/kg of body weight of gadopentetate dimeglumine (Magnevist; Schering, Berlin, Germany) in two planes. The section thickness was 4 5 mm and the intersection gap was mm. The FOV ( mm) and image matrix ( ) were identical to those used in the fast STIR sequence. There were two excitations, and the acquisition time for the fat-suppressed T1W sequence varied between 3 minutes 30 seconds and 5 minutes 12 seconds. Imaging Analysis The SDNR signals of the bone marrow and surrounding soft-tissue components were measured for both sequences as follows: In bone marrow abnormalities, a region of interest (ROI) was placed within the zone of abnormal signals of bone tumors and in the nearest normal-appearing part of the bone marrow. In contrast, in surrounding soft-tissue components, an ROI was placed within the zone of increased signals of the surrounding soft-tissue components and in the nearest most normal-appearing part of the soft tissues. The ROI within the area of bone marrow and surrounding softtissue components, and the ROI within the normalappearing part of bone marrow and soft tissues were identical in size. The SDNR was calculated as follows: SDNR (The mean SI of the abnormal lesion The mean SI of the nearest normal-appearing part of the abnormal lesion)/ SD of air near the abnormal lesion The SNR of the bone tumors was measured for both sequences by the following method: In the bone marrow component, an ROI was placed within the zone of increased signal of the bone marrow component. In contrast, in the surrounding soft-tissue component, an ROI was placed within the zone of increased signals of the surrounding soft-tissue component and in the nearest most normal-appearing part of the soft tissue. The SNR was calculated as follows: SNR The mean SI of the abnormal lesion/ SD of air near the abnormal lesion The SDNR and SNR of the bone marrow and surrounding soft-tissue components were also calculated separately and compared by means of a paired t-test. A P-value of 0.05 indicated a significant difference. We determined the volumes of the zones of increased signals of the bone marrow and surrounding soft-tissue components by measuring the coronal areas within every section and multiplying them by the section thickness plus the intersection gap. Both measurements

4 478 Tokuda et al. Table 1 SDNR Comparisons of STIR Vs. T1W-CEFS Images in 19 Bone Tumors STIR sequence T1W-CEFS sequence P value Bone marrow components (N 19) Surrounding soft tissue components (N 10) Bone marrow components with mineralization (N 7) Bone marrow components without mineralization (N 12) Benign lesions (N 5) Malignant lesions (N 14) Values are means SD. T1W-CEFS Sequence T1 weighted contrast-enhanced fat-suppressed sequence. were performed manually by one musculoskeletal radiologist (O.T.) using the standard MR console (Image VINS Pro; Yokogawa Electric Corp., Musashino, Tokyo, Japan). We then compared the volumes of the zones of increased signals of the bone marrow and surrounding soft-tissue components using a paired t-test. A P-value of 0.05 indicated a significant difference. An additional qualitative analysis was performed by two musculoskeletal radiologists (O.T. and N.H.). Both observers were blinded to the clinical data. Subjective imaging contrast (differentiation between the bone marrow component and normal-appearing part of the bone marrow, and the surrounding soft-tissue component and normal-appearing part of the soft tissue) was rated on a scale of 1 4 (1 poor, 2 acceptable, 3 good, 4 excellent). Furthermore, the observers classified benign and malignant tumors by the use of a five-point scale to assign a confidence level (1 definitely benign, 2 probably benign, 3 equivocal, 4 probably malignant, 5 definitely malignant). Two readings were performed. First, both observers analyzed T1W SE images together with fast STIR images. After two weeks, a second session was performed. During this second session, the T1W SE images were reviewed together with T1W-CEFS images. The ratings of both images were compared by means of a post-hoc test. A P-value of 0.05 indicated a significant difference. The confidence level ratings of the images were also used to calculate the sensitivity, specificity, and accuracy of each observer in diagnosing malignancy by the MR images. Ratings of 1 or 2 indicated a reading of a benign tumor, and ratings of 4 or 5 indicated a reading of a malignant tumor. A rating of 3 indicated incorrect readings. The sensitivity, specificity, and accuracy of the STIR images were compared with those of the T1W- CEFS images by the use of a chi-square for independence test. A P-value of 0.05 indicated a statistically significant difference. To assess inter- and intraobserver variability in the assignment of a confidence level to the lesion status, the weighted value (25) was calculated. The level of agreement was defined as follows: 0 indicated no agreement, indicated poor agreement, represented good agreement, and represented excellent agreement. RESULTS Quantitative Analysis The mean SDNR of the bone marrow components was significantly higher on the fast STIR images than on the T1W-CEFS images (P 0.05) (Table 1). However, there was no significant difference in the mean SDNR of the surrounding soft-tissue components between the fast STIR images and the T1W-CEFS images (Table 1). There were no significant differences in the mean SNR of the bone marrow components and the mean SNR of the surrounding soft-tissue components between the fast STIR images and the T1W-CEFS images. The mean SDNR of the bone marrow components without mineralization was significantly higher on images obtained with fast STIR than on those obtained with T1W-CEFS imaging (P 0.05) (Table 1). However, there was no significant difference in the mean SDNR of bone marrow components with mineralization between fast STIR and T1W-CEFS (Table 1). There were no significant differences in the mean SNR of the bone marrow components with and without mineralization between fast STIR and T1W-CEFS. There was no significant difference in the mean SDNR and SNR of the benign and malignant bone marrow components between fast STIR and T1W-CEFS. The mean volume of the bone marrow and surrounding soft-tissue components was significantly higher with the fast STIR sequence (mean cm 3 ; SD ) than with the T1W-CEFS sequence (mean cm 3 ;SD , P 0.05). Qualitative Analysis Observer ratings of STIR and T1W-CEFS images were compared in 19 bone tumors (Table 2). In bone tumors with mineralization, there were no significant differences between the mean ratings by both observers on images obtained with the fast STIR and T1W-CEFS sequences. In bone tumors without mineralization, the mean ratings by observer 2 were significantly higher on images obtained with fast STIR compared to the T1W- CEFS sequence (P 0.05). However, there were no significant differences in the mean ratings by observer 1 between images obtained with the fast STIR and T1W- CEFS sequences. In benign and malignant tumors, there were no significant differences in the mean ratings by both observers between the fast STIR and T1W-CEFS images. In all tumors, the mean ratings by observer 2 were significantly higher on images obtained with fast STIR compared to T1W-CEFS (P 0.05). However, there were no significant differences in the mean ratings by observer 1 between the fast STIR and T1W-CEFS images. In all images, the weighted value of both the STIR and T1W-CEFS images was These findings indicate

5 MRI of Bone Tumors 479 Table 2 Observer Rating of STIR Vs. T1W-CEFS Images in 19 Bone Tumors Observer 1 Observer 2 STIR T1W-CEFS P value STIR T1W-CEFS P value Bone tumors with mineralization (N 7) Bone tumors without mineralization (N 12) Benign tumors (N 5) Malignant tumors (N 14) All tumors (N 19) Values are means SD. T1W-CEFS Sequence T1 weighted contrast-enhanced fat-suppressed sequence. that interobserver agreement was excellent. The weighted value for observers 1 and 2 was These findings suggest that intraobserver agreement was also excellent. The sensitivity, specificity, and accuracy for the differentiation of benign from malignant tumors on fast STIR and T1W-CEFS images for both observers were calculated (Table 3). The mean sensitivity, specificity, and accuracy for both observers using the fast STIR sequence were 78.6%, 70%, and 76%, respectively. For the T1W-CEFS sequence, these values were 85.8%, 40%, and 73.7%, respectively. In all groups, there were no significant differences in mean sensitivity, specificity, and accuracy. In all images, the weighted value was 0.88 for the fast STIR images, and 0.85 for the T1W-CEFS images. These findings suggest that interobserver agreement was excellent. The weighted value was 0.88 for observer 1, and 0.96 for observer 2. These findings suggest that intraobserver agreement was also excellent. DISCUSSION The T1W-CEFS sequence has become one of the standard pulse sequences used to examine musculoskeletal (1,12,22), head and neck (20), renal (19), and adrenal lesions (16) because it can depress the high SI of fat and improve the detection of lesions that are in close contiguity. However, several studies (26,27) reported that frequency-selective, fat-suppressed conventional spinecho sequences resulted in suboptimal signal loss. This was attributed to a loss of signal from magnetic susceptibility artifacts, poor delineation of anatomic detail, Table 3 Differentiation of Benign fom Malignant Tumors on STIR and T1W-CEFS Images Observer 1 Observer 2 Mean STIR Sensitivity Specificity Accuracy T1W-CEFS Sensitivity Specificity Accuracy Values are percentages. and reduced tumor conspicuity in T2W imaging with fat saturation (26). Other reasons why frequency-selective, FS conventional spin-echo sequences result in suboptimal signal loss have been reported as follows: First, adipose tissue is partially composed of water protons, which will not be suppressed. Second, olefinic fats, which comprise 5% of the lipid present within adipose tissue, have resonant frequencies similar to those of water. The olefinic protons will not be suppressed. Finally, because of T1 relaxation after the saturation pulse and magnetic field inhomogeneity, some other lipid protons may not be suppressed and will create inhomogeneous signals (27). Most abnormal tissues have very high SI on STIR images, which have a combination of T1 and T2 contrasts. Additional advantages of STIR imaging include the suppression of fat signal from surrounding tissues, and the elimination of chemical shift artifacts at the interfaces between fat and water (3). According to May et al (12), the routine use of gadolinium in the initial MR evaluation of a possible primary musculoskeletal neoplasm is not justified. Hence, if the value of STIR imaging is equivalent or superior to that of T1-CEFS imaging for the evaluation of bone tumors, intravenous injections could be abbreviated, resulting in time and cost savings. According to Schmid et al (14), the mean volume of increased bone marrow abnormalities was slightly greater with fast STIR than with T1W-CEFS imaging. However, in the present study, the mean volumes of the bone marrow and surrounding soft-tissue components on images obtained with the fast STIR sequence were significantly greater than those obtained with the T1W- CEFS sequence. This may be because the present analysis differed from previous studies in that both the volumes of zones of bone marrow components and the soft-tissue components surrounding the bone were measured. This discrepancy may also be explained by the difference in patient populations between the two studies. The present study included 14 malignant bone tumors, whereas no malignant bone tumors were included in the study by Schmid et al (14). The mean volumes of the bone marrow and surrounding softtissue components on images obtained with the STIR sequence might be significantly greater than those obtained with T1W-CEFS sequence, since bone marrow edema tends to be primarily caused by malignant bone

6 480 Tokuda et al. tumors. Benign lesions (chondroblastoma, active infection) can also cause bone marrow edema. The margins of bone marrow abnormalities on the STIR images tend to be overestimated compared to those on T1W SE images (28). One can assume that the apparent increase in the musculoskeletal lesion extent was likely due to peritumoral edema; however, peritumoral brightening, which was demonstrated on both the STIR and T1W-CFSE images, may include not only the peritumorous edema but also viable tumors (3). According to Schmid et al (14), the mean SDNR of increased bone marrow abnormalities was higher on images obtained with T1W-CEFS than on those obtained with the STIR sequence. They suggested that the higher noise on images obtained with the STIR sequence was mainly caused by the lower number of excitations obtained with the STIR sequence compared to the T1W-CEFS sequence. In addition, the greater difference in the SI between the bone tumors and the normal-appearing part of the bone marrow obtained with the STIR vs. the T1W-CEFS sequence may influence the findings. In contrast, the mean SNRs of the groups of bone tumors, surrounding soft-tissue components, bone tumors with mineralization, and malignant lesions tended to be higher on images obtained with the T1W-CEFS compared to the fast STIR sequence, although there was no significant difference. The lower SNR with STIR compared to the SE sequences is one of the inevitable disadvantages of the STIR sequence. In the present study, the mean SDNR of bone marrow components without mineralization was significantly higher on images obtained with the fast STIR (81.54) than on those obtained with the T1W-CEFS sequence. However, when the SDNR of the bone tumors with mineralization on images obtained with STIR and those obtained with the T1W-CEFS sequence were compared, there was no significant difference. The T1W-CEFS sequence may be more useful than the fast STIR sequence for the evaluation of bone marrow components with mineralization, such as enchondroma, fibrous dysplasia (Fig. 1), adamantinoma, and osteosarcoma (Fig. 2). To our knowledge, no previous study has compared the STIR and T1W-CEFS sequences for the differentiation of benign from malignant bone tumors with and without surrounding soft-tissue components. In the present study, the specificity for the T1W-CEFS sequence by both observers was 40%; however, this may be explained by the small number of benign lesions in the present analysis. The mean sensitivity and accuracy for both sequences were almost identical, and there were no significant differences between benign and malignant lesions in terms of sensitivity, specificity, and accuracy. The fast STIR and T1W-CEFS sequences may be comparable for the differentiation of benign from malignant tumors. In the present study, the mean rating of subjective imaging contrast of bone tumors without mineralization and benign lesions using the fast STIR sequence by observer 2 was significantly higher compared to the T1W-CEFS sequence. However, there were no significant differences when the mean ratings of subjective imaging contrast of the other parts (all tumors, bone tumors with mineralization, malignant tumors) using fast STIR and T1W-CEFS sequences were compared. Interobserver and intraobserver agreement in the ratings of subjective imaging contrast was excellent, and this comparable agreement suggested that one sequence did not have an advantage over the other. The STIR sequence has significant disadvantages. First, the SNR is usually significantly lower with the STIR than with the SE sequence (3). In the present study, the mean SNR of both bone marrow and surrounding soft-tissue components on STIR was lower than that on the T1W-CEFS images. However, this is less of a problem in bone tumors, because most bone tumors are demonstrated in the extremities and vertebrae, and there is less motion-induced noise than in images of the abdomen. Second, the STIR sequence requires relatively long imaging times to acquire a limited number of imaging slices, resulting in vulnerability to motion artifacts and poor SNR (5). Third, STIR imaging is also limited in its ability to provide anatomic detail, because common landmarks that are usually outlined by fat planes may be difficult to identify (3). However, this disadvantage can be overcome by the additional use of T1W with STIR images for the evaluation of the bone marrow lesions, because T1W images are superior to STIR images in providing anatomic detail. Finally, any nonlipid tissue will show suppression if it has a short-t1 value similar to that of fat, because fat suppression on STIR images is based strictly on relaxation values (5). This advantage is the most important. Although the STIR sequence has several disadvantages, it is clearly a very useful technique for the evaluation of bone and soft-tissue tumors. The STIR sequence is routinely used by many radiologists to evaluate bone tumors. A further important point is that frequency-selective, fat-suppressed sequences require the use of high-field-strength systems, whereas the STIR sequence can be performed with low- or highfield-strength units (29). In conclusion, the fast STIR technique is an excellent method for evaluating bone marrow and the surrounding soft-tissue components, and contrast enhancement should not be routinely performed. In the present study, the STIR and T1W-CEFS sequences showed comparable results in differentiating benign from malignant tumors. Therefore, since the STIR technique offers time and cost savings, and causes less patient discomfort, it should be used instead of T1W-CEFS for the evaluation of bone tumors. REFERENCES 1. Gronemeyer SA, Kauffman WK, Rocha MS, Steen RG, Fletcher BD. Fat-saturated contrast-enhanced T1-weighted MRI in evaluation of osteosarcoma and Ewing sarcoma. J Magn Reson Imaging 1997;7: Tamada T, Nagai K, Iizuka M, et al. Comparison of whole-body MR imaging and bone scintigraphy in the detection of bone metastasis from breast cancer. Nippon Acta Radiologica 2000;60: Shuman WP, Patten RM, Baron RL, Liddell RM, Conrad EU, Richardson ML. Comparison of STIR and spin-echo MR imaging at 1.5T in 45 suspected extremity tumors: lesion conspicuity and extent. Radiology 1991;179: Erlemann R, Reiser MF, Peters PE, et al. Musculoskeletal neoplasms: static and dynamic Gd-DTPA-enhanced MR imaging. Radiology 1989;171:

7 MRI of Bone Tumors Mirowitz SA. Fast scanning and fat-suppression MR imaging of musculoskeletal disorders. AJR Am J Roentgenol 1993;161: Beltran JB, Simon DC, Katz WK, Weis LD. Increased MR signal intensity in skeletal muscle adjacent to malignant tumors: pathologic correlation and clinical relevance. Radiology 1987;162: Stimac GK, Porter BA, Olison DO, Gerlach R, Genton M. Gadolinium-DTPA-enhanced imaging of spinal neoplasms: preliminary investigation and comparison with unenhanced spin-echo and STIR sequences. AJR Am J Roentgenol 1988;151: Steinborn MM, Heuck AF, Tiling R, Bruegel M, Gauger L, Reiser M. Whole-body bone marrow MRI in patients with metastatic disease to the skeletal system. J Comput Assist Tomogr 1999;23: Herrlin K, Ling LB, Pettersson H, Willen H, Rydholm A. Gadolinium- DTPA enhancement of soft tissue tumors in magnetic resonance imaging. Acta Radiol 1990;31: Golfieri BR, Baddeley H, Pringle JS, Souhami MR. The role of the STIR sequence in magnetic resonance imaging examination of bone tumors. Br J Radiol 1989;63: Flickinger FW, Sanal SM. Bone marrow MRI: techniques and accuracy for detecting breast cancer metastasis. Magn Reson Imaging 1994;12: May DA, Good RB, Smith DK, Parsons TW. MR imaging of musculoskeletal tumors and tumor mimickers with intravenous gadolinium: experience with 242 patients. Skel Radiol 1997;26: Ba-Salamah A, Schibany N, Puig S, et al. Imaging articular cartilage defects in the ankle joint with 3D fat-suppressed echo planar imaging: comparison with conventional 3D fat-suppressed gradient echo imaging. J Magn Reson Imaging 2002;16: Schmid MR, Holder J, Vienne P, Binkert CA, Zanetti M. Bone marrow abnormalities of foot and ankle: STIR versus T1-weighted contrast-enhanced fat-suppressed spin-echo MR imaging. Radiology 2002;224: Mirowitz SA, Apicella P, Reinus WR, Hammerman AM. MR imaging of bone marrow lesions: relative conspicuousness on T1-weighted, fat-suppressed T2-weighted, and STIR images. AJR Am J Roentgenol 1993;162: Semelka RC, Shoenut JP, Lawrence PH, et al. Evaluation of adrenal masses with gadolinium enhancement and fat-suppressed MR imaging. J Magn Reson Imaging 1993;3: Tien RD, Olson EM, Zee CS. Disease of the lumbar spine: findings on fat-suppression MR imaging. AJR Am J Roentgenol 1992;159: Georgy BA, Hesselink JR. Evaluation of fat-suppression in contrast-enhanced MR of neoplastic and inflammatory spine disease. AJNR Am J Neuroradiol 1994;15: Semelka RC, Hricak H, Stevens SK, Finegold R, Tomei E, Carroll PR. Combined gadolinium-enhanced and fat-saturation MR imaging of renal masses. Radiology 1991;178: Barakos JA, Dillon WP, Chew WM. Orbit, skull base, and pharynx: contrast-enhanced fat-suppression MR imaging. Radiology 1991; 179: Ross MR, Schomer DF, Chappell P, Enzmann DR. MR imaging of head and neck tumors: comparison of T1-weighted fat-suppressed images with conventional T2-weighted and fast spin-echo T2- weighted images. AJR Am J Roentgenol 1994;163: Morrison WB, Schweitzer ME, Bock GW, et al. Diagnosis of osteomyelitis: utility of fat-suppressed contrast-enhanced MR imaging. Radiology 1993;189: Wolff SD, Balaban RS. Assessing contrast on MR images. Radiology 1997;202: Galvin DH, Hornstein AM, Levine SM, Terrek RM. Collision of a metastatic carcinoma to an enchondroma. Orthopedics 2000;23: Fleiss JL. The measurement of interrater agreement. In: Statistical methods for rates and proportions. 2nd ed. New York: Wiley; p King AD, Lam WW, Leung SF, et al. Comparison of T2-weighted fat suppressed turbo spin-echo and contrast enhanced T1-weighted spin-echo MRI in nasopharyngeal carcinoma. Br J Radiol 1997;70: Hirsch JA, Loevner LA, Yousem DM, et al. Gadolinium-enhanced fat-suppressed T1-weighted imaging of the head and neck: comparison of gradient and conventional SE sequences. J Comput Assist Tomogr 1998;22: Jones KM, Unger EC, Granstrom P, Seeger JF, Camoby RF, Yoshino M. Bone marrow imaging using STIR at 0.5 and 1.5T. Magn Reson Imaging 1992;10: Tien RD. Fat-suppression MR imaging in neuroradiology: techniques and clinical application. AJR Am J Roentgenol 1992;158: