magnetic resonance imaging of intracranial neoplasms at 0.02 tesla

5
A C T A A D I 0 L 0 G I C A Vol. 28 1987 Fasc. 4 July-August MAGNETIC RESONANCE IMAGING OF INTRACRANIAL NEOPLASMS AT 0.02 TESLA M. KORMANO, R. RAININKO and K. KATEVUO Abstract Sixty-five patients with single or multiple intracranial neo- plasms were examined with an MR imager operating at 0.02 tesla. In 56 patients the diagnosis was histologically confirmed. All patients had an abnormal CT finding. MR images were positive in 59 cases, while the lesion remained undetected or equivocal in 6 cases (2 pituitary adenomas and 4 meningiomas). The MR signal intensity of several meningiomas was equal to that of normal brain tissue. Some astrocytomas were better delineated on MRI than on CT. For the study of pituitary lesions, the spatial resolu- tion was unsatisfactory. The ultralow field MR imager was found to be sensitive for the detection of other intracranial neoplasms. Key words: Brain, MR studies; -, neoplasms; magnetic reso- nance, comparative studies; -, technique; computed tomogra- phy, comparative studies. Since the introduction of magnetic resonance imaging (MRI) into clinical practice, several reports have been published on its value in the diagnosis of intracranial tumors (4-6, 8, 9, 11, 13, 15, 18, 21). MRI has been considered a useful method, especially in patients with a suspected brain stem tumor. The optimum field strength in clinical MRI is still a controversial question. In most reports field strengths from 0.15 to 0.35 tesla (T) have been used. In a concise material (2,3), results obtained with field strengths of 0.12 or 1.5 T were compared; higher field strengths resulted in better anatomic delineation. To our knowledge 0.02 T is the lowest magnetic field strength used in clinical MRI. Our experience with MRI at 0.02 T in the diagnosis of intracranial neoplasms is now described and the results are compared with the findings at computed tomography (CT). Material and Methods The material consisted of 65 patients (33 men and 32 women, aged 18 to 77 years, median 56) with single or multiple intracranial neoplasms, referred for CT or MRI between May 1984 and September 1985. The patients were not premedicated or anesthetized. The MR imager (Acutscan, Instrumentarium, Finland) has a resistive magnet and operates at a field strength of 0.02 T. The slice thickness is 10 mm, and the imaging matrix is 256X 128. The field of view of the head coil is 30 cm allowing a spatial resolution of 3 to 5 mm. Multislice axial spin echo technique with TR 2000 ms and TE from 130 to 160 ms and with two acquistions was routinely used. Sagittal or coronal slices were frequently obtained, especially in the examination of a suspected sellar pro- cess. For better characterization of lesions, selected sin- gle slice images were produced using inversion recovery sequences (TR 1000 ms, TI 250, 200 and 150-100). The whole examination time varied from 60 to 80 minutes. A complete series of SE images was obtained in every case. Some patients were not able to cooperate long enough. Hence, in 36 cases only one or few IR images were available. Tumor detectability, tumor delineation and sig- nal intensity relative to surrounding tissues were ana- lyzed. All patients underwent contemporaneous non-contrast and contrast enhanced CT examinations. Nine patients also had an angiography and 18 patients a scintigraphy. Forty-six patients were operated upon and 2 patients au- topsied within one month after the MR examination. The final diagnoses are shown in the Table. In 56 patients the diagnosis was histologically confirmed. Three patients without a histologic confirmation had a pituitary tumor on CT and abnormal hormone values. Four patients with an unconfirmed diagnosis of brain metastases showed multi- From the Department of Diagnostic Radiology, University Central Hospital, SF-20520 Turku, Finland. Accepted for publi- cation 17 February 1987. - 24-878084 369 Acta Radiol Downloaded from informahealthcare.com by ThULB Jena on 11/21/14 For personal use only.

Upload: k

Post on 27-Mar-2017

215 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Magnetic Resonance Imaging of Intracranial Neoplasms at 0.02 Tesla

A C T A A D I 0 L 0 G I C A Vol. 28 1987 Fasc. 4 July-August

MAGNETIC RESONANCE IMAGING OF INTRACRANIAL NEOPLASMS AT 0.02 TESLA

M. KORMANO, R. RAININKO and K. KATEVUO

Abstract Sixty-five patients with single or multiple intracranial neo-

plasms were examined with an MR imager operating at 0.02 tesla. In 56 patients the diagnosis was histologically confirmed. All patients had an abnormal CT finding. MR images were positive in 59 cases, while the lesion remained undetected or equivocal in 6 cases (2 pituitary adenomas and 4 meningiomas). The MR signal intensity of several meningiomas was equal to that of normal brain tissue. Some astrocytomas were better delineated on MRI than on CT. For the study of pituitary lesions, the spatial resolu- tion was unsatisfactory. The ultralow field MR imager was found to be sensitive for the detection of other intracranial neoplasms.

Key words: Brain, MR studies; -, neoplasms; magnetic reso- nance, comparative studies; -, technique; computed tomogra- phy, comparative studies.

Since the introduction of magnetic resonance imaging (MRI) into clinical practice, several reports have been published on its value in the diagnosis of intracranial tumors (4-6, 8, 9, 11, 13, 15, 18, 21). MRI has been considered a useful method, especially in patients with a suspected brain stem tumor.

The optimum field strength in clinical MRI is still a controversial question. In most reports field strengths from 0.15 to 0.35 tesla (T) have been used. In a concise material (2,3), results obtained with field strengths of 0.12 or 1.5 T were compared; higher field strengths resulted in better anatomic delineation. To our knowledge 0.02 T is the lowest magnetic field strength used in clinical MRI. Our experience with MRI at 0.02 T in the diagnosis of intracranial neoplasms is now described and the results are compared with the findings at computed tomography (CT).

Material and Methods

The material consisted of 65 patients (33 men and 32 women, aged 18 to 77 years, median 56) with single or

multiple intracranial neoplasms, referred for CT or MRI between May 1984 and September 1985. The patients were not premedicated or anesthetized.

The MR imager (Acutscan, Instrumentarium, Finland) has a resistive magnet and operates at a field strength of 0.02 T. The slice thickness is 10 mm, and the imaging matrix is 256X 128. The field of view of the head coil is 30 cm allowing a spatial resolution of 3 to 5 mm. Multislice axial spin echo technique with TR 2000 ms and TE from 130 to 160 ms and with two acquistions was routinely used. Sagittal or coronal slices were frequently obtained, especially in the examination of a suspected sellar pro- cess. For better characterization of lesions, selected sin- gle slice images were produced using inversion recovery sequences (TR 1000 ms, TI 250, 200 and 150-100). The whole examination time varied from 60 to 80 minutes. A complete series of SE images was obtained in every case. Some patients were not able to cooperate long enough. Hence, in 36 cases only one or few IR images were available. Tumor detectability, tumor delineation and sig- nal intensity relative to surrounding tissues were ana- lyzed.

All patients underwent contemporaneous non-contrast and contrast enhanced CT examinations. Nine patients also had an angiography and 18 patients a scintigraphy. Forty-six patients were operated upon and 2 patients au- topsied within one month after the MR examination. The final diagnoses are shown in the Table. In 56 patients the diagnosis was histologically confirmed. Three patients without a histologic confirmation had a pituitary tumor on CT and abnormal hormone values. Four patients with an unconfirmed diagnosis of brain metastases showed multi-

From the Department of Diagnostic Radiology, University Central Hospital, SF-20520 Turku, Finland. Accepted for publi- cation 17 February 1987.

-

24-878084 369

Act

a R

adio

l Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

ThU

LB

Jen

a on

11/

21/1

4Fo

r pe

rson

al u

se o

nly.

Page 2: Magnetic Resonance Imaging of Intracranial Neoplasms at 0.02 Tesla

370 M . KORMANO, R. RAININKO AND K. KATEWO

Lymphoma Melanoma Hemanrrioblastoma Neuroeitodermd tumor 1 Pineal germinoma 1 Tumor of unknown origin (glioma?) 2 Total 65

pie intracranial lesions both on MRI and on CT; in 2 of them a tumor was also demonstrated at chest radiogra- phy. Two patients without histologic diagnosis revealed a single tumor both on MRI and on CT; these tumors were suspected to be gliomas.

a b

In 55 patients the lesions were supratentorial, including C d 8 neoplasms in the sellar region, and 6 were infratentorial. I,, 4 patients both supra- and infratentorial lesions were detected.

Fig. 1. Astrocytoma grade 11 in MR image SE 2000/130 (a, b). neoplasm appears to be a little larger on MRI than on contrast enhanced CT scans (c, d) in which it is poorly delineated.

Results

Detectability of lesions. The MR findings were positive in 59 patients. In 6 patients MR examinations were nega- tive or equivocal. Two of these negative cases were pitu- itary adenomas, which extended also into the suprasellar cistern on CT, and 4 were meningiomas. None of the 4 meningiomas undetected on MRI showed a surrounding edema on CT. The largest of the missed meningiomas was 3 cm in diameter at operation, and the smallest was a thin parasellar plaque. As re-evaluated in retrospect 2 missed cases of meningioma exhibited slight asymmetry of anato- my related to the tumor and could be suspected as being abnormal. In a further 2 cases a signal intensity of the lesion similar to that of the brain did not allow a positive diagnosis even in retrospect. CT showed abnormality in all 65 patients.

The extent of the neoplasms, including possible sur- rounding edema, was equally large on CT and MRI in 40 patients. The abnormality appeared larger on MRI than on CT in 13 patients and larger on CT than on MRI in 6 patients (excluding the 6 missed cases). Three astrocyto- mas appeared somewhat larger on MRI than on CT. These neoplasms were poorly delineated on CT, but were well differentiated from the normal brain parenchyma on MR images (Fig. 1). In 2 patients with multiple carcinomatous metastases on CT, one additional infratentorial metastasis was detected on MRI. In one patient with intracranial

a b

Fig. 2. Glioblastoma with surrounding edema. Cystic component is well seen both on MRI (SE 2000/160) (a) and on contrast enhanced CT (b).

metastases from a small cell carcinoma of the lung, CT demonstrated 10 metastases but MRI only 3, one of which was not detected on CT. In one patient with a carcinoma- tous metastasis in the pineal region, the tumor appeared larger on CT than on MR images.

Tumor delineation. Edema gave a strong signal on SE images. Differentiation 'of the neoplasm from the sur- rounding edema was often difficult. Edema was seen around the neoplasm on CT-scans of 40 patients. Among

Act

a R

adio

l Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

ThU

LB

Jen

a on

11/

21/1

4Fo

r pe

rson

al u

se o

nly.

Page 3: Magnetic Resonance Imaging of Intracranial Neoplasms at 0.02 Tesla

MAGNETIC RESONANCE IMAGING OF INTRACRANIAL NEOPLASMS AT 0.02 TESLA 37 1

a b

Fig. 3. Glioblastoma is much better delineated on MRI (SE 2000/145) (a) as compared with contrast enhanced CT (b).

Fig. 5. A parasagittal meningioma with large surrounding edema. Calcification is seen as a dark spot (arrow). SE 2000/160.

a

b

Fig. 4. Basophil pituitary adenoma. The neoplasm is not differen- tiated from CSF in a SE 2000/130 image (a) but is well differenti- ated in an IR 1600/250 image (b).

these patients, tumor margins were well seen in 22 pa- tients (Fig. 2) and partially so in 10 patients on the SE 2000445 MR images. In 8 cases the neoplasm could not be differentiated from edema on SE sequences. When a se- ries of IR images was available tumor margins were satis- factorily delineated.

Neoplasms without marked surrounding edema (18 pa- tients) were successfully delineated (Figs l , 3) except in the false negative cases: 4 meningiomas and some of the multiple metastases in a patient with pulmonary carcino- ma. Some astrocytomas and one glioblastoma, that were poorly delineated on CT, were well seen on MRI. The surgical findings agreed with the MRI findings, whenever the real extent of the neoplasm could be determined at operation.

All 6 pituitary adenomas and one spongioblastoma were covered by cerebrospinal fluid in the suprasellar cistern. In the SE 2000/150 sequence the signal intensity from the tumor and the cerebrospinal fluid could not be distin- guished from each other. The margins of the pituitary lesions were best demonstrated on IR 1000/250 images (Fig. 4).

Signal intensity In heavily T2 weighted SE 20001150 images the majority

of neoplasms other than meningiomas emitted a strong signal. Nine of the 15 meningiomas showed approximately the same signal intensity as did the brain. Even with supplementary IR images the signal from 8 meningiomas did not clearly differ from that of the brain.

Signal intensity from solid tumor tissue either on SE images or on the series of IR images with different inver- sion times showed no correlation with the histologic struc- ture or the grade of malignancy. Cystic parts of tumors emitted a strong signal on SE images. Usually the cystic and solid components of the tumor could be differentiated from each other (Fig. 2), unless the solid wall was very thin. The T2 weighted signals from edema, cysts, or ne- crotic tissue were often identical. Calcifications were seen as black areas on MRI (Fig. 5 ) , but they were more often detected on CT. In general, the analysis of the structure of the lesion (differentiation of edema, solid tumor, cystic or necrotic components, calcifications) was easier and more accurate on CT than on MRI. Thus, CT images were

Act

a R

adio

l Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

ThU

LB

Jen

a on

11/

21/1

4Fo

r pe

rson

al u

se o

nly.

Page 4: Magnetic Resonance Imaging of Intracranial Neoplasms at 0.02 Tesla

372 M. KORMANO, R . RAININKO AND K. KATEVUO

considered to have better differential diagnostic specific- ity than the ultralow field MR images.

Discussion

Enthusiasm for MRI in the detection of intracranial neoplasms is well justified, because of its sensitivity to changes in the state of water in tissues and better demon- stration of the posterior fossa and better spatial orientation as compared with CT (8, 15, 20). The lack of specificity of MRI in differentiating malignant versus benign neoplasia is currently recognized (6,8, 10,21). The latter as well as the substantially higher cost of MRI as compared with CT have been the main reasons why CT is currently consid- ered the primary imaging modality in cases of suspected intracranial neoplasms while MRI has been considered a complementary examination. An ultralow field MR imager is an interesting approach, since it brings the cost of MR imaging to the level of a CT examination. Also, it may even have better TI weighted contrast capability than higher field strength MR imagers (22). It is therefore important to investigate whether the inherently lower sig- nal to noise ratio, as compared with higher field strength MR imagers, allows enough spatial resolution to fully utilize the sensitivity of MR to changes in the shift of water in tissue. It has been proven to be diagnostic in intracranial hemorrhage (23) and in demyelinating dis- eases. In intracranial neoplasms structural information and consequently spatial resolution would appear more important for diagnosis. Low signal intensity favors the use of 10 mm or 15 mm slice thickness instead of 5 mm often used in high magnetic field imagers. These factors make ultralow field MR images less likely diagnostic of small tumors if they have a signal intensity close to that of neighboring tissue. Therefore, some false negative diag- noses in small tumors, e.g. pituitary adenomas, and me- ningiomas are more likely to occur at ultralow field MRI.

Meningiomas have been problematic also when other types of MR devices have been used. In the material of BRADLEY et coll. (9, 3 of the 22 meningiomas were not detected on MRI and the largest of the missed meningio- mas was 1.5 cm in diameter. Relaxation times of meningi- omas closely resemble those of gray matter. Therefore, small meningiomas without surrounding edema are diffi- cult to detect unless anatomy is demonstrated in great detail.

The superiority of high field strength in the diagnosis of pituitary lesions has been documented previously in pa- tients examined at different field strength (3). Large tu- mors in the sellar region are usually detected also at low field MRI.

Ultralow magnetic field strength is not only a limiting factor, it does also allow some advantages relative to higher field strengths due to greater contrast between T1 signal intensity of different tissues (12). New methods such as T1e dispersion imaging (22), which may further

improve tissue characterization, are available at a low magnetic field. Unfortunately the pulse sequences cur- rently used in routine examinations are heavily T2 weight- ed and may not utilize the T1 relaxation differences be- tween tissues.

The heavily T2 weighted pulse sequence used in the present study seems to be sensitive for detection of most brain lesions as reported also at higher magnetic fields (7, 24). Characterization of a lesion or its components may pose a problem at any field strength, and one of the problems is the poor differentiation of the neoplasm from the edema in some cases (4, 6-8, 18, 21).

The most important indications for the use of MRI in suspected brain tumors are the known occurrence of CT isodense tumors, such as oligodendroglioma and astrocy- toma grade I and 11, and of tumors in the cerebellum and brainstem (19). Our results are in agreement with investi- gators using higher field strengths. MRI seems to be supe- rior in defining the extent of gliomas poorly delineated at CT.

The MRI appearance of neoplasms did not correlate with the histologic diagnosis or with the malignancy of the neoplasm although several pulse sequences were used. Poor specificity of MRI has been generally recognized (4, 8, 1 l), because there is an overlap in the relaxation times of tumors of different histologic types (1, 4, 8, 11, 17). More information on the nature of the lesion was attained on CT, although correlation of the CT appearance and histology of gliomas has also been proven quite poor (14). Tissue characterization at MRI may improve with increas- ing experience.

The introduction of a low field and low cost MRI may place MRI of the brain in a totally new context by being economically competitive and sensitive. An ultralow field MRI shows potential as a clinically useful method which is sensitive enough to detect most intracranial abnormali- ties. After detection of a suspected neoplasm CT exami- nation is still important for better structural analysis.

Request for reprints: Dr Raili Raininko, Department of Diag- nostic Radiology, Meilahti Hospital, Helsinki University Central Hospital, SF-00290 Helsinki, Finland.

REFERENCES 1. ARAKI T., INOUYE T., SUZUKI H., MACHIDA T. and 110 M.:

Magnetic resonance imaging of brain tumors. Measurement of T1. Radiology 150 (1984), 95.

2. BILANWK L. T., ZIMMERMAN R. A., WEHRLI F. W et coll.: Cerebral magnetic resonance. Comparison of high and low field strength imaging. Radiology 153 (1984), 409.

3. - - - et coll.: Magnetic resonance imaging of pituitary lesions using 1.0 to 1.5 T field strength. Radiology 153 (1984), 415.

4. BRADAC G. B., SCHORNER W., BENDER A. and FELIX R.: MRI (NMR) in the diagnosis of brain-stem tumors. Neuroradio- logy 27 (1985), 208.

5. BRADLEY W. G., WALUCH V., YADLEY R. A . and WYCOFF R. R.: Comparison of CT and MR in 400 patients with suspected

Act

a R

adio

l Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

ThU

LB

Jen

a on

11/

21/1

4Fo

r pe

rson

al u

se o

nly.

Page 5: Magnetic Resonance Imaging of Intracranial Neoplasms at 0.02 Tesla

MAGNETIC RESONANCE IMAGING OF INTRACRANIAL NEOPLASMS AT 0.02 TESLA 373

disease of the brain and cervical spinal cord. Radiology 152 (1984), 695.

6. BRANT-ZAWADZKI M., BADAMI I., MILLS K. M., NORMAN D. and NEWTON T. H.: Primary intracranial tumor imaging. A comparison of magnetic resonance and CT. Radiology 150 (1984), 435.

7. - NORMAN D., NEWTON T. H. et coll.: Magnetic resonance of the brain. The optimal screening technique. Radiology 152 (1984), 71.

8. BYDDER G. M., STEINER R. E., YOUNG I. R. et coll.: Clinical NMR imaging of the brain: 140 cases. Amer. J. Roentgenol. 139 (1982), 215.

9. HAN J . S., BONSTELLE C. T., KAUFMAN B. et coll.: Magnetic resonance imaging in the evaluation of the brainstem. Radio- logy 150 (1984), 705.

10. HAUGHTON V. M., RIMM A. A., SOBOCINSKI K. A. et coll.: A blinded clinical comparison of MR imaging and CT in neuro- radiology. Radiology 160 (1986), 751.

1 1 . H U E ~ L E M. G., HAN J. S., KAUFMAN B. and BENSON J . E.: MR imaging of brain stem gliomas. J. Comput. Assist. To- mogr. 9 (1985), 263.

12. KOENIG S. H., BROWN 111 R. D., ADAMS D., EMERSON D. and HARRISON C. G.: Magnetic field dependence of llTl of pro- tons in tissue. Invest. Radiol. 19 (1984), 76.

13. KUCHARCZYK W., BRANT-ZAWADZKI M., SOBEL D. et coll.: Central nervous system tumors in children. Detection by magnetic resonance imaging. Radiology 155 (1985), 131.

14. LILJA A., BERGSTROM K., SPANNARE B. and OLSON Y.: Reli- ability of computed tomography in assessing histopathologi- cal features of malignant supratentorial gliomas. J. Comput. Assist. Tomogr. 5 (1981), 625.

15. MCCINNES B. D., BRADY T. J., NEW P. F. J. et coll.: Nuclear magnetic resonance (NMR) imaging of tumors in the posteri- or fossa. J. Comput. Assist. Tomogr. 7 (1983), 575.

16. MAWHINNEY R. R., BUCKLEY J. H., HOLLAND I. M. and WORTHINGTON B. S.: The value of magnetic resonance imag- ing in the diagnosis of intracranial meningiomas. Clin. Ra- dial. 376 (1986), 429.

17. MILLS C. M., CROOKS L. E., KAUFMAN L. and BRANT- ZAWADZKI M.: Cerebral abnormalities. Use of calculated T1 and T2 magnetic resonance images for diagnosis. Radiology 150 (1984), 87.

18. PETERMAN S. B., STEINER R. E., BYDDER G. M., THOMAS D. J., TOBIAS J. S. and YOUNG I. R.: Nuclear magnetic reso- nance imaging (NMR) (MRI), of brain stem tumors. Neuro- radiology 27 (1985), 202.

19. PETERSEN S. B., MULLER R. R. and RINCK P. A.: An introduc- tion to biomedical nuclear magnetic resonance. Georg Thieme Verlag, Stuttgart, New York 1985.

20. POMERANZ S. J., SOILA, K., TOBIAS J., SHELDON J. J., Ros P. and VIAMONTE M. JR: Sensitivity of MRI in metastatic neo- plasia. A case report. Magnet. Res. Imag. 3 (1985), 291.

21. RANDELL C. P., COLLINS A. G., YOUNG I. R. et coll.: Nuclear magnetic resonance imaging of posterior fossa tumors. Amer. J . Roentgenol. 141 (1983), 489.

22. SEPPONEN R. E., POHJONEN J. A., SIPPONEN J . T. and TANTTU J . I.: A method for T1 imaging. J . Comput. Assist. Tomogr. 9 (1985), 1007.

23. SIPFQNEN J. T., SEPPONEN R. E., TAN^ J. I. and SIVULA A.: Intracranial hematomas studied by MR imaging at 0.17 and 0.02 T. J. Comput. Assist. Tomogr. 9 (1985), 698.

24. SMITH A. S., WEINSTEIN M. A., MODIC M. T. et coll.: Magnet- ic resonance with marked T2-weighted images. Improved demonstration of brain lesions, tumor and edema. Amer. J. Roentgenol. 145 (1984), 949.

Act

a R

adio

l Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

ThU

LB

Jen

a on

11/

21/1

4Fo

r pe

rson

al u

se o

nly.