Editorials James
W. Reinig,
MD
Differentiation with MR Imaging:
O
great hopes for magnetic (MR) imaging was to be able to specifically characterize lesions, either on the basis of visual appearance or by the use of Ti and T2 values. In particular, it was hoped that benign and malignant lesions in the liver could be differentiated to avoid biopsy in patients discovered to have hepatic masses. For the past 7 years, the search for a reliable method of differentiation has led investigators to measure Ti and T2 values (1-li), lesion-to-liver signal intensity ratios (6-8,ii-i4), and lesionto-liver contrast-to-noise ratios (4,8,12, 13) and to describe morphologic characteristics (4,6,7,i3-i5) in an attempt to distinguish benign from malignant lesiorLs. Unfortunately, we are not much better at making this distinction today than at the dawn of MR imaging. Early in the MR imaging experience, a great effort was made to measure Ti and T2 values of every lesion that was encountered. A number of investigators compiled these figures and attempted to characterize lesions based on Ti and T2 values. There was early recognition that the Ti and T2 values of cavernou#{225} hemangiomas as a group were significantly greater than those of metastases or hepatocellular carcinoma (2-4). There was significant overlap between the values of individual benign lesions and those of malignant ones, however (5,7,9,iO). In response to the difficulty in performing Ti and T2 calculations (14), the realization that the measurements were not highly accurate (6,13), and a desire to come up with a better method of differentiating benign from malignant disease, other NE
of the
resonance
Index terms Angioma. gastrointestinal, 761.3194 #{149} Editorials #{149} Liver, cysts. 761.312. Liver neoplasms. diagnosis, 761.30. Liver plasms, MR studies, 761.30,761.1214 Radiology
1
1991;
From
Anne
179:601-602
Arundel
Annapolis, MD accepted March the author.
neo-
MRI.
235 Jennifer
Rd.
21401. Received March 7,1991; 8. Address reprint requests to
#{176}RSNA,1991 See
also
the
article
679) in this issue.
by Brown
et al (pp
675-
of Hepatic Lesions The Last Word?’ authors
measured
liver-to-lesion
con-
trast-to-noise ratios or differences in signal intensity (4,6-8,ii-i4). Again, classes of lesions could be described in general terms (hemangiomas had higher signal intensity than malignant lesions); however, it was often not possible to identify a particular lesion. Furthermore, all of these methods were tedious efforts requiring the radiologist to be present when the study was performed or to return to the console shortly thereafter to make the measurements. On some units, it was difficult or not possible with some software releases to perform Ti and T2 calculalions (14). Outside university settings, it was a difficult practice to make these calculations. More recent articles have described the use of morphologic characteristics to distinguish benign from malignant disease (6,i3-i5). The smoothness of the margins, homogeneity, and appearance of the internal architecture were reported to be of use in distinguishing among these classes of disease. Although smooth margins could be seen in significant numbers of both benign and malignant lesions, irregular margins were found much more often in malignant disease (6). Homogeneous lesions were usually benign, while inhomogeneous lesions had a high probability
of being
malignant
(6,i3).
In ad-
dition, lesions with a target, halo, and amorphous appearance were described as being specific for malignant disease (15). Unfortunately for the general reader, after similar studies based on the morphologic characteristics of liver lesions were performed, diametrically opposing opinions have been voiced in the literature. While one author states, “Qualitative data alone are sufficient quantitative data are not required or useful in distinguishing hemangioma from metastasis” (6), another author indicates, “We conclude that distinction between cavernous hemangioma and malignancy by visual analysis of lesion morphology is inaccurate” (14). There is little wonder that many studies ended up being read on the basis of the gut feeling of the radiologist. In the article “Focal Hepatic Lesions: Differentiation with MR Imaging at 0.5 T,” Brown et a! (i6) repeated the effort to describe specific appearances that
correspond to benign or malignant lesions. However, they examined the differentiation by means of visual analysis of the lesion homogeneity, shape, and signal intensity compared with that of normal liver parenchyma, spleen, and skeletal muscle on Ti-weighted, balanced, and T2-weighted images. Although this analysis could result in a large number of potential combinations of characteristics, the authors found that a few distinct scenarios were highly specific for malignancy. Irregular lesions were uniformly malignant in their series. Further, heterogeneous lesions were almost always malignant unless they were markedly hyperintense. Lesions that were heterogeneous and isointense or hyperintense compared with liver on a T2weighted image and isointense to the spleen on a balanced image had a greater than 99.9% chance of being malignant. Homogeneous lesions that were isointense or hyperintense to the spleen on balanced images and not markedly hyperintense compared with liver on a T2-weighted image had a greater
than
95%
chance
of being
ma-
lignant.
However, lesions that were homogeneous and markedly hyperintense to liver and spleen had a less than 5% chance of being malignant. Of the lesions examined in this series, 73% had a predicted likelihood of malignancy of greater than 95% or less than 5%. It is certainly desirable to derive a characterization scheme based only on visual analysis. The greater the operator-dependent interaction with the console
to make
measurements
or cal-
culations, the less the likelihood that the method will be used. Only when calculations turn out to be absolutely specific for benign or metastatic disease would it routinely be worth the effort. Further, the more extensive the description of malignant lesions, the less likely that it will be remembered. The analysis of the lesions by Brown et al (i6) is useful because it was derived from sequences routinely used for abdominal imaging and centers on a few specific descriptions that have a great chance of being benign or malignant. Therefore, the descriptions will find a place in day-to-day liver imaging. This is clearly not the last word on 601
the subject. A significant sions remain unclassified scheme. Further, authors previously
reported
number of leunder this (6,13) have
conflicting
data
(eg, Li et al [6] report 20.8% of their hemangiomas to have irregular margins). There is plenty of room for the ingenious investigator to come up with an even more useful method of distinguishing benign from malignant disease in the liver. Such a scenario might include the use of intravenous contrast material (17-20) to further evaluate the indeterminate lesions. It appears that one of the original challenges of hepatic MR imaging might well turn out to be one of the most lasting. a References 1.
2.
3.
602.
4.
5.
6.
7.
8.
9.
Moss AA, Goldberg HI, Stark DB, et al. Hepatic tumors: magnetic resonance and CT appearance. Radiology 1984; 150:141147. Ohtomo K, Itai Y, Furui S, Yashiro N, Yoshikawa K, ho M. Hepatic tumors: differentiation by transverse relaxation time (T2) of magnetic resonance imaging. Radiology 1985; 155:421-423. Itai Y, Ohtomo K, Furui 5, Yamauchi T, Minami M, Yashiro N. Noninvasive diagnosis of small cavernous hemangioma of the liver: advantage of MRI. AJR 1985; 145:1195-1199.
Radiology
10.
11.
12.
Stark DD, Felder RC, Wittenberg J. et al. Magnetic resonance imaging of cavernous hemangioma of the liver tissue-specific characterization. AJR 1985; 145:213-222. Ohtomo K, Itai Y, Yoshikawa K, Kokubo T, ho M. Hepatocellular carcinoma and cayernous hemangioma: differentiation with MR imaging. Radiology 1988; 168:621-. 623. Li KC, Glazer GM, Quint LE, et al. Distinction of hepatic cavernous hemangioma from hepatic metastases with MR imaging. Radiology 1988; 169:409-415. Lombardo DM, Baker ME, Spritzer CE, Binder R, Myers W, Herfkens RJ. Hepatic hemangiomas vs metastases: MR differentiation at 1ST. AJR 1990; 155:55-59. Egglin TK, Rummeny E, Stark DD, Wittenberg J, Saini S. Ferruci JT. Hepatic tumors: quantitative tissue characterization with MR imaging. Radiology 1990; 176:107-1 10. Patrizio G, Pavone P. Testa A, Marsili L, Tettamanti E, Passariello R. MR characterization of hepatic lesions by t-null inversion recovery sequence. J Comput Assist Tomogr 1990; 14:96-101. Ohtomo K, Itai Y, Matuoka !t, et al. Hepatocellular carcinoma: MR appearance mimicking cavernous hemangioma. Comput Assist Tornogr 1990; 14:650-652. Glazer GM, Aisen AM, Francis hR. Gyves, JW, Lande I. Adler DD. Hepatic cavernous hemangioma: magnetic resonance imaging. Radiology 1985; 155:417-420. Itoh K, Saini S. Hahn PF, Imam N, Ferrucci JT. Differentiation between small hepatic hemangiomas and metastases on MR images: importance of size-specific quantitative criteria. AJR 1990; 155:61-66.
13.
14.
15.
16.
17.
18.
19.
20.
Choi BI, Han MC, Kim CW. Small hepatocellular carcinoma versus small cavernous hemangioma: differentiation with MR imaging at 2.0 T. Radiology 1990; 176:103106. Mirowitz SA, Lee JKT, Heiken JP. Cayernous hemangioma of the liver: assessment of MR tissue specificity with a simphified T2 index. J Comput Assist Tomogr 1990; 14:223-2.28. Wittenberg J, Stark DD, Forman BH, et a1. Differentiation of hepatic metastases from hepatic hemangiomas and cysts by using MR imaging. AJR 1988; 151:79-84. Brown JJ, Lee JM, Lee JKT, Lom KV, Malchow S. Focal hepatic lesions: differentiation with MR imaging at 0.5 T. Radiology 1991; 179:000-000. Ohtomo K, Itai Y, Yoshikawa K, et al. Hepatic tumors: dynamic MR imaging. Radiology 1987; 163:27-31. Yoshida H, ltai Y, Ohtomo K, Kokubo T, Minami M, Yashiro N. Small hepatocellular carcinoma and cavernous hemangioma: differentiation with dynamic FLASH MR imaging with Gd-DTPA. Radiology 1989; 171:339-342. Edelman RR, Siegel JB, Singer A, Dupuis K, Longmaid HE. Dynamic MR imaging of the liver with Gd-DTPA: initial clinical results. AJR 1989; 153:1213-1219. Hamm B, Fischer E, Taupitz M. Differentiation of hepatic hemangiomas from metastases by dynamic contrast-enhanced MR imaging. J Comput Assist Tomogr 1990; 14:205-216.
June
1991
W. Reinig,
MD
Differentiation with MR Imaging:
O
great hopes for magnetic (MR) imaging was to be able to specifically characterize lesions, either on the basis of visual appearance or by the use of Ti and T2 values. In particular, it was hoped that benign and malignant lesions in the liver could be differentiated to avoid biopsy in patients discovered to have hepatic masses. For the past 7 years, the search for a reliable method of differentiation has led investigators to measure Ti and T2 values (1-li), lesion-to-liver signal intensity ratios (6-8,ii-i4), and lesionto-liver contrast-to-noise ratios (4,8,12, 13) and to describe morphologic characteristics (4,6,7,i3-i5) in an attempt to distinguish benign from malignant lesiorLs. Unfortunately, we are not much better at making this distinction today than at the dawn of MR imaging. Early in the MR imaging experience, a great effort was made to measure Ti and T2 values of every lesion that was encountered. A number of investigators compiled these figures and attempted to characterize lesions based on Ti and T2 values. There was early recognition that the Ti and T2 values of cavernou#{225} hemangiomas as a group were significantly greater than those of metastases or hepatocellular carcinoma (2-4). There was significant overlap between the values of individual benign lesions and those of malignant ones, however (5,7,9,iO). In response to the difficulty in performing Ti and T2 calculations (14), the realization that the measurements were not highly accurate (6,13), and a desire to come up with a better method of differentiating benign from malignant disease, other NE
of the
resonance
Index terms Angioma. gastrointestinal, 761.3194 #{149} Editorials #{149} Liver, cysts. 761.312. Liver neoplasms. diagnosis, 761.30. Liver plasms, MR studies, 761.30,761.1214 Radiology
1
1991;
From
Anne
179:601-602
Arundel
Annapolis, MD accepted March the author.
neo-
MRI.
235 Jennifer
Rd.
21401. Received March 7,1991; 8. Address reprint requests to
#{176}RSNA,1991 See
also
the
article
679) in this issue.
by Brown
et al (pp
675-
of Hepatic Lesions The Last Word?’ authors
measured
liver-to-lesion
con-
trast-to-noise ratios or differences in signal intensity (4,6-8,ii-i4). Again, classes of lesions could be described in general terms (hemangiomas had higher signal intensity than malignant lesions); however, it was often not possible to identify a particular lesion. Furthermore, all of these methods were tedious efforts requiring the radiologist to be present when the study was performed or to return to the console shortly thereafter to make the measurements. On some units, it was difficult or not possible with some software releases to perform Ti and T2 calculalions (14). Outside university settings, it was a difficult practice to make these calculations. More recent articles have described the use of morphologic characteristics to distinguish benign from malignant disease (6,i3-i5). The smoothness of the margins, homogeneity, and appearance of the internal architecture were reported to be of use in distinguishing among these classes of disease. Although smooth margins could be seen in significant numbers of both benign and malignant lesions, irregular margins were found much more often in malignant disease (6). Homogeneous lesions were usually benign, while inhomogeneous lesions had a high probability
of being
malignant
(6,i3).
In ad-
dition, lesions with a target, halo, and amorphous appearance were described as being specific for malignant disease (15). Unfortunately for the general reader, after similar studies based on the morphologic characteristics of liver lesions were performed, diametrically opposing opinions have been voiced in the literature. While one author states, “Qualitative data alone are sufficient quantitative data are not required or useful in distinguishing hemangioma from metastasis” (6), another author indicates, “We conclude that distinction between cavernous hemangioma and malignancy by visual analysis of lesion morphology is inaccurate” (14). There is little wonder that many studies ended up being read on the basis of the gut feeling of the radiologist. In the article “Focal Hepatic Lesions: Differentiation with MR Imaging at 0.5 T,” Brown et a! (i6) repeated the effort to describe specific appearances that
correspond to benign or malignant lesions. However, they examined the differentiation by means of visual analysis of the lesion homogeneity, shape, and signal intensity compared with that of normal liver parenchyma, spleen, and skeletal muscle on Ti-weighted, balanced, and T2-weighted images. Although this analysis could result in a large number of potential combinations of characteristics, the authors found that a few distinct scenarios were highly specific for malignancy. Irregular lesions were uniformly malignant in their series. Further, heterogeneous lesions were almost always malignant unless they were markedly hyperintense. Lesions that were heterogeneous and isointense or hyperintense compared with liver on a T2weighted image and isointense to the spleen on a balanced image had a greater than 99.9% chance of being malignant. Homogeneous lesions that were isointense or hyperintense to the spleen on balanced images and not markedly hyperintense compared with liver on a T2-weighted image had a greater
than
95%
chance
of being
ma-
lignant.
However, lesions that were homogeneous and markedly hyperintense to liver and spleen had a less than 5% chance of being malignant. Of the lesions examined in this series, 73% had a predicted likelihood of malignancy of greater than 95% or less than 5%. It is certainly desirable to derive a characterization scheme based only on visual analysis. The greater the operator-dependent interaction with the console
to make
measurements
or cal-
culations, the less the likelihood that the method will be used. Only when calculations turn out to be absolutely specific for benign or metastatic disease would it routinely be worth the effort. Further, the more extensive the description of malignant lesions, the less likely that it will be remembered. The analysis of the lesions by Brown et al (i6) is useful because it was derived from sequences routinely used for abdominal imaging and centers on a few specific descriptions that have a great chance of being benign or malignant. Therefore, the descriptions will find a place in day-to-day liver imaging. This is clearly not the last word on 601
the subject. A significant sions remain unclassified scheme. Further, authors previously
reported
number of leunder this (6,13) have
conflicting
data
(eg, Li et al [6] report 20.8% of their hemangiomas to have irregular margins). There is plenty of room for the ingenious investigator to come up with an even more useful method of distinguishing benign from malignant disease in the liver. Such a scenario might include the use of intravenous contrast material (17-20) to further evaluate the indeterminate lesions. It appears that one of the original challenges of hepatic MR imaging might well turn out to be one of the most lasting. a References 1.
2.
3.
602.
4.
5.
6.
7.
8.
9.
Moss AA, Goldberg HI, Stark DB, et al. Hepatic tumors: magnetic resonance and CT appearance. Radiology 1984; 150:141147. Ohtomo K, Itai Y, Furui S, Yashiro N, Yoshikawa K, ho M. Hepatic tumors: differentiation by transverse relaxation time (T2) of magnetic resonance imaging. Radiology 1985; 155:421-423. Itai Y, Ohtomo K, Furui 5, Yamauchi T, Minami M, Yashiro N. Noninvasive diagnosis of small cavernous hemangioma of the liver: advantage of MRI. AJR 1985; 145:1195-1199.
Radiology
10.
11.
12.
Stark DD, Felder RC, Wittenberg J. et al. Magnetic resonance imaging of cavernous hemangioma of the liver tissue-specific characterization. AJR 1985; 145:213-222. Ohtomo K, Itai Y, Yoshikawa K, Kokubo T, ho M. Hepatocellular carcinoma and cayernous hemangioma: differentiation with MR imaging. Radiology 1988; 168:621-. 623. Li KC, Glazer GM, Quint LE, et al. Distinction of hepatic cavernous hemangioma from hepatic metastases with MR imaging. Radiology 1988; 169:409-415. Lombardo DM, Baker ME, Spritzer CE, Binder R, Myers W, Herfkens RJ. Hepatic hemangiomas vs metastases: MR differentiation at 1ST. AJR 1990; 155:55-59. Egglin TK, Rummeny E, Stark DD, Wittenberg J, Saini S. Ferruci JT. Hepatic tumors: quantitative tissue characterization with MR imaging. Radiology 1990; 176:107-1 10. Patrizio G, Pavone P. Testa A, Marsili L, Tettamanti E, Passariello R. MR characterization of hepatic lesions by t-null inversion recovery sequence. J Comput Assist Tomogr 1990; 14:96-101. Ohtomo K, Itai Y, Matuoka !t, et al. Hepatocellular carcinoma: MR appearance mimicking cavernous hemangioma. Comput Assist Tornogr 1990; 14:650-652. Glazer GM, Aisen AM, Francis hR. Gyves, JW, Lande I. Adler DD. Hepatic cavernous hemangioma: magnetic resonance imaging. Radiology 1985; 155:417-420. Itoh K, Saini S. Hahn PF, Imam N, Ferrucci JT. Differentiation between small hepatic hemangiomas and metastases on MR images: importance of size-specific quantitative criteria. AJR 1990; 155:61-66.
13.
14.
15.
16.
17.
18.
19.
20.
Choi BI, Han MC, Kim CW. Small hepatocellular carcinoma versus small cavernous hemangioma: differentiation with MR imaging at 2.0 T. Radiology 1990; 176:103106. Mirowitz SA, Lee JKT, Heiken JP. Cayernous hemangioma of the liver: assessment of MR tissue specificity with a simphified T2 index. J Comput Assist Tomogr 1990; 14:223-2.28. Wittenberg J, Stark DD, Forman BH, et a1. Differentiation of hepatic metastases from hepatic hemangiomas and cysts by using MR imaging. AJR 1988; 151:79-84. Brown JJ, Lee JM, Lee JKT, Lom KV, Malchow S. Focal hepatic lesions: differentiation with MR imaging at 0.5 T. Radiology 1991; 179:000-000. Ohtomo K, Itai Y, Yoshikawa K, et al. Hepatic tumors: dynamic MR imaging. Radiology 1987; 163:27-31. Yoshida H, ltai Y, Ohtomo K, Kokubo T, Minami M, Yashiro N. Small hepatocellular carcinoma and cavernous hemangioma: differentiation with dynamic FLASH MR imaging with Gd-DTPA. Radiology 1989; 171:339-342. Edelman RR, Siegel JB, Singer A, Dupuis K, Longmaid HE. Dynamic MR imaging of the liver with Gd-DTPA: initial clinical results. AJR 1989; 153:1213-1219. Hamm B, Fischer E, Taupitz M. Differentiation of hepatic hemangiomas from metastases by dynamic contrast-enhanced MR imaging. J Comput Assist Tomogr 1990; 14:205-216.
June
1991
