- Email: [email protected]
Relationship between apparent diffusion coefficient and signal intensity in endometrial and other pelvic cysts
Magnetic Resonance Imaging 20 (2002) 463– 470
Relationship between apparent diffusion coefficient and signal intensity in endometrial and other pelvic cysts Takao Motekia,*, Hiroyuki Horikoshic, Keigo Endob a
The Department of Diagnostic Radiology, Gunma University Hospital, 3-39-15, Showa-machi, Maebashi, Gunma 371-0034, Japan b The Department of Nuclear Medicine, Gunma University Hospital, 3-39-15, Showa-machi, Maebashi, Gunma 371-0034, Japan c The Department of Radiology, Gunma Cancer Tomo Hospital, 617-1, Takabayashinishi-machi, Otashi, Gunma 373-0828, Japan Received 20 February 2002; accepted 22 May 2002
Abstract We evaluated whether apparent diffusion coefficient (ADC) value is more useful than signal intensity for differentiating endometrial cysts from other pelvic cysts. In an in vitro study, signal intensity and diffusion coefficients were measured in whole blood phantoms in which blood oxidation was gradually increased and concentration subsequently diluted. Although both signal intensity and diffusion value were largely affected by blood concentration, diffusion value was almost independent of blood oxidation and red blood cell lysis-related diminution of magnetic nonhomogeneity, both factors greatly affecting signal intensity on T1- and T2-weighted images. In an in vivo study, differentiation between endometrial and other pelvic cysts was attempted by means of ADC values and signal ratios of cysts to muscles on T1- and T2-weighted images (T1- and T2-ratios). Endometrial cysts tended to show lower T2-ratios, higher T1-ratios, and lower ADC values than other pelvic cysts ( p ⬍ 0.001). However, ADC values were not correlated with T1- and T2-ratios ( p ⬍ P0.15P). The ability of ADC value to discriminate between these two groups (discriminant rate, 91.4%) was higher than that of T2-ratio (71.4%) or T1-ratio (88.6%). If combined, ADC and T1-ratio (or T2-ratio) showed higher discriminant rate (94.3%) than the combination of T1- and T2 ratios (88.6%). ADC value might be useful for evaluating the blood concentration of a cystic lesion, because diffusion value is more closely related to blood concentration and almost independent of blood oxidation and red blood cell lysis that largely affect signal intensity. © 2002 Elsevier Science Inc. All rights reserved. Keywords: Diffusion imaging; Blood; Fast MR imaging; Cyst
1. Introduction Interest in the potential capability of magnetic resonance (MR) to image and measure molecular diffusion has been considerable [1– 4]. Diffusion measurements in vivo are useful for tissue characterization because they provide information on the mobility of water [1] or cell metabolites [5]. Diffusion measurements offer data on molecular displacements that are much smaller than the resolution of an MR image (typically a few millimeters). At present, the most important clinical application of this imaging technique is the detection and characterization of cerebral ischemia [6,7]. Recently, a diffusion-weighted ultrafast MRI technique has become available with the introduction of a new generation of MR scanners, and enabled in vivo mea* Corresponding author. Tel.: ⫹81-272-20-8621; fax: ⫹81-272-208409. E-mail address: [email protected] (T. Moteki).
surements of diffusion in abdominal organs such as liver and kidney using echo-planar imaging (EPI) [8,9]. Endometrial cysts exhibiting varying stages of hemorrhage tend to display a wide range of signal behavior. High intensity in the lesion on T1-weighted images is attributed to paramagnetic effects from methemoglobin [10], and varying intensity on T2-weighted images may be due to red blood cell lysis [11]. However, the relationship between signal intensity and diffusion coefficient in connection with blood oxidation and concentration has not been evaluated. Hence, this study was undertaken to determine whether blood oxidation-related and concentration-related changes of diffusion values were similar to ones of signal intensity on T1- and T2-weighted images. For this purpose, blood phantoms with gradually increased levels of oxidation, and at various blood concentrations, were used. In an in vivo study, differentiation between endometrial and other pelvic cysts was attempted using apparent diffusion coefficient (ADC) and signal intensity. In other words, we attempted to
0730-725X/02/$ – see front matter © 2002 Elsevier Science Inc. All rights reserved. PII: S 0 7 3 0 - 7 2 5 X ( 0 2 ) 0 0 5 2 4 - 6
464
T. Moteki et al. / Magnetic Resonance Imaging 20 (2002) 463– 470
determine whether ADC value were useful to evaluate hemorrhagic cysts.
2. Materials and methods 2.1. Phantom study Phantoms were developed to determine how diffusion value as well as signal intensity on T1- and T2-weighted images were affected by blood oxidation and blood concentration. Ten ml of blood was obtained from the medial cubital veins of a healthy human volunteer (hematocrit, 46%) and heparinized (100 U/ml). Immediately after extraction, the blood was transferred to a closed plastic syringe, and stored for 2 hours at 24°C room temperature with 5 ml of 100% N2. Then, 0.1 g/ml of sodium bicarbonate was added to raise pH to 8.6, which inhibits methemoglobin reductase [12]. Another phantom, to which 100% saline was added, was used to normalize signal intensity. After the first MR measurements, the blood phantom was ventilated with 5 ml of 100% O2. Immediately after the ventilation, second MR measurements were performed. Then, MR measurements were performed at 1.5 hr, 3 hrs, 1 day, 2 days, 3 days, 4 days, 6 days, 8 days, 10 days, 12 days and 14 days after the second MR measurement. Between MR measurements, the blood phantoms were ventilated with 100% O2. Immediately after the 14-day measurement, the blood phantom was gradually diluted (to 80, 60, 40, 20 and 10% of the original phantom contents) with saline, and MR measurements again were performed at each of these concentrations. A 1.5-T whole-body MR imager (Signa Horizon; GE, Milwaukee, USA) with head coil was used for the in vitro MR study. MR measurements for phantoms were performed at 24°C (room temperature). 2.1.1. Signal intensity measurement Signal intensity measurements were performed with the use of a T1-weighted spin-echo (SE) pulse sequence (repetition time/echo time (TR/TE ⫽ 500/14) and a T2weighted fast SE sequence (TR/TE ⫽ 5000/100, 8 echo train length). The parameters used in both sequences include: one acquisition of 128 lines of data, 256 ⴱ 128 matrix, and FOV of 20 cm. Signal intensity was measured from a round-shaped (200 mm2) region of interest (ROI) placed in the center of each phantom. And signal intensity ratios of the blood phantom to saline phantom (T1- and T2-ratios) were calculated to normalize the signal intensity of the MR measurements. 2.1.2. Diffusion coefficient measurement Diffusion-weighted images were acquired by using the diffusion-weighted, single-section, SE type of echo-planar imaging (EPI) sequence that combines the diffusion gradients before and after the 180-degree pulse along the slice select, phase encoding and frequency encoding directions.
The preparation pulse was followed by 16 shots of EPI with 2000 msec of TR and 50 msec of TE. A total of 128 lines of data were acquired. The other parameters were a field of vision (FOV) of 20 cm, a 128 ⫻ 128 matrix, a 10-mm section thickness, and one data acquisition. The value b is the gradient factor (sec/mm2) that depends on the magnetic field gradients applied before data acquisition, and given by b⫽
冘
1⫽X,Y,Z
␥2␦2G2(⌬⫺␦/3)
(1)
where 1 is axis of gradient pulses, ␥ is the gyromagnetic ratio (42.576 MHz/T), ␦ is the duration of the gradient (10 msec), ⌬ is the time between gradients (25 msec), G is the strength of the gradient (0 and 20 mT/m), and the b values (b1 and b0) given by the Eq. (1) are 186 and 0 sec/mm2, respectively. Signal measurements were performed using the same ROI as the T1- and T2-ratio measurements. When two diffusion-weighted images had different b values, diffusion value (D) (mm2/sec) were calculated by using the following equation: D⫽⫺[ln (S1)⫺ ln (S0)]/(b1⫺b0)
(2)
where S0 and S1 are the signal intensities at the ROI with two different b values (b0 and b1). 2.2. In vivo study The in vivo MR analysis was limited to 26 pelvic cystic lesions of 26 female patients (age range, 12–79 years; mean age, 44 years) that were operatively or laparoscopically proved after the MR study. The pelvic cystic lesions consisted of 14 endometrial cysts, 5 simple ovarian cysts, 2 corpus luteum cysts, 1 paratubal cyst, 1 paraovarian cyst, 1 tubo-ovarian cyst, 1 lymphocele, and 1 mesenteric cyst. These were classified to two groups; 14 endometrial and 12 other pelvic cysts. At 1.5-T whole-body imager with body coil was used for the in vivo MR study. T1-weighted SE (500/14 [TR/TE]), T2-weighted fast SE (5000/95–105, 8 echo train length) images were obtained in all patients. The other parameters were a 32 ⫻ 24 cm FOV, 512 ⫻ 224 matrix, a 7-mm slice thickness, and 4 acquisitions. In 11 patients, T1-weighted MR imaging was repeated after intravenous administration of 0.1 mmoles/kg of gadopentetate dimeglumine (GdDTPA) (Magnevist, Schering AG, Berlin, Germany). The enhanced studies were performed after the SE, fast SE, and EPI scans. EPI scans were performed using a single section SE type of preparation with 50 msec of TE modified by adding large diffusion gradients on both sides of the 180-degree pulse along the slice select, phase encoding, and frequency encoding directions. The use of a shorter TE was due to the need for retention of sufficient signal/noise ratio at the ROI, even if hypointensity on T2-weighted images was significant. The amplitude of the diffusion gradient pulses (G) was 20 and 0 mT/m, respectively, the duration of each diffusion gradient pulse (␦) was 10 msec, and the interval separating
T. Moteki et al. / Magnetic Resonance Imaging 20 (2002) 463– 470
465
Fig. 1. Changing rates of diffusion values, signal intensity ratios to saline on T1-weighted SE (TR/TE ⫽ 500/14 msec) and T2-weighted fast SE (5000/100) images (T1- and T2-ratios) as function of gradually progressed blood oxidation. Changing rates of T1- and T2-ratios are much larger than ones of diffusion value.
the onset of diffusion gradient pulses (⌬) was 25 ms. The b values (b1 and b0) given by the Eq. (1) were 186 and 0 sec/mm2, respectively. The preparation sequence was followed by a one-shot EPI readout. Fat suppression using water-selective excitation was introduced into the sequence, and 64 lines of k-space data were acquired. The other parameters were an FOV of 35 cm, a 64 ⫻ 128 matrix, a 10-mm slice thickness, and 4 data acquisitions with a 3.0second TR. The total EPI scan time was 10 sec, and we instructed the patients to hold their breath during scanning after a deep inspiration and consecutive expiration. These EPI were scanned at the same slice location showing the maximum diameter of the cystic lesion on axial T2weighted images. Acquired EPI images with and without diffusion gradients were visually evaluated at the same window and level settings. The signal intensity of the cystic contents of each lesion was measured on a set of EPI (T1- and T2-weighted) images. A circular ROI was placed within each cyst loculus of more than 2.0 cm in diameter on the selected slice, and 22 loculi of 14 endometrial cysts and 13 loculi of 12 other pelvic cysts were evaluated. The outer margin of the ROI was kept at least 5 mm away from the capsule and the septa. On T1- and T2-weighted images, another ROI also was placed wherever possible in the gluteus maximus muscle, posterior to the ROI of the cystic lesion. Ratios of cystic lesion signal to the muscle signal (T1-ratio and T2-ratio) were calculated for both T1- and T2-weighted images, respectively. We manually shifted the round measurement cursor to retain the same ROI as much as possible, when there was distortion of EPI images related to susceptibility artifacts. ADC measurements reported herein were made by means of analysis of ROI using the Eq. (2). To determine the significance of difference in ADC,
T1-ratio and T2-ratio between the endometrial cysts and other pelvic cysts, we used Student’s t-test for normally distributed data (p ⬎ 0.05 on F analysis). When data were not normally distributed (p ⬉ 0.05 on F analysis), we used Welch’s t-test. Differences were deemed statistically significant when the p value was less than 0.05 on the selected t-test. In both the endometrial and other pelvic cysts, correlation coefficients between ADC, T1-ratio, and T2-ratio were calculated to evaluate whether ADC was correlated well to T1-ratio or T2-ratio. Discriminant analysis was applied to obtain the cutoff value of the best ADC, T1-ratio and T2-ratio differentiating endometrial cysts from other pelvic cysts, and to obtain the discriminant rate when using one of or combination of these three values.
3. Results 3.1. Phantom study The rates of change of T1-ratio, T2-ratio and diffusion values as a function of time from the generation of the blood phantom are shown in Fig. 1. T1-ratio cumulatively increased until the 8th day and subsequently plateaued (from 0.965 to 2.23). T1 value of the phantom at the 14th day was 334 msec (using T1-weighted spine-echo sequences 600 – 1500/20 [TR/TE]), and this value almost corresponds to the T1 value of 100% methemoglobin in blood with a 50% hematocrit [11]. T2-ratio increased (from 1 to 3.34) until the 4th day, peaked between the 4th and 10th days, and then slightly decreased. Diffusion value changed little (from 0.882 to 1.08), and minimally decreased during the first 6 days.
466
T. Moteki et al. / Magnetic Resonance Imaging 20 (2002) 463– 470
Fig. 2. Changing rates of diffusion values, T1- and T2-ratios as function of the original blood phantom concentration (100, 80, 60, 40, 20 and 10% of the contents). Diffusion value and T1-ratio were linearly changed, although T2-ratio reveals convex curve with the peak around 20% blood phantom concentration.
The rates of change of T1-ratio, T2-ratio and diffusion values as a function of blood phantom concentration are shown in Fig. 2. T1-ratio was almost linearly decreased as blood concentration decreased. Diffusion value was almost linearly increased as blood concentration decreased. T2ratio was almost linearly increased as blood concentration decreased until the 40% blood phantom concentration was reached, thereafter it peaked at 20%, and then mildly decreased at 10%.
(31/35), and T2-ratio was 71.4% (25/35). ADC value in combination with T1-ratio (and/or T2-ratio) showed higher discriminant rate (94.3% (33/35)) than combination of T1and T2-ratios (88.6% (31/35)). All 4 loculi of 3 endometrial cysts that were incorrectly classified as other kinds of pelvic cysts based on T1-ratio and T2-ratio (Fig. 5) were correctly classified based on ADC value. On the other hand, all 3 other pelvic cysts that were incorrectly classified as endometrial cysts based on ADC value were correctly classified based on T1- and T2-ratios.
3.2. In vivo study Mean and standard deviation (mean ⫾ SD) of ADC value, T1- and T2-ratios of endometrial cysts were 0.00091 ⫾ 0.00047 mm2/sec, 2.88 ⫾ 1.08, and 3.77 ⫾ 1.79. These values of other pelvic cysts were 0.00282 ⫾ 0.00080 mm2/sec, 0.77 ⫾ 0.22, and 5.71 ⫾ 1.34. Five other ovarian cysts showed high ADC values greater than the diffusion coefficient of pure water at 37°C (0.00324 mm2/sec). Endometrial cysts were apt to have lower ADC, lower T2ratio, and higher T1-ratio than other pelvic cysts (Fig. 3). Differences in T1-ratio, T2-ratio, and ADC value between the two groups were significant (P ⬍ 0.001). However, these three values were not correlated well with each other (correlation coefficients ⬍ P0.15P) except for relatively higher correlation between T1-ratio and T2-ratio in endometrial cysts (correlation coefficient ⫽ ⫺0.46). The cutoff values of ADC, T1- and T2-ratios between the two groups obtained from discriminant analysis were 0.00186 mm2/sec, 1.82, and 4.73, respectively. Histograms of discriminant scores are presented in Fig. 4. Discriminant rate using ADC was 91.4% (32/35), T1-ratio was 88.6%
4. Discussion The ectopic endometrium is sensitive to ovarian hormonal stimulation and undergoes repeated cycles of hemorrhage, leading to the development of endometrial cysts. The varying age of the hemorrhage in the cysts tends to have a wide range of signal behavior [13,14]. High intensity in an endometrial cyst on T1-weighted image is mainly attributed to the paramagnetism of methemoglobin by which the proton-electron dipole-dipole interaction (PEDD) between nuclear magnetic moments occurs because the electron magnetic moment from the paramagnetism is 700 times greater than that of the proton [15]. In our phantom study, gradually increasing blood oxidation caused the T1ratio to increase. This reveals that signal intensity on T1weighted image is greatly affected by methemoglobin-related paramagnetic effects. T2-ratio increased until the 4th day, peaked between the 4th and 10th days, and slightly decreased subsequently. Since T1 shortening is observed at lower concentrations of the paramagnetism than are needed
T. Moteki et al. / Magnetic Resonance Imaging 20 (2002) 463– 470
467
Fig. 3. Paraovarian cyst. (a) A postcontrast T1-weighted image shows a homogeneous low intensity lesion without enhancing sold component. (b) On T2-weighted image, the lesion reveals homogeneous high intensity. (c) EPI with diffusion gradients (lower) demonstrates significant signal decay in the lesion, compared with EPI without diffusion gradients (upper). Calculated ADC value is 0.00340 mm2/sec.
468
T. Moteki et al. / Magnetic Resonance Imaging 20 (2002) 463– 470
Fig. 4. Histograms of discriminant scores between endometrial cysts and other pelvic cysts as to T2-ratio (T2), T1-ratio (T1), ADC value, and combinations of them. ADC value shows higher discriminant rate than T2-ratio or T1-ratio. ADC value in combination with T1-ratio (and/or T2-ratio) demonstrates higher discriminant rate than the combination of T1-ratio and T2-ratio.
to produce T2 shortening [15], the T2-ratio decrease after the 10th day may be caused by the paramagnetic effect from methemoglobin. However, paramagnetic effects cannot explain either the increase of T2-ratio until the 4th day or the occurrence of peaks between the 4th and 10th days. Since T2 shortening is mainly caused by the internal magnetic nonhomogeneity that leads to irreversible loss of phase coherence [15], the increase of free hemoglobin and the decrease of intracellular hemoglobin in relation to gradual red blood cell lysis is thought to reduce magnetic nonhomogeneity and thereby increase T2 ratio [11]. In spite of markedly varied T1- and T2-ratios of the blood phantom during 14 days (changed 127 and 234%, respectively), diffusion value was little changed (19.8%). These results reveal that the diffusion value is almost unaffected by both the methemoglobin-related paramagnetic effect and the diminution of magnetic nonhomogeneity by gradual red blood cell lysis that strongly affect signal intensity on T1- and T2-weighted images. A tendency for the diffusion value to decrease slightly during the first 6 days may reflect a small increase in the volume of bound water around hemoglobin, inasmuch as the amount of free hemoglobin increase (vide the following paragraph). With increasing protein concentration, the T1 and T2 relaxation time progressively shortens [16,17]. This effect has been ascribed to slowing of the rotational motion of water molecules as they bind to and form structure in the region immediately surrounding protein molecules (i.e., become bound water) [18]. In our phantom study, T1-ratio declined almost linearly as blood concentration decreased, and the relationship of T2-ratio to blood concentration described a convex curve with the peak around 20% blood
phantom concentration. Considering that whole blood has about 20% solutes (or 18.5% protein), these T1- and T2ratio results for blood concentration are very similar to the previously reported results for protein; signal intensity on T1-weighted SE (TR/TE ⫽ 400/30) and T2-weighted SE (2000/100) images revealed convex curves with peaks around 25% and 5% protein concentration, respectively [19]. In our study, diffusion value was a nearly linear function of blood concentration, which is almost the same result as reported for protein solution [20]. The diffusion value obtained for protein solution is considered to be mainly dependent on the free water fraction, since water bound to protein has little effect on diffusion values, as seen in the following expression, D ⬵ 共1 ⫺ 兲 D f (as D b⬍⬍D f)
(3)
Here D is the self-diffusion coefficient, D b is the selfdiffusion coefficient of “bound” water, D f the self-diffusion coefficient of “free” water, and the bounded water fraction (1 ⫺ the free water fraction) [21]. Therefore, diffusion value for blood is considered to be dependent on free water fraction, just as is the diffusion value for protein solution. Paramagnetic iron derived from methemoglobin in relation to blood oxidation is mainly responsible for the signal increase on T1-weighted images, and diminution of magnetic nonhomogeneity in relation to gradual red blood cell lysis causes the signal increase on T2-weighted images [11]. However, as shown in our phantom study, diffusion value is almost independent of both blood oxidation and red blood cell lysis. Hence, we consider that the extent of blood oxidation and red blood cell lysis present in endometrial cysts are the main reasons for the lower correlation of ADC
T. Moteki et al. / Magnetic Resonance Imaging 20 (2002) 463– 470
469
Fig. 5. Endometrial cyst that was incorrectly classified to another pelvis cyst when using T1-ratio and T2-ratio, and correctly diagnosed when using ADC value. Content of the lesion was aged blood. (a) A postcontrast T1-weighted image shows a low intensity lesion with thickened capsular structure. (b) On T2-weighted image, the lesion reveals homogeneous high intensity. (c) EPI with diffusion gradients (lower) demonstrates mild signal decay in the lesion compared with EPI without diffusion gradients (upper). Calculated ADC value is 0.00148 mm2/sec.
470
T. Moteki et al. / Magnetic Resonance Imaging 20 (2002) 463– 470
value to T1- or T2-ratio in endometrial cysts. Signal intensity and diffusion value behavior in blood (both functions of blood concentration) are very similar to the reported results for protein concentration, as described in the previous paragraph; diffusion value and signal intensity on T1-weighted image are linearly related to blood or protein concentration, and the relation of signal intensity on T2-weighted image to concentration forms a convex curve with peak at around 5% protein or 20% blood phantom concentration (having 19.6% whole blood (or 3.6% protein) concentration). These results also explain why the correlation of ADC value and T1-ratio to T2-ratio in endometrial and other pelvic cysts is lower while ADC, T1-ratio and T2-ratio differences between the two groups are statistically significant. However, these cannot explain why for other pelvic cysts, the correlation of ADC value to T1-ratio is also lower. This may be explained by the presence of minimal hemorrhagic components that affect T1-ratios and/or of pseudodiffusion effect such as turbulent flow in the cystic contents by physiologic motions that affect pure diffusion values [22] because 5/12 other ovarian cysts showed high ADC values greater than the diffusion coefficient of pure water at 37°C (0.00324 mm2/ sec). In our in vivo study, ADC value showed substantially comparable ability of differentiation between endometrial and other pelvic cysts with T1- or T2-ratio, and the combination of ADC value and T1-ratio (and/or T2-ratio) could improve the differentiation between the two groups in contrast with the combination of T1- and T2-ratios. Furthermore, the cases misdiagnosed by the combination of T1and T2-ratios were different from the ones misdiagnosed by ADC value, probably because of the lower correlation of ADC value to T1-ratio and T2-ratio in endometrial and other pelvic cysts. Therefore, we think that ADC value may become a new diagnostic tool for the estimation of blood and/or protein concentration within cystic lesions because diffusion value is a nearly linear function of blood and protein concentration, almost independent of blood oxidation levels (or methemoglobin-related paramagnetic effects), and also almost independent of red blood lysis (or diminution of magnetic nonuniformity by increased free hemoglobin) that significantly affect signal intensity.
5. Conclusion Even though it was poorly correlated with signal intensity, ADC value has about the same ability to differentiate endometrial from other pelvic cysts as does signal intensity on T1- and T2-weighted images. We believe poor correlation between ADC value and signal intensity occurs because ADC value is almost linearly dependent on blood concentration and almost independent of blood oxidation and red blood cell lysis, two factors that significantly affect signal intensity.
References [1] Le Bihan D, Breston E, Lallemand D, Grenier P, Cabanis EA, LavalJeantet M. MR imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurologic disorders. Radiology 1986; 161:401–7. [2] Ahn CB, Lee SY, Nalcioglu O, Cho ZH. The effects of random directional distributed flow in nuclear magnetic resonance imaging. Med Phys 1987;14:43– 8. [3] Le Bihan D, Breton E, Lallemand D, Aubin ML, Vignaud J, Lavaljeantet M. Separation of diffusion and perfusion in intravoxel incoherent motion (IVIM) MR imaging. Radiology 1988;168:497–505. [4] Young IR, Hall AS, Bryant DJ, Tomas DG, Gill SS, Dubowitz LM, Cowan G, Pennock JM, Bydder GM. Assessment of brain perfusion with MR imaging. J Comput Assist Tomogr 1988;12:721–7. [5] Moonen CTW, Van Zijl PCM, Le Bihan D, DesPres D. In vivo NMR diffusion spectroscopy. 31P application to phosphorus metabolites in muscle. Magn Reson Med 1990;13:467–77. [6] Moseley ME, Kucharczyk J, Mintorovitch J, Cohen Y, Kurhanewicz J, Derugin N, Asgari H, Norman D. Diffusion-weighted MR imaging of acute stroke: correlation with T2-weighted and magnetic susceptibility-enhanced MR imaging in cats. AJNR 1990;11:423–9. [7] Warach S, Chien D, Li W, Ronthal M, Edelman RR. Fast magnetic resonance diffusion-weighted imaging of acute human stroke. Neurology 1992;42:1717–23. [8] Namimoto T, Yamashita Y, Sumi S, Tang Y, Takahashi M. Focal liver masses: characterization with diffusion-weighted echo-planar MR imaging. Radiology 1997;204:739 – 44. [9] Namimoto T, Yamashita Y, Mitsuzaki K, Nakayama Y, Tang Y, Takahashi M. Measurement of the apparent diffusion coefficient in diffuse renal disease by diffusion-weighted echo-planar MR imaging. J Magn Reson Imaging 1999;9:832–7. [10] Sugimura K, Takemori M, Sugiura M, Okizuka H, Kano M, Ishida T. The value of magnetic resonance relaxation time in staging ovarian endometrial cysts. Br J Radiol 1992;65:502– 6. [11] Janick PA, Hackney DB, Grossman RI, Asakura T. MR imaging of various oxidation states of intracellular and extracellular hemoglobin. AJNR 1991;12:891–7. [12] Kuma F. Properties of methemoglobin reductase and kinetic study of methemoglonin reduction. J Biol Chem 1981;256:5518 –23. [13] Nishimura K, Togashi K, Itoh K, Fujisawa I, Noma S, Kawamura Y, Nakano Y, Itoh H, Torizuka K, Ozawa H. Endometrial cysts of the ovary: MR imaging. Radiology 1987;162:315– 8. [14] Zawin M, McCarthy S, Scoutt L, Comite F. Endometriosis: appearance and detection at MR imaging. Radiology 1989;171:693– 6. [15] William G, Bradley WG. Hemorrhage and brain iron. In: Stark DD, Bradley WG, editors. Magnetic Resonance Imaging, 2nd ed. St. Louis: Mosby Year Book, 1992, pp. 721– 69. [16] Koenig SH, Schillinger WE. Nuclear magnetic relaxation dispersion in protein solution. J Biol Chem 1969;244:6520 – 6. [17] Fullerton GD, Finnie MF, Hunter KE, Ord VA, Cameron IL. The influence of macromolecular polymerization on spine-lattice relaxation of aqueous solution. Magn Reson Imaging 1987;5:353–70. [18] Bloembergen N, Purcell EM, Pound RV. Relaxation effects in nuclear magnetic resonance absorption. Phys Rev 1948;73:679 –712. [19] Som PM, Dillon WP, Fullerton GD, Zimmerman RA, Rajagopalan B, Marom Z. Chronically obstructed sinonasal secretions: observations on T1 and T2 shortening. Radiology 1989;172:515–20. [20] Baranowska HM, Olszewski KJ. The hydration of proteins in solutions by self-diffusion coefficients. NMR study. Biochim Biophys Acta 1996;1289:312– 4. [21] Blinc A, Lahajnar G, Blinc R, Zidansek A, Sepe A. Proton NMR study of the state of water in fibrin gels, plasma, and blood clots. Magn Reson Med 1990;14:105–22. [22] Edelman RR, Wielopolski P, Schmitt F. Echo-planar MR imaging. Radiology 1994;192:600 –12.
Relationship between apparent diffusion coefficient and signal intensity in endometrial and other pelvic cysts Takao Motekia,*, Hiroyuki Horikoshic, Keigo Endob a
The Department of Diagnostic Radiology, Gunma University Hospital, 3-39-15, Showa-machi, Maebashi, Gunma 371-0034, Japan b The Department of Nuclear Medicine, Gunma University Hospital, 3-39-15, Showa-machi, Maebashi, Gunma 371-0034, Japan c The Department of Radiology, Gunma Cancer Tomo Hospital, 617-1, Takabayashinishi-machi, Otashi, Gunma 373-0828, Japan Received 20 February 2002; accepted 22 May 2002
Abstract We evaluated whether apparent diffusion coefficient (ADC) value is more useful than signal intensity for differentiating endometrial cysts from other pelvic cysts. In an in vitro study, signal intensity and diffusion coefficients were measured in whole blood phantoms in which blood oxidation was gradually increased and concentration subsequently diluted. Although both signal intensity and diffusion value were largely affected by blood concentration, diffusion value was almost independent of blood oxidation and red blood cell lysis-related diminution of magnetic nonhomogeneity, both factors greatly affecting signal intensity on T1- and T2-weighted images. In an in vivo study, differentiation between endometrial and other pelvic cysts was attempted by means of ADC values and signal ratios of cysts to muscles on T1- and T2-weighted images (T1- and T2-ratios). Endometrial cysts tended to show lower T2-ratios, higher T1-ratios, and lower ADC values than other pelvic cysts ( p ⬍ 0.001). However, ADC values were not correlated with T1- and T2-ratios ( p ⬍ P0.15P). The ability of ADC value to discriminate between these two groups (discriminant rate, 91.4%) was higher than that of T2-ratio (71.4%) or T1-ratio (88.6%). If combined, ADC and T1-ratio (or T2-ratio) showed higher discriminant rate (94.3%) than the combination of T1- and T2 ratios (88.6%). ADC value might be useful for evaluating the blood concentration of a cystic lesion, because diffusion value is more closely related to blood concentration and almost independent of blood oxidation and red blood cell lysis that largely affect signal intensity. © 2002 Elsevier Science Inc. All rights reserved. Keywords: Diffusion imaging; Blood; Fast MR imaging; Cyst
1. Introduction Interest in the potential capability of magnetic resonance (MR) to image and measure molecular diffusion has been considerable [1– 4]. Diffusion measurements in vivo are useful for tissue characterization because they provide information on the mobility of water [1] or cell metabolites [5]. Diffusion measurements offer data on molecular displacements that are much smaller than the resolution of an MR image (typically a few millimeters). At present, the most important clinical application of this imaging technique is the detection and characterization of cerebral ischemia [6,7]. Recently, a diffusion-weighted ultrafast MRI technique has become available with the introduction of a new generation of MR scanners, and enabled in vivo mea* Corresponding author. Tel.: ⫹81-272-20-8621; fax: ⫹81-272-208409. E-mail address: [email protected] (T. Moteki).
surements of diffusion in abdominal organs such as liver and kidney using echo-planar imaging (EPI) [8,9]. Endometrial cysts exhibiting varying stages of hemorrhage tend to display a wide range of signal behavior. High intensity in the lesion on T1-weighted images is attributed to paramagnetic effects from methemoglobin [10], and varying intensity on T2-weighted images may be due to red blood cell lysis [11]. However, the relationship between signal intensity and diffusion coefficient in connection with blood oxidation and concentration has not been evaluated. Hence, this study was undertaken to determine whether blood oxidation-related and concentration-related changes of diffusion values were similar to ones of signal intensity on T1- and T2-weighted images. For this purpose, blood phantoms with gradually increased levels of oxidation, and at various blood concentrations, were used. In an in vivo study, differentiation between endometrial and other pelvic cysts was attempted using apparent diffusion coefficient (ADC) and signal intensity. In other words, we attempted to
0730-725X/02/$ – see front matter © 2002 Elsevier Science Inc. All rights reserved. PII: S 0 7 3 0 - 7 2 5 X ( 0 2 ) 0 0 5 2 4 - 6
464
T. Moteki et al. / Magnetic Resonance Imaging 20 (2002) 463– 470
determine whether ADC value were useful to evaluate hemorrhagic cysts.
2. Materials and methods 2.1. Phantom study Phantoms were developed to determine how diffusion value as well as signal intensity on T1- and T2-weighted images were affected by blood oxidation and blood concentration. Ten ml of blood was obtained from the medial cubital veins of a healthy human volunteer (hematocrit, 46%) and heparinized (100 U/ml). Immediately after extraction, the blood was transferred to a closed plastic syringe, and stored for 2 hours at 24°C room temperature with 5 ml of 100% N2. Then, 0.1 g/ml of sodium bicarbonate was added to raise pH to 8.6, which inhibits methemoglobin reductase [12]. Another phantom, to which 100% saline was added, was used to normalize signal intensity. After the first MR measurements, the blood phantom was ventilated with 5 ml of 100% O2. Immediately after the ventilation, second MR measurements were performed. Then, MR measurements were performed at 1.5 hr, 3 hrs, 1 day, 2 days, 3 days, 4 days, 6 days, 8 days, 10 days, 12 days and 14 days after the second MR measurement. Between MR measurements, the blood phantoms were ventilated with 100% O2. Immediately after the 14-day measurement, the blood phantom was gradually diluted (to 80, 60, 40, 20 and 10% of the original phantom contents) with saline, and MR measurements again were performed at each of these concentrations. A 1.5-T whole-body MR imager (Signa Horizon; GE, Milwaukee, USA) with head coil was used for the in vitro MR study. MR measurements for phantoms were performed at 24°C (room temperature). 2.1.1. Signal intensity measurement Signal intensity measurements were performed with the use of a T1-weighted spin-echo (SE) pulse sequence (repetition time/echo time (TR/TE ⫽ 500/14) and a T2weighted fast SE sequence (TR/TE ⫽ 5000/100, 8 echo train length). The parameters used in both sequences include: one acquisition of 128 lines of data, 256 ⴱ 128 matrix, and FOV of 20 cm. Signal intensity was measured from a round-shaped (200 mm2) region of interest (ROI) placed in the center of each phantom. And signal intensity ratios of the blood phantom to saline phantom (T1- and T2-ratios) were calculated to normalize the signal intensity of the MR measurements. 2.1.2. Diffusion coefficient measurement Diffusion-weighted images were acquired by using the diffusion-weighted, single-section, SE type of echo-planar imaging (EPI) sequence that combines the diffusion gradients before and after the 180-degree pulse along the slice select, phase encoding and frequency encoding directions.
The preparation pulse was followed by 16 shots of EPI with 2000 msec of TR and 50 msec of TE. A total of 128 lines of data were acquired. The other parameters were a field of vision (FOV) of 20 cm, a 128 ⫻ 128 matrix, a 10-mm section thickness, and one data acquisition. The value b is the gradient factor (sec/mm2) that depends on the magnetic field gradients applied before data acquisition, and given by b⫽
冘
1⫽X,Y,Z
␥2␦2G2(⌬⫺␦/3)
(1)
where 1 is axis of gradient pulses, ␥ is the gyromagnetic ratio (42.576 MHz/T), ␦ is the duration of the gradient (10 msec), ⌬ is the time between gradients (25 msec), G is the strength of the gradient (0 and 20 mT/m), and the b values (b1 and b0) given by the Eq. (1) are 186 and 0 sec/mm2, respectively. Signal measurements were performed using the same ROI as the T1- and T2-ratio measurements. When two diffusion-weighted images had different b values, diffusion value (D) (mm2/sec) were calculated by using the following equation: D⫽⫺[ln (S1)⫺ ln (S0)]/(b1⫺b0)
(2)
where S0 and S1 are the signal intensities at the ROI with two different b values (b0 and b1). 2.2. In vivo study The in vivo MR analysis was limited to 26 pelvic cystic lesions of 26 female patients (age range, 12–79 years; mean age, 44 years) that were operatively or laparoscopically proved after the MR study. The pelvic cystic lesions consisted of 14 endometrial cysts, 5 simple ovarian cysts, 2 corpus luteum cysts, 1 paratubal cyst, 1 paraovarian cyst, 1 tubo-ovarian cyst, 1 lymphocele, and 1 mesenteric cyst. These were classified to two groups; 14 endometrial and 12 other pelvic cysts. At 1.5-T whole-body imager with body coil was used for the in vivo MR study. T1-weighted SE (500/14 [TR/TE]), T2-weighted fast SE (5000/95–105, 8 echo train length) images were obtained in all patients. The other parameters were a 32 ⫻ 24 cm FOV, 512 ⫻ 224 matrix, a 7-mm slice thickness, and 4 acquisitions. In 11 patients, T1-weighted MR imaging was repeated after intravenous administration of 0.1 mmoles/kg of gadopentetate dimeglumine (GdDTPA) (Magnevist, Schering AG, Berlin, Germany). The enhanced studies were performed after the SE, fast SE, and EPI scans. EPI scans were performed using a single section SE type of preparation with 50 msec of TE modified by adding large diffusion gradients on both sides of the 180-degree pulse along the slice select, phase encoding, and frequency encoding directions. The use of a shorter TE was due to the need for retention of sufficient signal/noise ratio at the ROI, even if hypointensity on T2-weighted images was significant. The amplitude of the diffusion gradient pulses (G) was 20 and 0 mT/m, respectively, the duration of each diffusion gradient pulse (␦) was 10 msec, and the interval separating
T. Moteki et al. / Magnetic Resonance Imaging 20 (2002) 463– 470
465
Fig. 1. Changing rates of diffusion values, signal intensity ratios to saline on T1-weighted SE (TR/TE ⫽ 500/14 msec) and T2-weighted fast SE (5000/100) images (T1- and T2-ratios) as function of gradually progressed blood oxidation. Changing rates of T1- and T2-ratios are much larger than ones of diffusion value.
the onset of diffusion gradient pulses (⌬) was 25 ms. The b values (b1 and b0) given by the Eq. (1) were 186 and 0 sec/mm2, respectively. The preparation sequence was followed by a one-shot EPI readout. Fat suppression using water-selective excitation was introduced into the sequence, and 64 lines of k-space data were acquired. The other parameters were an FOV of 35 cm, a 64 ⫻ 128 matrix, a 10-mm slice thickness, and 4 data acquisitions with a 3.0second TR. The total EPI scan time was 10 sec, and we instructed the patients to hold their breath during scanning after a deep inspiration and consecutive expiration. These EPI were scanned at the same slice location showing the maximum diameter of the cystic lesion on axial T2weighted images. Acquired EPI images with and without diffusion gradients were visually evaluated at the same window and level settings. The signal intensity of the cystic contents of each lesion was measured on a set of EPI (T1- and T2-weighted) images. A circular ROI was placed within each cyst loculus of more than 2.0 cm in diameter on the selected slice, and 22 loculi of 14 endometrial cysts and 13 loculi of 12 other pelvic cysts were evaluated. The outer margin of the ROI was kept at least 5 mm away from the capsule and the septa. On T1- and T2-weighted images, another ROI also was placed wherever possible in the gluteus maximus muscle, posterior to the ROI of the cystic lesion. Ratios of cystic lesion signal to the muscle signal (T1-ratio and T2-ratio) were calculated for both T1- and T2-weighted images, respectively. We manually shifted the round measurement cursor to retain the same ROI as much as possible, when there was distortion of EPI images related to susceptibility artifacts. ADC measurements reported herein were made by means of analysis of ROI using the Eq. (2). To determine the significance of difference in ADC,
T1-ratio and T2-ratio between the endometrial cysts and other pelvic cysts, we used Student’s t-test for normally distributed data (p ⬎ 0.05 on F analysis). When data were not normally distributed (p ⬉ 0.05 on F analysis), we used Welch’s t-test. Differences were deemed statistically significant when the p value was less than 0.05 on the selected t-test. In both the endometrial and other pelvic cysts, correlation coefficients between ADC, T1-ratio, and T2-ratio were calculated to evaluate whether ADC was correlated well to T1-ratio or T2-ratio. Discriminant analysis was applied to obtain the cutoff value of the best ADC, T1-ratio and T2-ratio differentiating endometrial cysts from other pelvic cysts, and to obtain the discriminant rate when using one of or combination of these three values.
3. Results 3.1. Phantom study The rates of change of T1-ratio, T2-ratio and diffusion values as a function of time from the generation of the blood phantom are shown in Fig. 1. T1-ratio cumulatively increased until the 8th day and subsequently plateaued (from 0.965 to 2.23). T1 value of the phantom at the 14th day was 334 msec (using T1-weighted spine-echo sequences 600 – 1500/20 [TR/TE]), and this value almost corresponds to the T1 value of 100% methemoglobin in blood with a 50% hematocrit [11]. T2-ratio increased (from 1 to 3.34) until the 4th day, peaked between the 4th and 10th days, and then slightly decreased. Diffusion value changed little (from 0.882 to 1.08), and minimally decreased during the first 6 days.
466
T. Moteki et al. / Magnetic Resonance Imaging 20 (2002) 463– 470
Fig. 2. Changing rates of diffusion values, T1- and T2-ratios as function of the original blood phantom concentration (100, 80, 60, 40, 20 and 10% of the contents). Diffusion value and T1-ratio were linearly changed, although T2-ratio reveals convex curve with the peak around 20% blood phantom concentration.
The rates of change of T1-ratio, T2-ratio and diffusion values as a function of blood phantom concentration are shown in Fig. 2. T1-ratio was almost linearly decreased as blood concentration decreased. Diffusion value was almost linearly increased as blood concentration decreased. T2ratio was almost linearly increased as blood concentration decreased until the 40% blood phantom concentration was reached, thereafter it peaked at 20%, and then mildly decreased at 10%.
(31/35), and T2-ratio was 71.4% (25/35). ADC value in combination with T1-ratio (and/or T2-ratio) showed higher discriminant rate (94.3% (33/35)) than combination of T1and T2-ratios (88.6% (31/35)). All 4 loculi of 3 endometrial cysts that were incorrectly classified as other kinds of pelvic cysts based on T1-ratio and T2-ratio (Fig. 5) were correctly classified based on ADC value. On the other hand, all 3 other pelvic cysts that were incorrectly classified as endometrial cysts based on ADC value were correctly classified based on T1- and T2-ratios.
3.2. In vivo study Mean and standard deviation (mean ⫾ SD) of ADC value, T1- and T2-ratios of endometrial cysts were 0.00091 ⫾ 0.00047 mm2/sec, 2.88 ⫾ 1.08, and 3.77 ⫾ 1.79. These values of other pelvic cysts were 0.00282 ⫾ 0.00080 mm2/sec, 0.77 ⫾ 0.22, and 5.71 ⫾ 1.34. Five other ovarian cysts showed high ADC values greater than the diffusion coefficient of pure water at 37°C (0.00324 mm2/sec). Endometrial cysts were apt to have lower ADC, lower T2ratio, and higher T1-ratio than other pelvic cysts (Fig. 3). Differences in T1-ratio, T2-ratio, and ADC value between the two groups were significant (P ⬍ 0.001). However, these three values were not correlated well with each other (correlation coefficients ⬍ P0.15P) except for relatively higher correlation between T1-ratio and T2-ratio in endometrial cysts (correlation coefficient ⫽ ⫺0.46). The cutoff values of ADC, T1- and T2-ratios between the two groups obtained from discriminant analysis were 0.00186 mm2/sec, 1.82, and 4.73, respectively. Histograms of discriminant scores are presented in Fig. 4. Discriminant rate using ADC was 91.4% (32/35), T1-ratio was 88.6%
4. Discussion The ectopic endometrium is sensitive to ovarian hormonal stimulation and undergoes repeated cycles of hemorrhage, leading to the development of endometrial cysts. The varying age of the hemorrhage in the cysts tends to have a wide range of signal behavior [13,14]. High intensity in an endometrial cyst on T1-weighted image is mainly attributed to the paramagnetism of methemoglobin by which the proton-electron dipole-dipole interaction (PEDD) between nuclear magnetic moments occurs because the electron magnetic moment from the paramagnetism is 700 times greater than that of the proton [15]. In our phantom study, gradually increasing blood oxidation caused the T1ratio to increase. This reveals that signal intensity on T1weighted image is greatly affected by methemoglobin-related paramagnetic effects. T2-ratio increased until the 4th day, peaked between the 4th and 10th days, and slightly decreased subsequently. Since T1 shortening is observed at lower concentrations of the paramagnetism than are needed
T. Moteki et al. / Magnetic Resonance Imaging 20 (2002) 463– 470
467
Fig. 3. Paraovarian cyst. (a) A postcontrast T1-weighted image shows a homogeneous low intensity lesion without enhancing sold component. (b) On T2-weighted image, the lesion reveals homogeneous high intensity. (c) EPI with diffusion gradients (lower) demonstrates significant signal decay in the lesion, compared with EPI without diffusion gradients (upper). Calculated ADC value is 0.00340 mm2/sec.
468
T. Moteki et al. / Magnetic Resonance Imaging 20 (2002) 463– 470
Fig. 4. Histograms of discriminant scores between endometrial cysts and other pelvic cysts as to T2-ratio (T2), T1-ratio (T1), ADC value, and combinations of them. ADC value shows higher discriminant rate than T2-ratio or T1-ratio. ADC value in combination with T1-ratio (and/or T2-ratio) demonstrates higher discriminant rate than the combination of T1-ratio and T2-ratio.
to produce T2 shortening [15], the T2-ratio decrease after the 10th day may be caused by the paramagnetic effect from methemoglobin. However, paramagnetic effects cannot explain either the increase of T2-ratio until the 4th day or the occurrence of peaks between the 4th and 10th days. Since T2 shortening is mainly caused by the internal magnetic nonhomogeneity that leads to irreversible loss of phase coherence [15], the increase of free hemoglobin and the decrease of intracellular hemoglobin in relation to gradual red blood cell lysis is thought to reduce magnetic nonhomogeneity and thereby increase T2 ratio [11]. In spite of markedly varied T1- and T2-ratios of the blood phantom during 14 days (changed 127 and 234%, respectively), diffusion value was little changed (19.8%). These results reveal that the diffusion value is almost unaffected by both the methemoglobin-related paramagnetic effect and the diminution of magnetic nonhomogeneity by gradual red blood cell lysis that strongly affect signal intensity on T1- and T2-weighted images. A tendency for the diffusion value to decrease slightly during the first 6 days may reflect a small increase in the volume of bound water around hemoglobin, inasmuch as the amount of free hemoglobin increase (vide the following paragraph). With increasing protein concentration, the T1 and T2 relaxation time progressively shortens [16,17]. This effect has been ascribed to slowing of the rotational motion of water molecules as they bind to and form structure in the region immediately surrounding protein molecules (i.e., become bound water) [18]. In our phantom study, T1-ratio declined almost linearly as blood concentration decreased, and the relationship of T2-ratio to blood concentration described a convex curve with the peak around 20% blood
phantom concentration. Considering that whole blood has about 20% solutes (or 18.5% protein), these T1- and T2ratio results for blood concentration are very similar to the previously reported results for protein; signal intensity on T1-weighted SE (TR/TE ⫽ 400/30) and T2-weighted SE (2000/100) images revealed convex curves with peaks around 25% and 5% protein concentration, respectively [19]. In our study, diffusion value was a nearly linear function of blood concentration, which is almost the same result as reported for protein solution [20]. The diffusion value obtained for protein solution is considered to be mainly dependent on the free water fraction, since water bound to protein has little effect on diffusion values, as seen in the following expression, D ⬵ 共1 ⫺ 兲 D f (as D b⬍⬍D f)
(3)
Here D is the self-diffusion coefficient, D b is the selfdiffusion coefficient of “bound” water, D f the self-diffusion coefficient of “free” water, and the bounded water fraction (1 ⫺ the free water fraction) [21]. Therefore, diffusion value for blood is considered to be dependent on free water fraction, just as is the diffusion value for protein solution. Paramagnetic iron derived from methemoglobin in relation to blood oxidation is mainly responsible for the signal increase on T1-weighted images, and diminution of magnetic nonhomogeneity in relation to gradual red blood cell lysis causes the signal increase on T2-weighted images [11]. However, as shown in our phantom study, diffusion value is almost independent of both blood oxidation and red blood cell lysis. Hence, we consider that the extent of blood oxidation and red blood cell lysis present in endometrial cysts are the main reasons for the lower correlation of ADC
T. Moteki et al. / Magnetic Resonance Imaging 20 (2002) 463– 470
469
Fig. 5. Endometrial cyst that was incorrectly classified to another pelvis cyst when using T1-ratio and T2-ratio, and correctly diagnosed when using ADC value. Content of the lesion was aged blood. (a) A postcontrast T1-weighted image shows a low intensity lesion with thickened capsular structure. (b) On T2-weighted image, the lesion reveals homogeneous high intensity. (c) EPI with diffusion gradients (lower) demonstrates mild signal decay in the lesion compared with EPI without diffusion gradients (upper). Calculated ADC value is 0.00148 mm2/sec.
470
T. Moteki et al. / Magnetic Resonance Imaging 20 (2002) 463– 470
value to T1- or T2-ratio in endometrial cysts. Signal intensity and diffusion value behavior in blood (both functions of blood concentration) are very similar to the reported results for protein concentration, as described in the previous paragraph; diffusion value and signal intensity on T1-weighted image are linearly related to blood or protein concentration, and the relation of signal intensity on T2-weighted image to concentration forms a convex curve with peak at around 5% protein or 20% blood phantom concentration (having 19.6% whole blood (or 3.6% protein) concentration). These results also explain why the correlation of ADC value and T1-ratio to T2-ratio in endometrial and other pelvic cysts is lower while ADC, T1-ratio and T2-ratio differences between the two groups are statistically significant. However, these cannot explain why for other pelvic cysts, the correlation of ADC value to T1-ratio is also lower. This may be explained by the presence of minimal hemorrhagic components that affect T1-ratios and/or of pseudodiffusion effect such as turbulent flow in the cystic contents by physiologic motions that affect pure diffusion values [22] because 5/12 other ovarian cysts showed high ADC values greater than the diffusion coefficient of pure water at 37°C (0.00324 mm2/ sec). In our in vivo study, ADC value showed substantially comparable ability of differentiation between endometrial and other pelvic cysts with T1- or T2-ratio, and the combination of ADC value and T1-ratio (and/or T2-ratio) could improve the differentiation between the two groups in contrast with the combination of T1- and T2-ratios. Furthermore, the cases misdiagnosed by the combination of T1and T2-ratios were different from the ones misdiagnosed by ADC value, probably because of the lower correlation of ADC value to T1-ratio and T2-ratio in endometrial and other pelvic cysts. Therefore, we think that ADC value may become a new diagnostic tool for the estimation of blood and/or protein concentration within cystic lesions because diffusion value is a nearly linear function of blood and protein concentration, almost independent of blood oxidation levels (or methemoglobin-related paramagnetic effects), and also almost independent of red blood lysis (or diminution of magnetic nonuniformity by increased free hemoglobin) that significantly affect signal intensity.
5. Conclusion Even though it was poorly correlated with signal intensity, ADC value has about the same ability to differentiate endometrial from other pelvic cysts as does signal intensity on T1- and T2-weighted images. We believe poor correlation between ADC value and signal intensity occurs because ADC value is almost linearly dependent on blood concentration and almost independent of blood oxidation and red blood cell lysis, two factors that significantly affect signal intensity.
References [1] Le Bihan D, Breston E, Lallemand D, Grenier P, Cabanis EA, LavalJeantet M. MR imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurologic disorders. Radiology 1986; 161:401–7. [2] Ahn CB, Lee SY, Nalcioglu O, Cho ZH. The effects of random directional distributed flow in nuclear magnetic resonance imaging. Med Phys 1987;14:43– 8. [3] Le Bihan D, Breton E, Lallemand D, Aubin ML, Vignaud J, Lavaljeantet M. Separation of diffusion and perfusion in intravoxel incoherent motion (IVIM) MR imaging. Radiology 1988;168:497–505. [4] Young IR, Hall AS, Bryant DJ, Tomas DG, Gill SS, Dubowitz LM, Cowan G, Pennock JM, Bydder GM. Assessment of brain perfusion with MR imaging. J Comput Assist Tomogr 1988;12:721–7. [5] Moonen CTW, Van Zijl PCM, Le Bihan D, DesPres D. In vivo NMR diffusion spectroscopy. 31P application to phosphorus metabolites in muscle. Magn Reson Med 1990;13:467–77. [6] Moseley ME, Kucharczyk J, Mintorovitch J, Cohen Y, Kurhanewicz J, Derugin N, Asgari H, Norman D. Diffusion-weighted MR imaging of acute stroke: correlation with T2-weighted and magnetic susceptibility-enhanced MR imaging in cats. AJNR 1990;11:423–9. [7] Warach S, Chien D, Li W, Ronthal M, Edelman RR. Fast magnetic resonance diffusion-weighted imaging of acute human stroke. Neurology 1992;42:1717–23. [8] Namimoto T, Yamashita Y, Sumi S, Tang Y, Takahashi M. Focal liver masses: characterization with diffusion-weighted echo-planar MR imaging. Radiology 1997;204:739 – 44. [9] Namimoto T, Yamashita Y, Mitsuzaki K, Nakayama Y, Tang Y, Takahashi M. Measurement of the apparent diffusion coefficient in diffuse renal disease by diffusion-weighted echo-planar MR imaging. J Magn Reson Imaging 1999;9:832–7. [10] Sugimura K, Takemori M, Sugiura M, Okizuka H, Kano M, Ishida T. The value of magnetic resonance relaxation time in staging ovarian endometrial cysts. Br J Radiol 1992;65:502– 6. [11] Janick PA, Hackney DB, Grossman RI, Asakura T. MR imaging of various oxidation states of intracellular and extracellular hemoglobin. AJNR 1991;12:891–7. [12] Kuma F. Properties of methemoglobin reductase and kinetic study of methemoglonin reduction. J Biol Chem 1981;256:5518 –23. [13] Nishimura K, Togashi K, Itoh K, Fujisawa I, Noma S, Kawamura Y, Nakano Y, Itoh H, Torizuka K, Ozawa H. Endometrial cysts of the ovary: MR imaging. Radiology 1987;162:315– 8. [14] Zawin M, McCarthy S, Scoutt L, Comite F. Endometriosis: appearance and detection at MR imaging. Radiology 1989;171:693– 6. [15] William G, Bradley WG. Hemorrhage and brain iron. In: Stark DD, Bradley WG, editors. Magnetic Resonance Imaging, 2nd ed. St. Louis: Mosby Year Book, 1992, pp. 721– 69. [16] Koenig SH, Schillinger WE. Nuclear magnetic relaxation dispersion in protein solution. J Biol Chem 1969;244:6520 – 6. [17] Fullerton GD, Finnie MF, Hunter KE, Ord VA, Cameron IL. The influence of macromolecular polymerization on spine-lattice relaxation of aqueous solution. Magn Reson Imaging 1987;5:353–70. [18] Bloembergen N, Purcell EM, Pound RV. Relaxation effects in nuclear magnetic resonance absorption. Phys Rev 1948;73:679 –712. [19] Som PM, Dillon WP, Fullerton GD, Zimmerman RA, Rajagopalan B, Marom Z. Chronically obstructed sinonasal secretions: observations on T1 and T2 shortening. Radiology 1989;172:515–20. [20] Baranowska HM, Olszewski KJ. The hydration of proteins in solutions by self-diffusion coefficients. NMR study. Biochim Biophys Acta 1996;1289:312– 4. [21] Blinc A, Lahajnar G, Blinc R, Zidansek A, Sepe A. Proton NMR study of the state of water in fibrin gels, plasma, and blood clots. Magn Reson Med 1990;14:105–22. [22] Edelman RR, Wielopolski P, Schmitt F. Echo-planar MR imaging. Radiology 1994;192:600 –12.