Original Research

T2-Weighted MRI of Rectosigmoid Carcinoma:Comparison of Respiratory-Triggered Fast Spin-Echo, Breathhold Fast-Recovery Fast Spin-Echo,and Breathhold Single-Shot Fast Spin-EchoSequences

Shuhei Yamashita, MD,1,2* Takayuki Masui, MD,1 Motoyuki Katayama, MD,1

Kimihiko Sato, MD,1 Nobuko Yoshizawa, MD,2 Hidekazu Seo, MD,1 andHarumi Sakahara, MD2

Purpose: To compare the abilities of T2-weighted (T2W)imaging using respiratory-triggered fast spin-echo (RT-FSE), breathhold fast-recovery FSE (BH-FRFSE), and BHsingle-shot FSE (BH-SSFSE) sequences without an endo-rectal coil to detect rectosigmoid carcinomas.

Materials and Methods: Forty patients (stage: pT0, 1;pTis-2, 15; pT3-4, 24) were included in the study. All ex-aminations were performed on a 1.5T magnet with aphased-array coil and the patients were studied in theprone position with per-anal air injection. Qualitative andquantitative evaluations were performed.

Results: Motion artifact was the most prominent with theRT-FSE sequence, and the least prominent with the BH-SSFSE sequence. Scores for depiction of the rectal walllayer, tumor recognition, and overall image quality were thehighest with the BH-FRFSE sequence. On the basis of areceiver operating characteristic (ROC) analysis, the detec-tion rate of tumor invasion through the rectal wall washigher with the BH-FRFSE sequence (Az � 0.9077) thanwith the RT-FSE (Az � 0.7762, p � 0.05) or BH-SSFSE(Az � 0.8602) sequence. Tumor-to-fat contrast was highestwith the BH-FRFSE sequence (P � 0.017).

Conclusion: The BH-FRFSE sequence may be the firstchoice for rectosigmoid T2W imaging in the prone positionwith per-anal air injection for patients who can hold theirbreath stably.

Key Words: rectosigmoid carcinoma; magnetic resonanceimaging; comparative study; pulse sequences; T2-weightedimagingJ. Magn. Reson. Imaging 2007;25:511–516.© 2007 Wiley-Liss, Inc.

MAGNETIC RESONANCE IMAGING (MRI) is one of theimportant modalities for diagnosing rectosigmoid car-cinoma (1–7). T2-weighted (T2W) images can demon-strate the rectal wall layers, such as the submucosallayer and muscularis propria (8), Thus, tumor penetra-tion and invasion of bowel wall layers can be evaluatedwell with T2W images.

MRI with the use of an endorectal coil can demon-strate rectal layers in detail, and is reported to be usefulfor tumor staging (9–14). Because insertion of the coilinto the rectum is an invasive procedure, a less invasiveway to perform MRI is desirable. MRI using a phased-array coil, which is less invasive than an endorectalcoil, can be performed for patients with a suspiciousrectosigmoid carcinoma. To evaluate the rectosigmoidin detail, its extension by inflation with room air may berequired, and to keep the air in the rectum, the patientshave to maintain a prone position. When we performMRI in this way, motion artifacts caused by respirationand peristalsis are problematic, although the rectum isa relatively fixed part of the colon. The turbo spin-echo(TSE) (or fast spin-echo (FSE)) sequence is now widelyused for evaluation of the abdomen and pelvis, and acombination with a respiratory-triggering (RT) tech-nique may be effective for reducing motion artifacts(15–17).

Another way to reduce respiratory motion artifacts isto use a breathholding technique. By using a longerecho train length (ETL) and a shorter interecho spacewithout reduction of spatial resolution, FSE T2W im-ages with a shorter repetition time (TR) within 2000msec using an ETL of about 16 can be obtained during

1Department of Radiology, Seirei Hamamatsu General Hospital, Shi-zuoka, Japan.2Department of Radiology, Hamamatsu University School of Medicine,Shizuoka, Japan.Presented in part at the 12th Annual Meeting of ISMRM, Kyoto, Japan,2004.*Address reprint requests to: S.Y., Department of Radiology, HamamatsuUniversity School of Medicine, 1-20-1 Handayama, Hamamatsu, Shi-zuoka, 431-3192 Japan. E-mail: shuheiy@hama-med.ac.jpReceived October 5, 2005; Accepted September 27, 2006.DOI 10.1002/jmri.20827Published online in Wiley InterScience (www.interscience.wiley.com).

JOURNAL OF MAGNETIC RESONANCE IMAGING 25:511–516 (2007)

© 2007 Wiley-Liss, Inc. 511

a single breathhold (BH) of less than 30 seconds. How-ever, T2W images obtained with an FSE sequence withshorter TR may suffer from lower SIs due to protonsaturation (18). A fast-recovery FSE (FRFSE) sequencehas sequential 180° and –90° radiofrequency (RF)pulses that are applied at the end of the FSE sequenceto recover saturated spins (19). By using this method,even with shorter TRs (within 2000 msec), BH-T2Wimages with an acceptable signal-to-noise ratio (SNR)can be obtained. Half-Fourier single-shot (SS) rapid ac-quisition techniques (i.e., the half-Fourier acquisitionSSTSE or SSFSE sequence) can also be utilized forBH-T2W imaging (20).

The purpose of this prospective study was to evaluateand compare the abilities of T2W imaging using RT-FSE, BH-FRFSE, and BH-SSFSE sequences for the de-tection of rectosigmoid carcinomas.

MATERIALS AND METHODS

Patient Selection

Between August 2000 and September 2001, 55 consec-utive patients who were suspected of having rectosig-moid carcinoma underwent pelvic MRI in our hospitalbefore surgery. Fifteen patients were excluded for thefollowing reasons: surgery was not performed in fivepatients, lesions were not rectosigmoid carcinoma insix patients (carcinoid tumor in two, teratoma in one,endometriosis in one, peritoneal dissemination of ovar-ian carcinoma in one, and abscess in one), and theimaging protocol was incomplete in four patients. Theremaining 40 patients (30 men and 10 women, 44–89years old, mean age � 62.1 years) were included in thisstudy. The postoperative pathologic stages were as fol-lows: pT0: 1 (operation was performed after endoscopicmucosal resection, and no tumor remained), pTis: 1,pT1: 5, pT2: 9, and pT3: 24 (21).

MRI Technique

All MR examinations were performed on a 1.5T super-conducting magnet (Signa Horizon LX Echospeed; Gen-eral Electric Medical Systems, Milwaukee, WI). The sys-tem provides a maximum gradient strength of 23 mT/mwith a peak slew rate of 120 mT/m/msec. A pelvicphased-array coil was utilized (a torso phased-arraycoil was used in one patient). An antiperistaltic drug(7.5 mg of timepidium bromide; Sesden; Tanabe Seiy-aku Co., Ltd., Osaka, Japan) was administered to 25patients (19 men and six women) who were less than 70years old and had no contraindications. All patientswere studied in the prone position with per-anal injec-tion of approximately 400–600 mL of room air. Threetypes of T2W images with 5-mm thickness and 1-mminterslice gap were obtained in the correspondingplane. The imaging parameters for the T2W imagingwere as follows: 1) RT-FSE sequence: TR � 4000–7000msec, effective TE (eTE) � 85 msec, ETL � 16, receiverbandwidth (RBW) � 62.5 kHz, matrix � 256 � 256, fieldof view (FOV) � 20 � 20–24 � 24 cm, number of exci-tations (NEX) � 2, respiratory trigger point � 20–30%,acquisition time � 2–4 minutes; 2) BH-FRFSE se-quence: TR � 2000 msec, eTE � 85 msec, ETL � 17,

RBW � 31.2 kHz, matrix � 256 � 160, FOV � 20 �20–24 � 24 cm, NEX � 1, acquisition time � 24 sec-onds; 3) BH-SSFSE sequence: TR � �, eTE � 96 msec,RBW � 62.5 kHz, matrix � 256 � 192, FOV � 20 �18–24 � 24 cm, NEX � 0.5, acquisition time � 24seconds.

Evaluations

All MR images were stored in a workstation (AdvantageWorkstation 4.0; General Electric Medical Systems,Milwaukee, WI, USA). The images were displayed on a21-inch display monitor and evaluations were per-formed.

Image Quality

All MR images were randomly displayed and indepen-dently reviewed by two radiologists without awarenessof the patient’s history, clinical or other radiologicalfindings, name, age, or imaging parameters. Motionartifact and blurring were graded on a five-point scale(1 � unacceptable; 2 � severe; 3 � moderate; 4 � mild;and 5 � absent). Overall quality, depiction of submu-cosa, depiction of muscularis propria, border betweentumor and rectal lumen, border between tumor andnormal rectal wall, and tumor recognition were alsograded on a five-point scale (1 � unacceptable; 2 �poor; 3 � fair; 4 � good; and 5 � excellent). The resultswere statistically analyzed by means of a Wilcoxonsigned-rank test with Bonferroni correction.

Tumor Staging

The existence or absence of penetration through therectal wall by the tumor (T3 or T4) was evaluated by tworadiologists on a five-point scale (1 � definitely absent;2 � probably absent; 3 � possibly present; 4 � probablypresent; and 5 � definitely present). Then, receiver op-erating characteristic (ROC) analyses were performed.For each imaging sequence, a binominal ROC curvewas fitted to each interpreter’s confidence rating byusing a maximum-likelihood estimation from the ROC-KIT 0.9.1 B program (Metz C, University of Chicago,Chicago, IL, USA) (22). The observer performance witheach imaging sequence for each interpreter was esti-mated by calculating the area under the ROC curve(Az). Differences between the ROC curves of individualinterpreters and pooled data of the two interpreterswere tested by using the area test with a univariatez-score test of the difference between the areas underthe two ROC curves.

Quantitative Evaluations

Operator-defined regions of interest (ROIs) were placedon the corresponding sites of the tumor, skeletal mus-cle, and fat with three types of T2W images. The signalintensities (SIs) of these regions were measured. Thenthe tumor-to-muscle contrast (| SIs of the tumor – SIsof the skeletal muscle |/SIs of the skeletal muscle) andtumor-to-fat contrast (| SIs of the tumor – SIs of the fat|/SIs of the fat) were calculated. The results were ana-

512 Yamashita et al.

lyzed by means of a paired t-test with Bonferroni cor-rection.

RESULTS

Image Quality

The results of the evaluations of image quality are sum-marized in Table 1. Motion artifact was most prominentwith the RT-FSE sequence, and the least prominentwith the BH-SSFSE sequence among three sequences.In eight patients (observer 1: four patients; observer 2:seven patients), poor image quality caused by motionartifact (graded as unacceptable or severe on the five-point scale) with the RT-FSE sequence was noted (Fig.1). With the BH-FRFSE sequence, prominent motionartifacts were caused by poor breathholding in two pa-tients (observer 1: one patient; observer 2: two patients;Fig. 2). Motion artifact with the antiperistaltic drug wasRT-FSE:BH-FRFSE:BH-SSFSE � 3.78:4.20:4.80, andwithout using the antiperistaltic drug it was RT-FSE:BH-FRFSE:BH-SSFSE � 3.10:4.27:4.93. Blurring arti-facts were more severe with the BH-SSFSE sequencethan with the other two sequences. Overall image qual-ity with the BH-FRFSE sequence was better than thatwith other two sequences. Overall image quality usingthe antiperistaltic drug was RT-FSE:BH-FRFSE:BH-

SSFSE � 4.02:4.48:3.58 and without using the drug itwas RT-FSE:BH-FRFSE:BH-SSFSE � 3.23:4.33:3.63.Depiction of the layers of the rectal wall (submucosaand muscularis propria) was better with the BH-FRFSEsequence than with the other two sequences (Fig. 3).Depiction of the border of the tumor and tumor recog-nition were better with the BH-FRFSE sequence thanwith the other two sequences.

Tumor Staging

The results of the ROC analyses in the evaluation of theexistence or absence of penetration through the rectalwall by the tumor are summarized in Table 2. With thepooled data of the two observers, the Az value for thecomposite ROC curves with the BH-FRFSE sequence(Az � 0.9077) was significantly higher compared to theRT-FSE sequence (Az � 0.7762, P � 0.05).

Quantitative Analysis

The results of the quantitative evaluations are summa-rized in Table 3. Although there were no statisticallysignificant differences, tumor-to-muscle contrast withthe BH-FRFSE sequence was lower than with the othertwo sequences. Tumor-to-fat contrast with BH-FRFSEsequence was significantly higher than with the other

Table 1Qualitative Evaluations of the Artifacts, Image Quality and Depiction of Tumor with Three MR Pulse Sequences*

RT-FSE BH-FRFSE BH-SSFSE

Motion artifact 3.53 � 1.01a,b 4.23 � 0.76c 4.85 � 0.36Blurring artifact 4.49 � 0.55a,b 4.79 � 0.41c 3.24 � 0.72Overall quality 3.73 � 1.12a 4.43 � 0.69c 3.60 � 0.65Depiction of submucosa 2.39 � 1.26b 2.68 � 1.14c 1.76 � 0.86Depiction of muscularis propria 3.05 � 1.28b 3.51 � 1.03c 2.40 � 0.96Border between tumor and rectal lumen 3.74 � 1.19a 4.18 � 0.94c 3.55 � 1.05Border between tumor and normal rectal wall 3.56 � 1.13a 4.10 � 0.89c 3.46 � 0.88Tumor recognition 3.99 � 1.17a 4.45 � 0.85c 3.91 � 1.07

*All numbers refer to mean ranks � SD.aSignificant difference between RT-FSE and BH-FRFSE (P � 0.017 using Wilcoxon signed rank test with Bonferroni correction).bSignificant difference between RT-FSE and BH-SSFSE, (P � 0.017 using Wilcoxon signed rank test with Bonferroni correction).cSignificant difference between BH-FRFSE and BH-SSFSE (P � 0.017 using Wilcoxon signed rank test with Bonferroni correction).RT-FSE � respiratory-triggered fast spin-echo sequence, BH-FRFSE � breathhold fast-recovery fast spin-echo sequence, BH-SSFSE �breathhold single-shot fast spin-echo sequence.

Figure 1. A 66-year-old man with rectal carcinoma (T3, long arrows). a: RT-FSE (TR/eTE � 7059/87 msec, acquisition time �3 minutes 46 seconds for 24 sections). b: BH-FRFSE (TR/eTE 2000/84 � msec, acquisition time � 24 seconds for 12 sections).c: BH-SSFSE (TR/eTE � �/95.5 msec, acquisition time � 20 seconds for 12 sections) images demonstrate the thickened rectalwall. Prominent motion artifact was noted with the (a) RT-FSE sequence (short arrows). Blurring artifact was severe with the (c)BH-SSFSE sequence. Depiction of the layers of the rectal wall was better with the (b) BH-FRFSE sequence than with the othertwo sequences (short arrows). This tumor was evaluated as penetrating through the rectal wall because it extends beyond thecontour of the rectum.

MRI of Rectosigmoid Carcinoma 513

two sequences. The RT-FSE and BH-SSFSE sequenceswere equivalent in terms of tumor-to-muscle contrastand tumor-to-fat contrast.

DISCUSSION

In this study, T2W imaging with the BH-FRFSE se-quence had higher scores compared to the other twosequences with respect to overall image quality, recog-nition of tumor, and depiction of rectal wall layers.Although motion artifact was the least prominent withthe BH-SSFSE sequence, poor image quality was ob-served because of blurring artifact, which is thought tobe caused by T2-filtering effects (23–25).

In this study, we could not place an ROI in the rectalwall because the normal rectal wall is a very thin struc-ture. Thus we were unable to calculate the SNR of thecolon or the CNR between the lesion and normal rectalwall. In other organs, such as the uterus (19) and liver(26), it has been reported that the SNR and CNR of T2Wimages with the FRFSE sequence are lower compared tothe FSE sequence.

However, in the present study, T2W imaging withBH-FRFSE provided better image quality. This may bebecause the SI of the lesion with the BH-FRFSE se-quence was lower than with the other sequence be-

cause of its lower SNR, which resulted in high contrastbetween the lesion and the perirectal adipose tissue.Theoretically, contrast of the rectal layer is thought tobe highest with the RT-FSE sequence. This advantagemay be offset by image degradation caused by respira-tory motion, which is exaggerated by the prone positioncompared to the supine position.

There were almost no motion artifacts with the BH-SSFSE sequence; however, image quality was insuffi-cient to depict the rectal wall layer. This may be due tothe blurring effect or lower SI because of the half NEXused. It may be possible to reduce the blurring effect byshortening the ETL using parallel imaging (27).

We used an antiperistaltic drug for the patients whohad no contraindications. Our results demonstrate thatthe antiperistaltic drug had an obvious influence on theimage quality of RT-FSE due to the reduction of motionartifact. In the BH-FRFSE and BH-SSFSE sequences,peristaltic motion artifact was minimized because oftheir shorter scanning times.

Shortened imaging times may also be advantageousfor decreasing motion artifacts. Steady-state free pre-cession (SSFP) imaging (28) using techniques such asTrueFISP, FIESTA, and balanced FFE may be used toevaluate rectosigmoid carcinomas. However, image

Figure 2. A 51-year-old man with rectal carcinoma (T3). a: RT-FSE (TR/eTE � 7826/87.3 msec, acquisition time � 4 minutes11 seconds for 20 sections). b: BH-FRFSE (TR/eTE � 2000/85.8 msec, acquisition time � 24 seconds for 10 sections). c:BH-SSFSE (TR/eTE � �/95.5 msec, acquisition time � 15 seconds for 10 sections) images demonstrate the rectal tumorpredominantly in the posterior wall. In this patient, although there was no motion artifact due to peristalsis, motion artifact ofthe abdominal wall due to incomplete breathholding was prominent with the (b) BH-FRFSE sequence. With the (c) BH-SSFSEsequence, there were no motion artifacts.

Figure 3. A 56-year-old man with rectal carcinoma (T2, arrows). a: Respiratory-triggered FSE (TR/eTE � 7059/83.2 msec,acquisition time � 3 minutes 53 seconds for 20 sections). b: BH-FRFSE (TR/eTE � 2000/83.0 msec, acquisition time � 24seconds for 12 sections). c: BH-SS-FSE (TR/eTE � �/97.6 msec, acquisition time � 20 seconds for 12 sections) imagesdemonstrate the nodular rectal tumor in the right posterior wall. Motion artifact was noted with the (a) RT-FSE sequence.Blurring artifact was noted with the (c) BH-SSFSE sequence. Depiction of the layers of the rectal wall was better with the (b)BH-FRFSE sequence than with the other two sequences. We concluded there were no penetration through the rectal wallbecause the low SI of the muscularis propria was not interrupted.

514 Yamashita et al.

contrasts of soft tissues are characterized by T2/T1ratios. Thus, further investigation is required.

This study has some limitations. We were unable tocalculate the SNR and CNR because of the difficulty ofplacing an ROI in the rectal wall. When we comparedthe techniques in terms of the diagnostic value of tumorstaging, we evaluated in only one direction. In clinicalpractice, we diagnose tumor staging in multiple direc-tions. However, in this study, we acquired three types ofT2W images in one direction and compared these im-ages. To investigate the diagnostic value of each se-quence precisely, evaluation of images in multiple di-rections would be desirable. In this study we used thepooled data of two interpreters for the ROC analysesbecause of the insufficient number of patients. The pos-sibility that statistically significant overstaging oc-curred cannot be excluded.

MRI of rectosigmoid carcinomas using an endorectalcoil is advantageous (9–14) because it offers high spa-tial resolution resulting in detailed visualization of le-sions. With this technique, layers of the rectal wall,such as the mucosal layer, can be depicted. However,because MRI with an endorectal coil has a limited FOV,it is difficult to evaluate the mesorectum or surround-ing structures (29). The endorectal coil also causes pa-tients discomfort or pain. Considering these limitationsof endorectal coil, we believe that MRI with a phased-array coil is still useful.

In conclusion, for T2W imaging with a phased-arraycoil in the prone position with per-anal air injection, theBH-FRFSE sequence may be useful for depictingrectosigmoid carcinomas and evaluating local invasionbecause it produces less motion artifacts and betterdepiction of the muscularis propria compared to the

other two sequences. RT-FSE sequence is potentially auseful method for patients who can breathe regularly.For patients who cannot hold their breath stably, theBH-SSFSE sequence can be utilized.

REFERENCES

1. Guinet C, Buy JN, Ghossain MA, et al. Comparison of magneticresonance imaging and computed tomography in the preoperativestaging of rectal cancer. Arch Surg 1990;125:385–388.

2. McNicholas MM, Joyce WP, Dolan J, Gibney RG, MacErlaine DP,Hyland J. Magnetic resonance imaging of rectal carcinoma: a pro-spective study. Br J Surg 1994;81:911–914.

3. Hadfield MB, Nicholson AA, MacDonald AW, et al. Preoperativestaging of rectal carcinoma by magnetic resonance imaging with apelvic phased-array coil. Br J Surg 1997;84:529–531.

4. Brown G, Richards CJ, Newcombe RG, et al. Rectal carcinoma:thin-section MR imaging for staging in 28 patients. Radiology 1999;211:215–222.

5. Blomqvist L, Machado M, Rubio C, et al. Rectal tumour staging: MRimaging using pelvic phased-array and endorectal coils vs endo-scopic ultrasonography. Eur Radiol 2000;10:653–660.

6. Laghi A, Ferri M, Catalano C, et al. Local staging of rectal cancerwith MRI using a phased array body coil. Abdom Imaging 2002;27:425–431.

7. Low RN, McCue M, Barone R, Saleh F, Song T. MR staging ofprimary colorectal carcinoma: comparison with surgical and his-topathologic findings. Abdom Imaging 2003;28:784–793.

8. Elmas N, Killi RM, Sever A. Colorectal carcinoma: radiological di-agnosis and staging. Eur J Radiol 2002;42:206–223.

9. Chan TW, Kressel HY, Milestone B, et al. Rectal carcinoma: stagingat MR imaging with endorectal surface coil. Work in progress.Radiology 1991;181:461–467.

10. Schnall MD, Furth EE, Rosato EF, Kressel HY. Rectal tumor stage:correlation of endorectal MR imaging and pathologic findings. Ra-diology 1994;190:709–714.

11. Joosten FB, Jansen JB, Joosten HJ, Rosenbusch G. Staging ofrectal carcinoma using MR double surface coil, MR endorectal coil,and intrarectal ultrasound: correlation with histopathologic find-ings. J Comput Assist Tomogr 1995;19:752–758.

12. Drew PJ, Farouk R, Turnbull LW, Ward SC, Hartley JE, MonsonJR. Preoperative magnetic resonance staging of rectal cancer withan endorectal coil and dynamic gadolinium enhancement. Br JSurg 1999;86:250–254.

13. Gualdi GF, Casciani E, Guadalaxara A, d’Orta C, Polettini E, Pap-palardo G. Local staging of rectal cancer with transrectal ultra-sound and endorectal magnetic resonance imaging: comparisonwith histologic findings. Dis Colon Rectum 2000;43:338–345.

14. Maldjian C, Smith R, Kilger A, Schnall M, Ginsberg G, Kochman M.Endorectal surface coil MR imaging as a staging technique forrectal carcinoma: a comparison study to rectal endosonography.Abdom Imaging 2000;25:75–80.

15. Low RN, Francis IR, Sigeti JS, Foo TK. Abdominal MR imaging:comparison of T2-weighted fast and conventional spin-echo, andcontrast-enhanced fast multiplanar spoiled gradient-recalled im-aging. Radiology 1993;186:803–811.

16. Catasca JV, Mirowitz SA. T2-weighted MR imaging of the abdomen:fast spin-echo vs conventional spin-echo sequences. AJR Am JRoentgenol 1994;162:61–67.

17. Low RN, Alzate GD, Shimakawa A. Motion suppression in MRimaging of the liver: comparison of respiratory-triggered and non-triggered fast spin-echo sequences. AJR Am J Roentgenol 1997;168:225–231.

18. Yamashita Y, Yamamoto H, Namimoto T, Abe Y, Takahashi M.Phased array breath-hold versus non-breath-hold MR imaging offocal liver lesions: a prospective comparative study. J Magn ResonImaging 1997;7:292–297.

19. Masui T, Katayama M, Kobayashi S, Sakahara H, Ito T, Nozaki A.T2-weighted MRI of the female pelvis: comparison of breath-holdfast-recovery fast spin-echo and nonbreath-hold fast spin-echosequences. J Magn Reson Imaging 2001;13:930–937.

20. Chung JJ, Semelka RC, Martin DR, Marcos HB. Colon diseases:MR evaluation using combined T2-weighted single-shot echo trainspin-echo and gadolinium-enhanced spoiled gradient-echo se-quences. J Magn Reson Imaging 2000;12:297–305.

Table 2Az Values Obtained with Three MR Pulse Sequences

RT-FSE BH-FRFSE BH-SSFSE

Observer 1 0.7804 0.9078 0.8850Observer 2 0.7705 0.9140 0.8438Pooled 0.7762a 0.9077 0.8602

aSignificant difference between RT-FSE and BH-FRFSE (P � 0.05using univariate z-score test).RT-FSE � respiratory-triggered fast spin-echo sequence, BH-FRFSE � breathhold fast-recovery fast spin-echo sequence, BH-SSFSE � breathhold single-shot fast spin-echo sequence.

Table 3Quantitative Evaluations With Three MR Pulse Sequences*

RT-FSE BH-FRFSE BH-SSFSE

Tumor-to-musclecontrast 0.45 � 0.63 0.35 � 0.53 0.46 � 0.50

Tumor-to-fat contrast 0.48 � 0.13a 0.58 � 0.08b 0.51 � 0.13

*All numbers refer to mean ranks � SD.aSignificant difference between RT-FSE and BH-FRFSE (P � 0.017using paired t-test with Bonferroni correction).bSignificant difference between BH-FRFSE and BH-SSFSE (P �0.017 using paired t-test with Bonferroni correction).RT-FSE � respiratory-triggered fast spin-echo sequence, BH-FRFSE � breathhold fast-recovery fast spin-echo sequence, BH-SSFSE � breathhold single-shot fast spin-echo sequence.

MRI of Rectosigmoid Carcinoma 515

21. Sobin LH, Wittekind CH. TNM classification of malignant tumours.6th ed. New York: Wiley; 2002. p 72–76.

22. Metz CE. Some practical issues of experimental design and dataanalysis in radiological ROC studies. Invest Radiol 1989;24:234–245.

23. Coates GG, Borrello JA, McFarland EG, Mirowitz SA, Brown JJ. He-patic T2-weighted MRI: a prospective comparison of sequences,including breath-hold, half-Fourier turbo spin echo (HASTE). JMagn Reson Imaging 1998;8:642–649.

24. Constable RT, Gore JC. The loss of small objects in variable TEimaging: implications for FSE, RARE, and EPI. Magn Reson Med1992;28:9–24.

25. Constable RT, Anderson AW, Zhong J, Gore JC. Factors influencingcontrast in fast spin-echo MR imaging. Magn Reson Imaging 1992;10:497–511.

26. Katayama M, Masui T, Kobayashi S, et al. Fat-suppressed T2-weighted MRI of the liver: comparison of respiratory-triggered fastspin-echo, breath-hold single-shot fast spin-echo, and breath-holdfast-recovery fast spin-echo sequences. J Magn Reson Imaging2001;14:439–449.

27. Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. SENSE:sensitivity encoding for fast MRI. Magn Reson Med 1999;42:952–962.

28. Haacke EM, Wielopolski PA, Tkach JA, Modic MT. Steady-state freeprecession imaging in the presence of motion: application for im-proved visualization of the cerebrospinal fluid. Radiology 1990;175:545–552.

29. Pijl ME, Chaoui AS, Wahl RL, van Oostayen JA. Radiology of colo-rectal cancer. Eur J Cancer 2002;38:887–898.

516 Yamashita et al.