Signal-to-noise ratio and section thickness in two-dimensional versus three-dimensional Fourier transform MR imaging

Usefulness of pituitary high-resolution 3D MRI with deep-learning-based reconstruction for perioperative evaluation of pituitary adenomas

PURPOSE

To evaluate the diagnostic value of T1-weighted 3D fast spin-echo sequence (CUBE) with deep learning-based reconstruction (DLR) for depiction of pituitary adenoma and parasellar regions on contrast-enhanced MRI.

METHODS

We evaluated 24 patients with pituitary adenoma or residual tumor using CUBE with and without DLR, 1-mm slice thickness 2D T1WI (1-mm 2D T1WI) with DLR, and 3D spoiled gradient echo sequence (SPGR) as contrast-enhanced MRI. Depiction scores of pituitary adenoma and parasellar regions were assigned by two neuroradiologists, and contrast-to-noise ratio (CNR) was calculated.

RESULTS

CUBE with DLR showed significantly higher scores for depicting pituitary adenoma or residual tumor compared to CUBE without DLR, 1-mm 2D T1WI with DLR, and SPGR (p < 0.01). The depiction score for delineation of the boundary between adenoma and the cavernous sinus was higher for CUBE with DLR than for 1-mm 2D T1WI with DLR (p = 0.01), but the difference was not significant when compared to SPGR (p = 0.20). CUBE with DLR had better interobserver agreement for evaluating adenomas than 1-mm 2D T1WI with DLR (Kappa values, 0.75 vs. 0.41). The CNR of the adenoma to the brain parenchyma increased to a ratio of 3.6 (obtained by dividing 13.7, CNR of CUBE with DLR, by 3.8, that without DLR, p < 0.01). CUBE with DLR had a significantly higher CNR than SPGR, but not 1-mm 2D T1WI with DLR.

CONCLUSION

On the contrast-enhanced MRI, compared to CUBE without DLR, 1-mm 2D T1WI with DLR and SPGR, CUBE with DLR improves the depiction of pituitary adenoma and parasellar regions.

Three-Tesla imaging of the pituitary and parasellar region: T1-weighted 3-dimensional fast spin echo cube outperforms conventional 2-dimensional magnetic resonance imaging

OBJECTIVE

We explored how a novel T1-weighted 3-dimensional (3D) fast spin echo (FSE) sequence (Cube; GE, Waukesha, Wis) might outperform conventional 2-dimensional (2D) FSE techniques for contrast-enhanced imaging of the pituitary and parasellar region.

METHODS

Ninety-one patients were imaged with 3D Cube and conventional 2D FSE on a 3.0-T magnetic resonance scanner. Two neuroradiologists independently assessed images for anatomical delineation (infundibulum, optic apparatus, and cavernous sinus), degree of artifact, and confidence in lesion definition or exclusion using a 5-point scale. In addition, the readers were asked to rank overall preference.

RESULTS

Readers A and B found 3D Cube to be better or equal to 2D FSE in 84% and 86% of the cases. Three-dimensional Cube provided significantly better images than 2D FSE with respect to delineation of the infundibulum (P < 0.0001), cavernous sinus (P < 0.0001), optic apparatus (P = 0.002 for reader A and P = 0.265 for reader B), and fewer artifacts at the sellar floor (P < 0.0001). Three-dimensional Cube provided greater lesion conspicuity or confidence in lesion exclusion (P < 0.0001).

CONCLUSIONS

Three-dimensional Cube provides superior quality with thinner slices as well as diminished artifact and can replace conventional 2D FSE sequences for routine evaluations of the pituitary and parasellar region.

Qualitative and quantitative assessment of wrist MRI at 3.0T: comparison between isotropic 3D turbo spin echo and isotropic 3D fast field echo and 2D turbo spin echo

BACKGROUND

Isotropic three-dimensional (3D) magnetic resonance imaging (MRI) has been applied to various joints. However, comparison for image quality between isotropic 3D MRI and two-dimensional (2D) turbo spin echo (TSE) sequence of the wrist at a 3T MR system has not been investigated.

PURPOSE

To compare the image quality of isotropic 3D MRI including TSE intermediate-weighted (VISTA) sequence and fast field echo (FFE) sequence with 2D TSE intermediate-weighted sequence of the wrist joint at 3.0 T.

MATERIAL AND METHODS

MRI was performed in 10 wrists of 10 healthy volunteers with isotropic 3D sequences (VISTA and FFE) and 2D TSE intermediate-weighted sequences at 3.0 T. The signal-to-noise ratio (SNR) was obtained by imaging phantom and noise-only image. Contrast ratios (CRs) were calculated between fluid and cartilage, triangular fibrocartilage complex (TFCC), and the scapholunate ligament. Two radiologists independently assessed the visibility of TFCC, carpal ligaments, cartilage, tendons and nerves with a four-point grading scale. Statistical analysis to compare CRs (one way ANOVA with a Tukey test) and grades of visibility (Kruskal-Wallis test) between three sequences and those for inter-observer agreement (kappa analysis) were performed.

RESULTS

The SNR of 2D TSE (46.26) was higher than those of VISTA (23.34) and 3D FFE (19.41). CRs were superior in 2D TSE than VISTA (P = 0.02) for fluid-cartilage and in 2D TSE than 3D FFE (P < 0.01) for fluid-TFCC. The visibility was best in 2D TSE (P < 0.01) for TFCC and in VISTA (P = 0.01) for scapholunate ligament. The visibility was better in 2D TSE and 3D FFE (P = 0.04) for cartilage and in VISTA than 3D FFE (P < 0.01) for TFCC. The inter-observer agreement for the visibility of anatomic structures was moderate or substantial.

CONCLUSION

Image quality of 2D TSE was superior to isotropic 3D MR imaging for cartilage, and TFCC. 3D FFE has better visibility for cartilage than VISTA and VISTA has superior visibility for TFCC to 3D FFE and the visibility for scapholunate ligament was best on VISTA.

Qualitative and quantitative assessment of isotropic ankle magnetic resonance imaging: three-dimensional isotropic intermediate-weighted turbo spin echo versus three-dimensional isotropic fast field echo sequences

Objective

To compare the image quality of volume isotropic turbo spin echo acquisition (VISTA) imaging method with that of the three-dimensional (3D) isotropic fast field echo (FFE) imaging method applied for ankle joint imaging.

Materials and Methods

MR imaging of the ankles of 10 healthy volunteers was performed with VISTA and 3D FFE sequences by using a 3.0 T machine. Two radiologists retrospectively assessed the tissue contrast between fluid and cartilage (F-C), and fluid and the Achilles tendon (F-T) with use of a 4-point scale. For a quantitative analysis, signal-to-noise ratio (SNR) was obtained by imaging phantom, and the contrast ratios (CRs) were calculated between F-T and F-C. Statistical analyses for differences in grades of tissue contrast and CRs were performed.

Results

VISTA had significantly superior grades in tissue contrast of F-T (p = 0.001). Results of 3D FFE had superior grades in tissue contrast of F-C, but these result were not statistically significant (p = 0.157). VISTA had significantly superior CRs in F-T (p = 0.002), and 3D FFE had superior CRs in F-C (p = 0.003). The SNR of VISTA was higher than that of 3D FFE (49.24 vs. 15.94).

Conclusion

VISTA demonstrates superior tissue contrast between fluid and the Achiles tendon in terms of quantitative and qualitative analysis, while 3D FFE shows superior tissue contrast between fluid and cartilage in terms of quantitative analysis.

3D isotropic turbo spin-echo intermediate-weighted sequence with refocusing control in knee imaging: comparison study with 3D isotropic fast-field echo sequence

BACKGROUND

Three-dimensional (3D) turbo spin-echo (TSE) images have been used in imaging of the extremities and comparable diagnostic performance to two-dimensional (2D) TSE images has been reported in several studies. However, comparison of the 3D isotropic TSE intermediate-weighted sequence and 3D FFE sequence in terms of image quality has not been investigated.

PURPOSE

To compare the image quality of a 3D isotropic TSE intermediate-weighted sequence with refocusing control (volume isotropic turbo spin echo acquisition [VISTA]) and a 3D isotropic fast-field echo (FFE) sequence of the knee joint.

MATERIAL AND METHODS

3.0 T knee MRI with VISTA and 3D FFE sequences was performed in 10 healthy volunteers (3 men, 7 women; age range 26-30 years). Two radiologists with specialties in the musculoskeletal system assessed tissue contrast between the fluid-cruciate ligament (F-L), fluid-meniscus (F-M), and fluid-cartilage (F-C) based on a 4-point scale (1, poor; 2, fair; 3, good; and 4, excellent). Statistical analysis for inter-observer agreement and differences in grades in tissue contrast between VISTA and 3D FFE images (Wilcoxon signed-rank) were performed. For a quantitative analysis, the signal-to-noise ratio (SNR) was obtained by imaging phantom and noise-only image. Image contrast ratios (CRs) were calculated between F-L, F-M, and F-C in volunteer images of VISTA and 3D FFE and compared statistically with a paired t-test.

RESULTS

Based on qualitative analysis, VISTA had statistically superior grades of tissue contrast in F-L (P < 0.001) and F-M (P < 0.001). 3D FFE had superior but not statistically significant (P = 0.317) grades in F-C. Based on quantitative analysis, the SNR of the phantom imaging was higher in VISTA than that in 3D FFE (28.18 vs. 14.90). VISTA had superior CRs in F-L (P < 0.001) and F-M (P < 0.001). 3D FFE had superior CR in F-C (P = 0.038).

CONCLUSION

The VISTA sequence was superior in tissue contrast between F-M and F-L, and 3D FFE was superior in tissue contrast between F-C, subjectively and quantitatively.

Muscle fat fraction in neuromuscular disorders: dual-echo dual-flip-angle spoiled gradient-recalled MR imaging technique for quantification--a feasibility study

PURPOSE

To prospectively evaluate the muscle fat fraction (MFF) measured with dual-echo dual-flip-angle spoiled gradient-recalled acquisition in the steady state (SPGR) magnetic resonance (MR) imaging technique by using muscle biopsy as the reference standard.

MATERIALS AND METHODS

After ethics approval, written informed consent from all patients was obtained. Twenty-seven consecutive patients, evaluated at the Neuromuscular Disorders Center with a possible diagnosis of neuromuscular disorder, were prospectively studied with MR imaging of the lower extremities to quantify muscle fatty infiltration by means of MFF calculation. Spin-density- and T1-weighted fast SPGR in-phase and opposed-phase dual-echo sequences were performed, respectively, with 20° and 80° flip angles. Round regions of interest were drawn by consensus on selected MR sections corresponding to anticipated biopsy sites. These were marked on the patient's skin with a pen by using the infrared spider light of the system, and subsequent muscle biopsy was performed. MR images with regions of interest were stored on a secondary console where the MFF calculation was performed by another radiologist blinded to the biopsy results. MFFs calculated with dual-echo dual-flip-angle SPGR MR imaging and biopsy were compared by using a paired t test, Pearson correlation coefficient, and Bland-Altman plots. P value of < .05 was considered to indicate a statistically significant difference.

RESULTS

The mean MFFs obtained with dual-echo dual-flip-angle SPGR MR imaging and biopsy were 20.3% (range, 1.7%-45.1%) and 20.6% (range, 3%-46.1%), respectively. The mean difference, standard deviation of the difference, and t value were -0.3, 1.3, and -1.3 (P > .2), respectively. The Pearson correlation coefficient was 0.995; with the Bland-Altman method, all data points were within the ± 2 SDs limits of agreement.

CONCLUSION

The results show that dual-echo dual-flip-angle SPGR MR imaging technique provides reliable calculation of MFF, consistent with biopsy measurements.

Diagnosis of internal derangement of the knee at 3.0-T MR imaging: 3D isotropic intermediate-weighted versus 2D sequences

PURPOSE

To compare three-dimensional (3D) isotropic fast spin-echo (SE) intermediate-weighted magnetic resonance (MR) imaging with two-dimensional (2D) fast SE MR imaging-both performed at 3.0 T-for performance in the diagnosis of internal derangements of the knee.

MATERIALS AND METHODS

The institutional review board approved this HIPAA-compliant study, and the requirement for informed consent was waived. The authors retrospectively reviewed 87 knee MR images obtained in 85 patients who had undergone both 3D isotropic and 2D MR examinations of the knee at 3.0 T and subsequent arthroscopic surgery. The 2D MR images included intermediate-weighted coronal and sagittal images, intermediate-weighted axial images with fat saturation, and T2-weighted sagittal images. The 3D isotropic MR images were obtained with multiplanar reformation (MPR), a fast SE intermediate-weighted sequence, and a reconstruction voxel size of 0.5 x 0.5 x 0.5 mm. Two radiologists retrospectively and independently evaluated the 2D and 3D data sets, at different sessions, for the presence of medial meniscus (MM), lateral meniscus (LM), anterior cruciate ligament (ACL), and posterior cruciate ligament (PCL) tears. These interpretations were compared with the arthroscopic surgery findings. The statistical differences between the sensitivities, specificities, and accuracies of the two methods were determined at McNemar testing, with surgical findings serving as the reference standard. Interobserver agreement was calculated by using kappa coefficients.

RESULTS

For both reviewers, the sensitivity, specificity, and accuracy of both MR techniques were higher than 95% for the diagnosis of ACL and PCL tears, higher than 85% for the diagnosis of MM tears, and higher than 80% for the diagnosis of LM tears. There were no significant differences in sensitivity, specificity, or accuracy between the two methods. Interobserver agreement for evaluation of all lesions was excellent and ranged from 0.81 (LM tears evaluated with 3D and 2D sequences) to 0.93 (ACL tears evaluated with 3D and 2D sequences, PCL tears evaluated with 2D sequence, and MM tears evaluated with 3D sequence).

CONCLUSION

The performance of 3D isotropic fast SE intermediate-weighted MR imaging with MPR was not significantly different from that of 2D MR imaging in the diagnosis of cruciate ligament and meniscal tears of the knee.

Functional MR imaging of the female pelvis

Recent developments in MR techniques have magnified the roles and potential of MRI in the female pelvis. This article reviews the techniques and clinical applications of functional MRI (fMRI) of the female pelvis, including cine MRI, diffusion-weighted MRI (DWI), and dynamic contrast-enhanced (DCE)-MRI. Cine MRI is a useful tool for evaluating uterine contractility, including sustained contraction and peristalsis, in a variety of conditions and gynecologic disorders, and for evaluating pelvic-floor weakness. DWI can demonstrate abnormal signals in pathologic foci based on differences in molecular diffusion. It also enables the quantitative evaluation of the apparent diffusion coefficient (ADC), which may be useful for distinguishing malignant from benign tissues and monitoring therapeutic outcome. DCE-MRI has the potential to improve tumor detection and local staging, and can also provide quantitative information about perfusion of the tumor, which may be useful for both monitoring therapeutic effects and predicting therapeutic outcome. Understanding the roles played by functional MR techniques in the female pelvic region is beneficial not only for determining clinical applications, but also for developing further investigations with MRI.

Isotropic 3D fast spin-echo with proton-density-like contrast: a comprehensive approach to musculoskeletal MRI

OBJECTIVE

Scanning time considerations have restricted routine use of 3D Fourier transform (3DFT)-encoded MRI to gradient-recalled echo sequences. We sought to combine isotropic 3DFT acquisition with fast spin-echo at a practical scan duration. This strategy offers versatile image contrast for musculoskeletal evaluation and facilitates image reformation tailored to the depiction of small anatomic features.

CONCLUSION

Isotropic 3DFT fast spin-echo is feasible on current MRI scanners and has the potential to improve musculoskeletal evaluation.

Current status of body MR imaging: fast MR imaging and diffusion-weighted imaging

The recent technical advances in fast MR imaging have greatly enhanced the clinical value of MR imaging of the body. Advances in T(1)-weighted images have enabled the acquisition of dynamic contrast-enhanced MR imaging, which is currently central to hepatic MR imaging for detection and characterization of liver tumors and is also useful for the evaluation of myometrial invasion in uterine cor-pus cancer. Advances in rapid T(2)-weighted MR imaging with single-shot fast spin-echo images have enabled MR cholangiopancreatography and MR urography. Application of respiratory triggering can also provide T(2)-weighted images of high quality. Cine MR imaging utilizing ultrafast MR sequences enables the assessment of the respiratory motion of the lung for evaluating thoracic wall invasion by tumors. Diffusion-weighted images can provide excellent tissue contrast based on molecular diffusion and have the potential to demonstrate malignant tumors. Quantitative measurement of apparent diffusion coefficient values may also be valuable in distinguishing malignancies from benign lesions.

Comparison of gadobenate dimeglumine-enhanced dynamic MRI and 16-MDCT for the detection of hepatocellular carcinoma

OBJECTIVE

The objective of our study was to compare the diagnostic performance of gadobenate dimeglumine-enhanced MRI with that of 16-MDCT for the detection of hepatocellular carcinoma using receiver operating characteristic (ROC) curve analysis.

MATERIALS AND METHODS

Thirty-one patients with 53 hepatocellular carcinomas underwent gadobenate dimeglumine-enhanced dynamic MRI and multiphasic CT using 16-MDCT within a mean interval of 5 days (range, 3-9 days). The dynamic MRI examination was performed using 3D fat-saturated volumetric interpolated imaging and sensitivity encoding on a 1.5-T unit. Both dynamic MRI and multiphasic MDCT included dual arterial phase images. Three observers independently interpreted the CT and MR images in random order, separately, and without patient identifiers. The diagnostic accuracy of each technique was evaluated using the alternative-free response ROC method. The sensitivity and positive predictive values were also calculated.

RESULTS

The sensitivities of gadobenate dimeglumine-enhanced MRI for all observers were significantly higher than those of MDCT for all the lesions and for lesions 1.0 cm or smaller (p < 0.05); however, for lesions larger than 1.0 cm, the sensitivities of the two imaging techniques were similar. The mean area under the ROC curve (A(z)) of gadobenate dimeglumine-enhanced MRI (0.87 +/- 0.03 [SD]) was higher than that of MDCT (0.83 +/- 0.04), but no significant difference was found between them (p = 0.31). The number of false-positive findings on dynamic MRI was slightly higher than on MDCT, but no significant difference in the positive predictive value between the two imaging techniques was detected (observer 1, p = 0.06; observer 2, p = 0.13; observer 3, p = 1.00).

CONCLUSION

Gadobenate dimeglumine-enhanced MRI has a higher sensitivity for small hepatocellular carcinomas (</= 1 cm) but a higher false-positive rate due to nonspecific enhancement of benign lesions, such as arterioportal shunt, leading to no significant difference of overall accuracy when compared with MDCT.

Three-dimensional dynamic liver MR imaging using sensitivity encoding for detection of hepatocellular carcinomas: comparison with superparamagnetic iron oxide-enhanced mr imaging

PURPOSE

To assess the diagnostic performance of three-dimensional dynamic liver imaging with sensitivity encoding (SENSE), including double arterial phase images and increased resolution, by comparing it to superparamagnetic iron oxide (SPIO)-enhanced magnetic resonance (MR) imaging for the detection of hypervascular hepatocellular carcinoma (HCC).

MATERIALS AND METHODS

Twenty-seven consecutive patients with 50 HCCs underwent Gd-BOPTA-enhanced dynamic imaging using SENSE and SPIO-enhanced MR imaging with at least a 24-hour interval between examinations. Using a three-dimensional gradient-echo technique applying SENSE, dynamic imaging consisting of double arterial phase-, portal phase- and delayed phase-images, was obtained. Using T2-weighted turbo spin-echo and T2*-weighted fast imaging with steady-state precession sequence, SPIO-enhanced MR imaging was obtained. For qualitative analysis, the diagnostic accuracy of both MR examinations for detecting the 50 HCCs was evaluated using the alternative free-response receiver operating characteristic method. Sensitivity and positive predictive value were also evaluated.

RESULTS

The mean sensitivity and positive predictive value of three-dimensional dynamic imaging with SENSE were 91.3% and 89.2%, respectively, and those of SPIO-enhanced imaging were 77.3% and 92.6 %, respectively. There was a significant difference in sensitivity between the two images (P <0.05). The mean Az value of three-dimensional dynamic imaging with SENSE (0.97 +/- 0.01) was significantly higher than that of SPIO-enhanced imaging (0.90 +/- 0.02) (P=0.00).

CONCLUSION

Three-dimensional dynamic liver MR imaging using SENSE for acquiring double arterial phase images is more efficient than SPIO-enhanced MR imaging for detecting HCCs.

Parallel acquisition techniques for accelerated volumetric interpolated breath-hold examination magnetic resonance imaging of the upper abdomen: Assessment of image quality and lesion conspicuity

PURPOSE

To evaluate the impact of parallel acquisition techniques (PATs) on image quality and detection of liver metastases using three-dimensional volumetric interpolated breath-hold examination (VIBE) for clinical liver imaging.

MATERIALS AND METHODS

Forty-nine patients with various primary malignancies underwent abdominal dynamic contrast-enhanced three-dimensional VIBE magnetic resonance imaging (MRI) (1.5 T) using a standard phased array coil. Recently introduced Generalized Autocalibrating Partially Parallel Acquisition (GRAPPA) and SENSitivity Encoding (mSENSE) PAT reconstruction algorithms were added to reduce scan time twofold. Overall image quality, motion, and aliasing artifacts were classified on a 5-point scale. Signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) measurements were performed for quantitative comparison. All sequences were evaluated concerning the number of detected lesions.

RESULTS

PAT resulted in a reduction of data acquisition time from 23 to 13 seconds. Both GRAPPA and mSENSE data sets yielded 30% less SNR (34.8 +/- 14.1 and 33.1 +/- 13.3, P < 0.001) and 35% less CNR (21.2 +/- 15.0 and 20.9 +/- 12.7, P < 0.05) in comparison to unaccelerated VIBE (SNR = 50.8 +/- 20.3/CNR = 32.5 +/- 19.1). Similarly, PAT revealed lower-image-quality scores than unaccelerated VIBE. GRAPPA resulted in more fold-over artifacts than mSENSE. mSENSE revealed slightly fewer motion artifacts than no PAT. The unaccelerated late-venous-phase VIBE sequence revealed 146 lesions in the same patients. Accelerated images with mSENSE reconstruction detected 138 lesions. GRAPPA revealed 127 lesions, and thus performed inferior to mSENSE.

CONCLUSION

At least for arrays with small numbers of elements, such as arrays used in this study, the PAT-induced reduction in scanning times must be weighed against compromises in image quality, which translate into poorer diagnostic performance regarding detection of small hepatic lesions. Thus, the PAT implementations tested in this study should probably be reserved for patients unable to hold their breaths for regular three-dimensional VIBE data sets.

Magnetic resonance imaging of the palmar aspect of the equine podotrochlear apparatus: normal appearance

The purpose of this study was to describe the normal magnetic resonance (MR) imaging characteristics of the palmar structures of the equine podotrochlear apparatus by means of retrospective evaluation of MR imaging studies of 16 cadaver limbs. The articular aspect of the distal sesamoid bone was not evaluated in this study. Equine digits were imaged with a human knee radiofrequency coil in a 1.5 T magnetic field, using spin echo (SE) T1-weighted, turbo spin echo proton density (TSE PD)-weighted with and without fat saturation (FS), and FS TSE T2-weighted sequences. The limbs were dissected after imaging to validate the absence of gross abnormalities of the flexor aspect of the distal sesamoid bone, of the deep digital flexor tendon, and the distal impar sesamoidean ligament. Seven deep digital flexor tendons were subjected to histologic examination to exclude any microscopic tendon pathology. The anatomic structures of the podotrochlear apparatus were easily identified on MR images. Compact bone of the flexor cortex of the distal sesamoid bone had low intensity signal on all sequences. In 11 digits an increased signal was seen within the thickness of the sagittal eminence of the flexor cortex in SE T1-weighted images and in TSE PD-weighted images without FS. Trabecular bone had a granular appearance and high signal in SE T1-weighted sequences and TSE images without FS. The deep digital flexor tendon had low signal on FS T2-weighted images, while on short echo time sequences (T1- and PD-weighted sequences), the tendon signal varied depending on the relative orientation between its fibers and the static magnetic field. Seven tendons had stippled appearance due to small intratendonous foci of slightly increased signal on transverse T1-weighted images. MR imaging provides a thorough evaluation of the anatomical structure of the podotrochlear apparatus: A good knowledge of the MR imaging appearance and anatomy and an awareness of potential pitfalls will improve diagnostic specificity for the detection of pathologic changes.

Suppression of cerebrospinal fluid and blood flow artifacts in FLAIR MR imaging with a single-slab three-dimensional pulse sequence: initial experience

The authors compared high-signal-intensity flow-related artifacts present with a conventional two-dimensional (2D) fluid-attenuated inversion recovery (FLAIR) sequence with those seen with a single-slab, three-dimensional (3D) FLAIR sequence. Four readers graded the subarachnoid space and intraventricular artifacts, the pulsation artifacts, and the conspicuity of cranial nerves in the posterior fossa. For all comparisons, differences between 2D and 3D images were highly statistically significant, with 3D imaging being superior in all cases.

Evaluation of syringomyelia with three-dimensional constructive interference in a steady state (CISS) sequence

The purpose of this study was to evaluate a three-dimensional (3D) constructive interference in steady state (CISS) sequence in the assessment of syringomyelia. Eleven patients with syringomyelia were prospectively studied with magnetic resonance imaging. All patients underwent sagittal imaging with T1- and T2-weighted spin-echo (SE), and 3D-CISS sequences. The SE and 3D-CISS images, as well as multiplanar reconstruction (MPR) images of the 3D-CISS sequence, were analyzed with regard to image quality, degree of artifacts, visualization of the extent and internal structure of the syringomyelia, and contrast-to-noise ratio (CNR) of the fluid within the syringomyelia. Contrast between the spinal cord and cerebrospinal fluid (CSF), as well as delineation was significantly poorer for the T1-weighted SE sequence than for the 3D-CISS sequence (P < 0.01), while there was no significant difference between the T2-weighted SE sequence and the 3D-CISS sequence. Artifacts induced by CSF flow were significantly more for the T2-weighted SE sequence than for the 3D-CISS sequence (P < 0.01). Although the extent of syringomyelia was delineated equally among the three sequences in 9 of 11 patients, it was better for the 3D-CISS sequence than for the SE sequences in the remaining two. Septation and communication between the cavities were best detected by the 3D-CISS MPR images. The CNR of the 3D-CISS sequence was significantly higher than that of the SE sequence (P < 0.01). The 3D-CISS sequence demonstrates the extent and internal structures of syringomyelia better than conventional SE sequences and should be added to SE sequences in the evaluation of syringomyelia.

MRI simulation-based evaluation of image-processing and classification methods

With the increased interest in computer-aided image analysis methods, there is a greater need for objective methods of algorithm evaluation. Validation of in vivo MRI studies is complicated by a lack of reference data and the difficulty of constructing anatomically realistic physical phantoms. We present here an extensible MRI simulator that efficiently generates realistic three-dimensional (3-D) brain images using a hybrid Bloch equation and tissue template simulation that accounts for image contrast, partial volume, and noise. This allows image analysis methods to be evaluated with controlled degradations of image data.

Abdominal MR imaging with a volumetric interpolated breath-hold examination

PURPOSE

To compare a T1-weighted, three-dimensional (3D), gradient-echo (GRE) sequence for magnetic resonance (MR) imaging of the body (volumetric interpolated breath-hold examination, or VIBE) with a two-dimensional (2D) GRE breath-hold equivalent.

MATERIALS AND METHODS

Twenty consecutive patients underwent 1.5-T MR imaging. The examinations included pre- and postcontrast (20 mL gadopentetate dimeglumine) fat-saturated 2D GRE breath-hold imaging and fat-saturated volumetric interpolated breath-hold imaging before, during (arterial phase), and after injection, with thin (2-mm source images) and thick (8-mm reconstruction images) sections. The three images were compared qualitatively and quantitatively (signal-to-noise ratio [SNR] and contrast-to-noise ratio [CNR]).

RESULTS

Qualitatively, the 2-mm source images had poorer pancreatic edge definition on precontrast images compared with the other two data sets (P < .05). On gadolinium-enhanced images, scores for clarity of pancreatic edge, number of vessels visualized, and arterial ghosting were significantly lower for the postcontrast 2D GRE images. Quantitatively, SNR measurements in the liver, aorta, and renal cortex on pre- and postcontrast images were significantly higher for the 8-mm reconstruction images than for the 2D GRE or 2-mm source images (P < .05). Aorta-to-fat CNR was significantly higher on the 8-mm reconstruction images.

CONCLUSION

Fat-saturated volumetric interpolated breath-hold images have quality comparable to that of conventional fat-saturated 2D GRE images.

Low-field magnetic resonance imaging of the canine stifle joint: normal anatomy

Low-field magnetic resonance imaging (MRI) was performed on the stifle joints of four normal adult mongrel dogs using a 0.064 Tesla scanner. Markers were placed on each stifle joint to serve as reference points for comparing gross sections with the images. A T1-weighted sequence was used to image one stifle joint on each dog in the sagittal plane and the other stifle joint in the dorsal plane. The dogs were euthanized immediately following MRI and the stifle joints frozen intact. Each stifle joint was then embedded in paraffin, again frozen, and sectioned using the markers as reference points. On T1-weighted images, synovial fluid had low signal intensity (dark) compared to the infrapatellar fat pad which had a high signal intensity (bright). Articular cartilage was visualized as an intermediate bright signal and was separated from trabecular bone by a dark line representing subchondral bone. Menisci, fibrous joint capsule, and ligamentous structures appeared dark. In the true sagittal plane, the entire caudal cruciate ligament was often seen within one image slice. The patella was visualized as an intermediate bright signal (trabecular bone) surrounded by a low intensity signal (cortical bone). The trochlea and the intercondylar notch were difficult areas to analyze due to signal volume averaging of the curved surface of these areas and the presence of several types of tissues.

Assessment of pituitary microadenomas: comparison between 2D and 3D MR sequences

The aim of this study was to compare the diagnostic value and image quality of two different T1-weighted MR sequences in the evaluation of PRL-secreting pituitary microadenomas. Twenty-four patients with a clinical diagnosis of pituitary microprolactinoma were prospectively examined with both a 2D SE and a 3D turbo-SE T1-weighted coronal sequence, before and after intravenous contrast medium administration. Evaluation of MR images was done only on postcontrast images and considered the number of slices in which the lesions were visible and image quality, which was evaluated by both a subjective score and objective parameters (signal-to-noise and contrast-to-noise ratios). Pituitary microprolactinomas are visible in a higher number of slices in 3D TSE sequences; the subjective scores for image quality and signal-to-noise ratios were similar in both 2D and 3D sequences; the contrast-to-noise ratio was always higher in 3D sequences. In patients with hyperprolactinemia, the authors recommend using coronal 3D TSE T1-weighted sequences for evaluation of the pituitary region. 2D SE T1-weighted sequences may be considered if a shorter examination time is required (i.e., for claustrophobic patients) and in assessment or follow-up of microadenomas when a larger size lesion is clinically suspected.

Magnetic resonance imaging of the brain with gadopentetate dimeglumine-DTPA: comparison of T1-weighted spin-echo and 3D gradient-echo sequences

Short TR, short TE, high resolution, 3D gradient-recalled echo (GRE) imaging was evaluated for lesion detection in the brain. High resolution 3D GRE data acquisition was used to reduce partial volume effects and flow artifacts, to better visualize smaller structures, to minimize signal losses caused by field inhomogeneities, and to allow better image reformatting. Spin-echo (SE) and 3D GRE approaches were compared for lesion detection after the administration of an MR contrast agent, gadopentetate dimeglumine. Preliminary clinical studies demonstrated that the signal-to-noise ratio (SNR) in each slice of the GRE scan was worse than that of the SE scan because of the much thicker slices acquired with the SE technique. However, by averaging two adjacent 3D slices, the SNR of the two methods was essentially equivalent. In the averaged GRE slices, large lesions were seen just as well as in the SE images. More importantly, small lesions were better visualized in the thin 3D GRE images than in the thick SE images for the lesions studied in this work and the protocols used. These observations were confirmed by theoretical simulations.

3D echo planar imaging: application to the human head

In this work, the authors present 3D images acquired from the human head using echo planar encoding for two of the three dimensions of k-space. The third dimension of k-space is filled by selecting and phase encoding a slab of spins as in conventional 3D steady state (GRASS based) acquisition regimens. Using this approach, a 128 x 64 x 64 3D data matrix could be obtained in 3.4-4.7 sec using effective TE values of 24 and 34 ms, respectively. High quality 3D images could be acquired once phase ghosts present on 2D images were minimized through proper adjustments of scanner hardware.

Segmentation of brain parenchyma and cerebrospinal fluid in multispectral magnetic resonance images

Presents a new method to segment brain parenchyma and cerebrospinal fluid spaces automatically in routine axial spin echo multispectral MR images. The algorithm simultaneously incorporates information about anatomical boundaries (shape) and tissue signature (grey scale) using a priori knowledge. The head and brain are divided into four regions and seven different tissue types. Each tissue type c is modeled by a multivariate Gaussian distribution N(mu(c),Sigma(c)). Each region is associated with a finite mixture density corresponding to its constituent tissue types. Initial estimates of tissue parameters {mu(c),Sigma(c )}(c=1,...,7) are obtained from k-means clustering of a single slice used for training. The first algorithmic step uses the EM-algorithm for adjusting the initial tissue parameter estimates to the MR data of new patients. The second step uses a recently developed model of dynamic contours to detect three simply closed nonintersecting curves in the plane, constituting the arachnoid/dura mater boundary of the brain, the border between the subarachnoid space and brain parenchyma, and the inner border of the parenchyma toward the lateral ventricles. The model, which is formulated by energy functions in a Bayesian framework, incorporates a priori knowledge, smoothness constraints, and updated tissue type parameters. Satisfactory maximum a posteriori probability estimates of the closed contour curves defined by the model were found using simulated annealing.

Water and fat MR imaging with chemical shift selective 3D steady state methods

A new 3D acquisition regimen that enables the collection of conventional, water-suppressed, and fat-suppressed images with no increase in scan times compared with currently implemented 3D sequences is presented. The method is based on conventional 3D steady state with interleaved selective excitation of the fat resonance resulting in acquisition of a fat-based image during the TR period experienced by the water spins. This new sequence is relatively tolerant to susceptibility artifacts and results in excellent water-based images. Because the idea, which we propose, is independent of the type of steady-state imaging protocol utilized, it can be easily applied with regimens that are specifically tailored to enhance contrast.

High-resolution variable flip angle 3D MR imaging of trabecular microstructure in vivo

Two conceptually related variable-flip-angle 3D spin-echo pulse sequences were designed for imaging at voxel sizes of 2-5 x 10(-3) mm3 corresponding to pixel areas of less than 100 x 100 microns2 and section thicknesses on the order of 300-400 microns on a conventional 1.5 T MR imaging system equipped with 1 G/cm imaging field gradients, providing 12 sections in 10 min imaging time. The pulse sequences make use of the concept of restoring longitudinal magnetization inverted by the 180 degrees phase reversal pulse and are derivatives of pulse sequences previously dubbed "FATE" and "RASEE." It is shown that even in the small-voxel regime (< 10(-2) mm3 voxel size) and at echo times on the order of 10 ms, gradient echo images are sensitive to intrinsic fields causing artifactual boundary effects, including signal loss from intravoxel phase scrambling and spatial mismapping. At this resolution the variable flip-angle spin-echo pulse sequences are demonstrated to be better suited for imaging magnetically heterogeneous systems such as trabecular bone microstructure in vivo. These pulse sequences are found to be substantially less sensitive to distortions from magnetic dipole fields occurring at the boundaries of two phases of different magnetic permeability.

Rapid three-dimensional T1-weighted MR imaging with the MP-RAGE sequence

The authors investigated the application of three-dimensional (3D) magnetization-prepared rapid gradient-echo (MP-RAGE) imaging to the acquisition of small (32 x 128 x 256) T1-weighted 3D data sets with imaging times of approximately 1 minute. A theoretical model was used to study the contrast behavior of brain tissue. On the basis of these theoretical results, 3D MP-RAGE sequences were implemented on a 1.5-T whole-body imager. Thirty-two-section 3D data sets demonstrating good signal-to-noise ratios and resolution and strong T1-weighted contrast were obtained in 1 minute. Compared with standard short TR/TE spin-echo sequences with the same imaging times and comparable sequence parameters, the 3D MP-RAGE sequence delivered increases of more than 50% in the white matter/gray matter signal difference-to-noise and white matter signal-to-noise ratios, and provided almost twice as many sections. These sequences may find a clinical role in 3D scout imaging and screening and in patients with claustrophobia or trauma.

Magnetic resonance imaging of the temporomandibular joint: preliminary evaluation of partial flip angle three-dimensional volume acquisitions against conventional single and multiecho pulse sequences

The authors imaged 20 temporomandibular joints (TMJs) in 20 patients with high field surface coil magnetic resonance imaging, to compare the accuracy and the clinical utility of thin section partial flip angle (PFA) three dimensional (3D) volume acquisitions with T1-weighted images (T1WI), spin echo, multiecho, long TR/short and long TE images, PFA gradient recalled acquisition in the steady state (GRASS) images in the diagnosis of internal derangement of the TMJ. Pathological conditions demonstrated on imaging studies included disk derangement, joint effusion, and osteocartilaginous lesions, such as degenerative arthritis, osteoporosis, osteochondritis dissecans and early avascular necrosis of the mandibular condyle.

Measurement of nonuniform current density by magnetic resonance

A noninvasive tissue current measurement technique and its use in measuring a nonuniform current density are described. This current density image is created by measuring the magnetic field arising from these currents and taking its curl. These magnetic fields are proportional to the phase component of a complex magnetic resonance image. Measurements of all three components of a quasistatic nonuniform current density in a phantom are described. Expected current density calculations from a numerical solution for the magnetic field which was created by the phantom are presented for comparison. The results of a numerical simulation of the experiment, which used this field solution and which included the effects of slice selection and sampling, are also presented. The experimental and simulated results are quantitatively compared. It is concluded that the principle source of systematic error was the finite slice thickness, which causes blurring of boundaries.

Low-field 3-DFT MRI: conceptual, analytical, and experimental aspects

Three-dimensional Fourier transform (3-DFT) magnetic resonance imaging (MRI) offers advantages in terms of signal-to-noise (S/N) per unit of time for the case where a large number of slices is desired. This advantage is enhanced when the relaxation time, T(1 ), is short. Because time limitations in 3-DFT imaging force the use of short time intervals, TR, between excitations of a slice, lesion contrast is often undesirable at mid-and high-field strength even when the S/N is good. At low fields, where T(1) values are short, high S/N and contrast can both be achieved with 3-DFT MR images. The conceptual and analytical aspects of low-field 3-DFT MRI are presented and demonstrated at 640 G.

Magnetic resonance imaging of the thoracic cavity using a paused 3DFT acquisition technique

A new pulse sequence designed for magnetic resonance imaging of the entire thoracic cavity is described. This sequence, called 3DPAUSE, is a rapid three-dimensional Fourier transform (3DFT) sequences with periodic pauses for breathing and additional rf pulses after each pause to restore the magnetization to steady-state before data acquisition resumes. Cardiac motion artifacts are effectively removed by signal averaging. Respiratory motion artifacts are removed by breath hold. Image artifacts caused by an inadequate number of pauses or by inappropriate placement of the pauses within a scan are shown, and ways to avoid these artifacts are discussed. 3DPAUSE provides the ability to acquire three-dimensional arrays in the thoracic cavity with minimal artifacts from respiratory and cardiac motions in a clinically reasonable time.

Noninvasive determination of coronary artery bypass graft patency by cine magnetic resonance imaging

Cine magnetic resonance imaging (MRI) is a gradient-recalled, retrospectively gated, fast-scan technique that depicts laminar flowing blood as bright signal and has been proposed as a useful method for determination of coronary artery bypass graft (CABG) patency. Therefore, we performed a blinded prospective study to assess the value of cine MRI determination of CABG patency in 20 patients with 45 CABG proximal anastomoses who were undergoing repeat angiography. Ten normal subjects served as controls to define normal intrathoracic vascular patterns. There were 21 left anterior descending (LAD) grafts, of which four were left internal mammary (LIMA), 12 left circumflex (Cx), and 12 right coronary (RCA) grafts. After localizing spin-echo coronal images were obtained, multiple axial multislice interleaved cine MRI acquisitions, each consisting of two to four 5-10-mm-thick slices at eight to 24 frames per cardiac cycle, were obtained from the superior main pulmonary artery to the inferior left ventricle. Each acquisition took 5-8 minutes with a subsequent 5-10 minutes of computer image reconstruction. Total study time per patient was 50-75 minutes. Known to cine MRI interpreters were the original surgical CABG insertions but not the angiographic findings. A graft was called patent if a bright graft flow signal, not corresponding to a normal vessel, was identified on multiple frames at multiple levels abutting the great vessels or epicardial surface of the heart. Angiographically, there were 33 patent grafts, of which 29 were identified as patent by cine MRI (sensitivity, 88%).(ABSTRACT TRUNCATED AT 250 WORDS)