USE OF THE T2∗ -WEIGHTED GRADIENT RECALLED ECHO SEQUENCE FOR MAGNETIC RESONANCE IMAGING OF THE CANINE AND FELINE BRAIN AMY W. HODSHON, SILKE HECHT, WILLIAM B. THOMAS
T2∗ -weighted magnetic resonance imaging (MRI) has been reported to help improve detection of intracranial hemorrhage and is widely used in human neuroimaging. To assess the utility of this technique in small animals, interpretations based on this sequence were compared with those based on paired T2-weighted and fluid-attenuated inversion recovery (FLAIR) sequences in 200 dogs and cats that underwent brain MRI for suspected intracranial disease. Two sets of images (T2 + FLAIR and T2∗ ) were reviewed separately in random order unaccompanied by patient information and were interpreted as normal or abnormal based on whether intracranial abnormalities were seen. The number and location of intracranial lesions were recorded. Eighty-five studies were considered normal and 88 were considered abnormal based on both sets of images, with good agreement (κ = 0.731) between the two. Susceptibility artifact was present in 33 cases (16.5%) on T2∗ -weighted images. In 12 cases (6%) a total of 69 lesions were seen on T2∗ -weighted images that were not seen on T2/FLAIR, all of which were associated with susceptibility artifact caused by presumed intracranial hemorrhage. Pseudolesions were seen on T2∗ -weighted images in five cases, none of which were associated with susceptibility artifact. Abnormalities were seen on T2/FLAIR images that were not seen on T2∗ -weighted images in 35 cases, confirming that T2∗ does not replace standard spin echo sequences. These results support inclusion of T2∗ -weighted sequences in small animal brain MRI studies and indicate that that a large number of abnormalities (especially hemorrhagic lesions) can go undetected if it is not performed. C 2014 American College of Veterinary Radiology. Key words: brain, cat, dog, gradient echo, MRI, T2∗ .
generation of a spin echo.1 Gradient echo sequences instead use the frequency encoding gradient to rephase the magnetic moments, but this gradient reversal does not cancel out the loss of phase coherence caused by magnetic field inhomogeneities.2 The resultant signal void in the image is referred to as susceptibility artifact. Inhomogeneity in the magnetic field can be caused by the presence of an air-tissue interface, imperfections in the magnet, calcification, or paramagnetic or ferromagnetic substances. T2∗ weighted pulse sequences use a long echo time and small flip angle to maximize the amount of T2∗ shortening and thereby the degree of susceptibility artifact.3 T2∗ -weighted imaging is particularly sensitive to hemorrhage because certain hemoglobin breakdown products are paramagnetic and therefore create local magnetic field inhomogeneities. In veterinary patients, the use of T2∗ -weighted gradient recalled echo imaging to detect hemorrhage has been reported in cases of primary cerebral hemangioma, metastatic hemangiosarcoma, pituitary adenoma and adenocarcinoma, astrocytoma, oligodendroglioma,
Introduction 2∗ -WEIGHTED GRADIENT-RECALLED ECHO MRI creates contrast among tissues based on both the rate of T2 decay (spin–spin relaxation) of each tissue as well as the rate of net magnetization vector dephasing caused by external and local magnetic field inhomogeneities that together are called T2∗ shortening or decay. The presence of magnetic field inhomogeneities exponentially speeds up loss of transverse magnetization. This allows T2∗ weighting to highlight local inhomogeneities that can go undetected on spin echo sequences, as the latter sequences compensate for magnetic field inhomogeneities by interposing one or more 180° refocusing pulses between the initial excitation pulse and the
T
From the Department of Small Animal Clinical Sciences, College of Veterinary Medicine, University of Tennessee, Knoxville, TN 379964544. Results of the study were presented at the ACVR Annual Scientific Meeting, October 8 – 11th 2013, Savannah GA, and at the UTVM CEM and Public Health Research Symposium, May 20th – 21st 2013, Knoxville, TN. Address correspondence and reprints to Amy Hodshon at the above address. E-mail: [email protected]. Received September 23, 2013; accepted for publication January 27, 2014. doi: 10.1111/vru.12164
Vet Radiol Ultrasound, Vol. 55, No. 6, 2014, pp 599–606.
599
600
HODSHON, HECHT, AND THOMAS
intravascular lymphoma, meningioma, Angiostrongylus vasorum infection, bacterial septicemia, hemorrhagic infarction, hemorrhagic stroke, cerebral microbleeds, subdural hematoma, spinal extradural hematoma associated with intervertebral disk herniation, hemorrhagic myelomalacia, traumatic intramedullary spinal cord hemorrhage, ischemic myelopathy, acute noncompressive nucleus pulposus extrusion, and necrotizing myopathy.4–22 However, the routine use of T2∗ -weighted gradient recalled echo pulse sequences in a large series of canine and feline patients has not been reported. This study compares the T2∗ -weighted gradient echo sequence with two spin echo based sequences—T2weighted and T2-weighted fluid-attenuated inversion recovery (FLAIR)—in dogs and cats with suspected brain disease with the goal of determining the utility of these sequences in detecting abnormalities, in particular areas of suspected hemorrhage.
Materials and Methods Medical records at the University of Tennessee Veterinary Medical Center were searched for dogs and cats that underwent a brain MRI examination due to clinical suspicion of intracranial disease. Patients were selected by one investigator (A.H.) by including the 200 most recent patients undergoing brain MRI; for patients who underwent more than one examination, only the initial study was included. All images were acquired using a 1.0 Tesla magnet.∗ A receive-only knee flex coil was used for patients weighing less than approximately 10 kg, and a receive-only body flex coil was used for patients weighing over approximately 10 kg. All studies included T2-weighted, T2-weighted FLAIR (hereafter referred to as “FLAIR”), and T2∗ -weighted sequences. Imaging parameters were tailored to each patient. In general, for T2-weighted images, repetition time (TR) was between 2000 and 5000 ms, echo time (TE) was 98 ms, and echo train length was 13. For FLAIR images, repetition time was approximately 8000 ms, echo time was approximately 90 ms, inversion time (TI) was 2300 ms, and echo train length was 19. For T2∗ -weighted images, repetition time was between 700 and 1000 ms, echo time was 26 ms, and flip angle was 20°. Slice thickness in the majority of patients was 3 mm with an interslice gap of 0.3 mm; in some larger patients, slice thickness was increased up to 4 mm. Transverse images of each sequence were reviewed by an observer (S.H.) who was blinded to patient signalment, history, and clinical findings. Paired T2-weighted and FLAIR images were reviewed from all 200 cases in a randomized order; T2∗ -weighted images were reviewed separately several ∗ Magnetom
Harmony, Siemens Medical Solutions, Malvern, PA
2014
weeks after the T2/FLAIR evaluation in a different randomized order. Images were evaluated for intracranial lesions, and the following were determined: normal or abnormal image, number of lesions (with the possibility for “diffuse/too numerous to count” lesions), intra-axial or extraaxial location (with the possibilities for “both” and “cannot tell”), and, for T2∗ -weighted images, whether susceptibility artifact(s) was present. Magnetic resonance imaging artifacts other than susceptibility artifact were not evaluated in the frame of this study. T2-weighted and FLAIR images were considered abnormal if a lesion was detected on either the T2-weighted or FLAIR images (not necessarily both); similarly, both sets of images were taken into account to determine number of lesions and lesion location. The statistical analysis was performed by a universityemployed statistician. The percentage of cases considered abnormal on T2 and FLAIR images was compared to the percentage of cases considered abnormal on T2∗ . The percentage of cases with intracranial susceptibility artifact present on T2∗ -weighted images was calculated. The number of cases considered abnormal on T2 and FLAIR images but considered normal on T2∗ was reported, and vice versa. Results of T2 and FLAIR sequences were considered different from results of T2∗ sequence if there were differences as to normal vs. abnormal, lesion location, the number of lesions (by more than one), or whether lesions were discrete and countable or diffuse/too numerous to count. Agreement between the T2/FLAIR sequences and the T2∗ sequence (normal vs. abnormal) was calculated using the κ statistic. Agreement as to the number of lesions (in cases considered abnormal by both T2/FLAIR and T2∗ ) was assessed using analysis of variance (ANOVA). Cases in which intracranial susceptibility artifact was seen and cases that differed between T2/FLAIR and T2∗ weighted sequences as to normal versus abnormal or number of lesions were reviewed by two investigators (S.H. and A.H.) by examining all imaging sequences acquired, along with the case signalment, history, clinical findings, and other diagnostic test results. In particular, T2∗ and T2/FLAIR images were examined together, and it was noted whether lesion(s) that had been missed on one of the sequences could be seen on that sequence in retrospect (i.e., could one legitimately be expected to see and report it). Abnormalities recorded as lesions on one sequence that were deemed to be normal anatomic structures based on all other available imaging sequences were recorded as pseudolesions. Clinical diagnoses (and definitive diagnoses, when available) were recorded. Diagnoses were grouped into the following categories for the purpose of analysis: mass (meningioma, glioma, pituitary, other/unknown), infiltrative or diffuse parenchymal disease [meningoencephalitis of unknown etiology (such as granulomatous or necrotizing meningoencephalitis), infectious meningoencephalitis,
T2∗ GRE SEQUENCE IN BRAIN MR IMAGING
VOL. 55, NO. 6
TABLE 1. Summarized Findings for Each Set of Images and Cases in which There was Agreement Between T2/FLAIR and T2∗ Sequences Number of cases (%)
All Normal brain Abnormal brain Number of lesions 1 2 3 4 ࣙ5 TNTC† /diffuse Lesion location Intra-axial Extra-axial I + E‡ Unable to determine † ‡
T2 + T2 FLAIR
T2∗
Sequences in agreement
200 (100.0) 91 (45.5) 109 (54.5)
200 (100.0) 106 (53.0) 94 (47.0)
173 (86.5) 85 (42.5) 88 (44.0)
74 (68) 16 (15) 7 (6) 1 (
T2∗ -weighted magnetic resonance imaging (MRI) has been reported to help improve detection of intracranial hemorrhage and is widely used in human neuroimaging. To assess the utility of this technique in small animals, interpretations based on this sequence were compared with those based on paired T2-weighted and fluid-attenuated inversion recovery (FLAIR) sequences in 200 dogs and cats that underwent brain MRI for suspected intracranial disease. Two sets of images (T2 + FLAIR and T2∗ ) were reviewed separately in random order unaccompanied by patient information and were interpreted as normal or abnormal based on whether intracranial abnormalities were seen. The number and location of intracranial lesions were recorded. Eighty-five studies were considered normal and 88 were considered abnormal based on both sets of images, with good agreement (κ = 0.731) between the two. Susceptibility artifact was present in 33 cases (16.5%) on T2∗ -weighted images. In 12 cases (6%) a total of 69 lesions were seen on T2∗ -weighted images that were not seen on T2/FLAIR, all of which were associated with susceptibility artifact caused by presumed intracranial hemorrhage. Pseudolesions were seen on T2∗ -weighted images in five cases, none of which were associated with susceptibility artifact. Abnormalities were seen on T2/FLAIR images that were not seen on T2∗ -weighted images in 35 cases, confirming that T2∗ does not replace standard spin echo sequences. These results support inclusion of T2∗ -weighted sequences in small animal brain MRI studies and indicate that that a large number of abnormalities (especially hemorrhagic lesions) can go undetected if it is not performed. C 2014 American College of Veterinary Radiology. Key words: brain, cat, dog, gradient echo, MRI, T2∗ .
generation of a spin echo.1 Gradient echo sequences instead use the frequency encoding gradient to rephase the magnetic moments, but this gradient reversal does not cancel out the loss of phase coherence caused by magnetic field inhomogeneities.2 The resultant signal void in the image is referred to as susceptibility artifact. Inhomogeneity in the magnetic field can be caused by the presence of an air-tissue interface, imperfections in the magnet, calcification, or paramagnetic or ferromagnetic substances. T2∗ weighted pulse sequences use a long echo time and small flip angle to maximize the amount of T2∗ shortening and thereby the degree of susceptibility artifact.3 T2∗ -weighted imaging is particularly sensitive to hemorrhage because certain hemoglobin breakdown products are paramagnetic and therefore create local magnetic field inhomogeneities. In veterinary patients, the use of T2∗ -weighted gradient recalled echo imaging to detect hemorrhage has been reported in cases of primary cerebral hemangioma, metastatic hemangiosarcoma, pituitary adenoma and adenocarcinoma, astrocytoma, oligodendroglioma,
Introduction 2∗ -WEIGHTED GRADIENT-RECALLED ECHO MRI creates contrast among tissues based on both the rate of T2 decay (spin–spin relaxation) of each tissue as well as the rate of net magnetization vector dephasing caused by external and local magnetic field inhomogeneities that together are called T2∗ shortening or decay. The presence of magnetic field inhomogeneities exponentially speeds up loss of transverse magnetization. This allows T2∗ weighting to highlight local inhomogeneities that can go undetected on spin echo sequences, as the latter sequences compensate for magnetic field inhomogeneities by interposing one or more 180° refocusing pulses between the initial excitation pulse and the
T
From the Department of Small Animal Clinical Sciences, College of Veterinary Medicine, University of Tennessee, Knoxville, TN 379964544. Results of the study were presented at the ACVR Annual Scientific Meeting, October 8 – 11th 2013, Savannah GA, and at the UTVM CEM and Public Health Research Symposium, May 20th – 21st 2013, Knoxville, TN. Address correspondence and reprints to Amy Hodshon at the above address. E-mail: [email protected]. Received September 23, 2013; accepted for publication January 27, 2014. doi: 10.1111/vru.12164
Vet Radiol Ultrasound, Vol. 55, No. 6, 2014, pp 599–606.
599
600
HODSHON, HECHT, AND THOMAS
intravascular lymphoma, meningioma, Angiostrongylus vasorum infection, bacterial septicemia, hemorrhagic infarction, hemorrhagic stroke, cerebral microbleeds, subdural hematoma, spinal extradural hematoma associated with intervertebral disk herniation, hemorrhagic myelomalacia, traumatic intramedullary spinal cord hemorrhage, ischemic myelopathy, acute noncompressive nucleus pulposus extrusion, and necrotizing myopathy.4–22 However, the routine use of T2∗ -weighted gradient recalled echo pulse sequences in a large series of canine and feline patients has not been reported. This study compares the T2∗ -weighted gradient echo sequence with two spin echo based sequences—T2weighted and T2-weighted fluid-attenuated inversion recovery (FLAIR)—in dogs and cats with suspected brain disease with the goal of determining the utility of these sequences in detecting abnormalities, in particular areas of suspected hemorrhage.
Materials and Methods Medical records at the University of Tennessee Veterinary Medical Center were searched for dogs and cats that underwent a brain MRI examination due to clinical suspicion of intracranial disease. Patients were selected by one investigator (A.H.) by including the 200 most recent patients undergoing brain MRI; for patients who underwent more than one examination, only the initial study was included. All images were acquired using a 1.0 Tesla magnet.∗ A receive-only knee flex coil was used for patients weighing less than approximately 10 kg, and a receive-only body flex coil was used for patients weighing over approximately 10 kg. All studies included T2-weighted, T2-weighted FLAIR (hereafter referred to as “FLAIR”), and T2∗ -weighted sequences. Imaging parameters were tailored to each patient. In general, for T2-weighted images, repetition time (TR) was between 2000 and 5000 ms, echo time (TE) was 98 ms, and echo train length was 13. For FLAIR images, repetition time was approximately 8000 ms, echo time was approximately 90 ms, inversion time (TI) was 2300 ms, and echo train length was 19. For T2∗ -weighted images, repetition time was between 700 and 1000 ms, echo time was 26 ms, and flip angle was 20°. Slice thickness in the majority of patients was 3 mm with an interslice gap of 0.3 mm; in some larger patients, slice thickness was increased up to 4 mm. Transverse images of each sequence were reviewed by an observer (S.H.) who was blinded to patient signalment, history, and clinical findings. Paired T2-weighted and FLAIR images were reviewed from all 200 cases in a randomized order; T2∗ -weighted images were reviewed separately several ∗ Magnetom
Harmony, Siemens Medical Solutions, Malvern, PA
2014
weeks after the T2/FLAIR evaluation in a different randomized order. Images were evaluated for intracranial lesions, and the following were determined: normal or abnormal image, number of lesions (with the possibility for “diffuse/too numerous to count” lesions), intra-axial or extraaxial location (with the possibilities for “both” and “cannot tell”), and, for T2∗ -weighted images, whether susceptibility artifact(s) was present. Magnetic resonance imaging artifacts other than susceptibility artifact were not evaluated in the frame of this study. T2-weighted and FLAIR images were considered abnormal if a lesion was detected on either the T2-weighted or FLAIR images (not necessarily both); similarly, both sets of images were taken into account to determine number of lesions and lesion location. The statistical analysis was performed by a universityemployed statistician. The percentage of cases considered abnormal on T2 and FLAIR images was compared to the percentage of cases considered abnormal on T2∗ . The percentage of cases with intracranial susceptibility artifact present on T2∗ -weighted images was calculated. The number of cases considered abnormal on T2 and FLAIR images but considered normal on T2∗ was reported, and vice versa. Results of T2 and FLAIR sequences were considered different from results of T2∗ sequence if there were differences as to normal vs. abnormal, lesion location, the number of lesions (by more than one), or whether lesions were discrete and countable or diffuse/too numerous to count. Agreement between the T2/FLAIR sequences and the T2∗ sequence (normal vs. abnormal) was calculated using the κ statistic. Agreement as to the number of lesions (in cases considered abnormal by both T2/FLAIR and T2∗ ) was assessed using analysis of variance (ANOVA). Cases in which intracranial susceptibility artifact was seen and cases that differed between T2/FLAIR and T2∗ weighted sequences as to normal versus abnormal or number of lesions were reviewed by two investigators (S.H. and A.H.) by examining all imaging sequences acquired, along with the case signalment, history, clinical findings, and other diagnostic test results. In particular, T2∗ and T2/FLAIR images were examined together, and it was noted whether lesion(s) that had been missed on one of the sequences could be seen on that sequence in retrospect (i.e., could one legitimately be expected to see and report it). Abnormalities recorded as lesions on one sequence that were deemed to be normal anatomic structures based on all other available imaging sequences were recorded as pseudolesions. Clinical diagnoses (and definitive diagnoses, when available) were recorded. Diagnoses were grouped into the following categories for the purpose of analysis: mass (meningioma, glioma, pituitary, other/unknown), infiltrative or diffuse parenchymal disease [meningoencephalitis of unknown etiology (such as granulomatous or necrotizing meningoencephalitis), infectious meningoencephalitis,
T2∗ GRE SEQUENCE IN BRAIN MR IMAGING
VOL. 55, NO. 6
TABLE 1. Summarized Findings for Each Set of Images and Cases in which There was Agreement Between T2/FLAIR and T2∗ Sequences Number of cases (%)
All Normal brain Abnormal brain Number of lesions 1 2 3 4 ࣙ5 TNTC† /diffuse Lesion location Intra-axial Extra-axial I + E‡ Unable to determine † ‡
T2 + T2 FLAIR
T2∗
Sequences in agreement
200 (100.0) 91 (45.5) 109 (54.5)
200 (100.0) 106 (53.0) 94 (47.0)
173 (86.5) 85 (42.5) 88 (44.0)
74 (68) 16 (15) 7 (6) 1 (
