Three‐dimensional T1‐weighted gradient echo is a suitable alternative to two‐dimensional T1‐weighted spin echo for imaging the canine brain

Abstract Volumetric imaging (VOL), a three‐dimensional magnetic resonance imaging (MRI) technique, has been described in the literature for evaluation of the human brain. It offers several advantages over conventional two‐dimensional (2D) spin echo (SE), allowing rapid, whole‐brain, isotropic imaging with submillimeter voxels. This retrospective, observational study compares the use of 2D T1‐weighted SE (T1W SE), with T1W VOL, for the evaluation of dogs with clinical signs of intracranial disease. Brain MRI images from 160 dogs who had T1W SE and T1W VOL sequences acquired pre‐ and postcontrast, were reviewed for presence and characteristics of intracranial lesions. Twenty‐nine of 160 patients were found to have intracranial lesions, all visible on both sequences. Significantly better grey‐white matter (GWM) differentiation was identified with T1W VOL (P < .001), with fair agreement between the two sequences (weighted κ = 0.35). Excluding a mild reduction in lesion intensity in three dogs precontrast on the T1W VOL images compared to T1W SE, and meningeal enhancement noted on the T1W VOL images in one dog, not identified on T1W SE, there was otherwise complete agreement between the two sequences. The T1W VOL sequence provided equivalent lesion evaluation and significantly improved GWM differentiation. Images acquired were of comparable diagnostic quality to those produced using a conventional T1W SE technique, for assessment of lesion appearance, number, location, mass effect, and postcontrast enhancement. T1W VOL, therefore, provides a suitable alternative T1W sequence for canine brain evaluation and can facilitate a reduction in total image acquisition time.


INTRODUCTION
Magnetic resonance imaging is the modality of choice for imaging the canine brain, since it provides excellent tissue contrast resolution and anatomic detail in addition to multiplanar image acquisition, without the use of ionizing radiation. [1][2][3] Increased caseloads have driven the need to acquire high-quality images more rapidly. Shorter scan times facilitate a higher case throughput, but are also advantageous to the patient as the duration of anesthesia is reduced.
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. Traditionally, SE sequences have been used to acquire twodimensional (2D), cross-sectional images of the canine brain. Gradient echo (GE) sequences typically use smaller flip angles than SE sequences (<90 • ) and a gradient to rephase the spins, rather than a 180 • radiofrequency pulse. Use of a gradient increases the speed of rephasing, and smaller flip angles mean less time is required for relaxation; both strategies permit a shorter time to echo (TE) and repetition time (TR) than in SE, thus allowing studies to be performed more rapidly. 4 Relatively short acquisition times permit the use of GE for Vet Radiol Ultrasound. 2019;60:543-551.
wileyonlinelibrary.com/journal/vru 543 three-dimensional (3D) acquisition, also known as volumetric imaging (VOL). This is the simultaneous acquisition of data from an entire volume of tissue, in a single acquisition using a nonselective excitation pulse. [5][6][7][8] Use of GE to acquire VOL images of the brain has been frequently described in the human literature. 7,[9][10][11][12][13] Spoiled GE sequences, for example, a fast low-angle shot (FLASH) sequence, can be used to generate T1W VOL images, with good temporal and spatial resolution. 4,6,7 Spoiled gradient echo uses a steady state, by using a very short TR and a medium flip angle, and a short TE to minimize T2* effects. 4 T1W VOL images are acquired in very thin slices (<1 mm) ideally with isotropic voxels, without a slice gap, that allow image reformatting so that the voxels can be displayed as a new matrix of pixels without loss of spatial resolution. 5,6,14

RESULTS
A total of 175 dogs were initially identified. Of these, 15 were excluded based on the previously detailed exclusion criteria. Thus, 160 dogs were included in analyses for the current study. The study population had a median age of 5 years (range 4 months to 14 years 10 months, with an interquartile range of 2 years 10 months to 8 years 4 months).
There were 62 neutered females, 56 neutered males, 24 entire males, and 18 entire females. Forty breeds were represented, the most common of which were: cross-breeds (28) Figure 1. An example of the difference seen in GWM differentiation between the two sequences is shown in Figure 2.  Figure 3); also a 13-year-old female neutered Patterdale Terrier cross, with a hemorrhagic mass lesion within the right forebrain). The fourth patient was a 5-year-old female neutered Lurcher presented for blindness. She was found to have a single, intra-axial, right parietal lobe hemorrhagic mass lesion, thought most likely to be neoplastic; it was hyperechoic precontrast, heterogeneously enhancing postcontrast, exerting a mass effect on both sequences. The inconsistency between the two sequences was the meningeal enhancement (of the pachymeninges) recorded as absent on the T1W SE sequence, yet present on the T1W VOL. (see Figure 4).

DISCUSSION
The findings of our study demonstrated that the T1W VOL sequence provided equivalent lesion evaluation to T1W SE imaging, for assessment of lesion appearance and enhancement. Additionally, significantly improved GWM differentiation was identified with the T1W VOL images compared to T1W SE. As hypothesized T1W VOL imaging can, therefore, provide a suitable alternative to conventional T1W SE MRI, for routine evaluation of the canine brain.
Volumetric imaging has several advantages over conventional 2D imaging. [5][6][7][8] The slice thickness can be much less than in conventional imaging, ≤0.9 mm in this study compared to 2-3 mm for the SE sequence, permitting high spatial resolution. Thinner slices and the lack of slice gap can lessen the likelihood of very small lesions being missed, and reduce partial volume effects. In 2D imaging, where acquisition is performed one slice at a time, slice thickness will affect the signal-to-noise ratio (SNR). In VOL, data are acquired simultaneously from an entire volume of tissue in a single acquisition and divided into slices by a slice select gradient, in a process known as slice encoding. 4 TA B L E 3 Summarized findings for each sequence showing the number (n) * and percentage (%) of patients exhibiting each feature evaluated and the number and percentage of patients in which there was agreement between the T1W SE and T1W VOL sequences  Since a whole volume of tissue is excited and there are no gaps, the SNR is increased. 4 Additionally because the data are collected from a slab of tissue rather than a single slice, this can be reformatted using MPR to allow assessment of the region of interest in any plane. 5,6,10 The use of small isotropic voxels (such as in this study) gives MPR's with high spatial resolution, which is equal regardless of plane. 4,16 This allows detailed evaluation of the brain and may permit improved detection of small intracranial lesions compared with 2D SE imaging. [17][18][19] The use of VOL imaging to acquire 3D data sets, negates the need for supplemental sequences in orthoganol imaging planes, reducing acquisition time compared to acquisition of 2D images in all three planes. [5][6][7]10,16 The T1W VOL sequence used in this study was a FLASH sequence, the use of which has been described in the human literature to obtain high resolution, very thin section, T1W images of the central nervous system. 6  imaging, data for the whole region of interest are collected throughout the image acquisition period. 6 In this study, the T1W VOL sequence produced brain images with significantly better GWM differentiation than the images acquired using the T1W SE sequence (P < .001). Improved GWM differentiation permits more detailed evaluation of the internal structure of the brain and may aid the localization and characterization of lesions/pathology, especially those resulting in a reduction in normal GWM distinction. Evidence from the human literature corroborates our finding of a superior GWM differentiation using a T1W VOL technique. 6,7,9,12 The greater GWM differentiation results from stronger T1W contrast achieved by the GE sequence. 21 These studies all employ a short-TR GE technique using a magnetization-preparation pulse (MP-RAGE), a 3D-Turbo FLASH technique, to produce T1W images. 21  A definitive diagnosis was not reached in many of the cases due to an absence of histopathological confirmation, and lack of relevant blood (infectious disease serology) or CSF analysis results. It is generally accepted that definitive diagnosis is not possible on the basis of MRI alone, because imaging features of neoplastic and certain nonneoplastic diseases are not sufficiently specific. 24,25 Histological examination is typically required for definitive diagnosis of intracranial neoplasms. 26,27 Magnetic resonance imaging is however regularly used for presumptive differentiation of neoplasia from inflammatory disease, and to provide probable or prioritized differential diagnoses to facilitate optimal patient management. 24,25,[28][29][30] During this study, we evaluated MRI signs that have previously been identified as being significantly associated with neoplasia, that is, a solitary lesion, presence of mass effect and contrast enhancement, 24 extra-axial location, 25 and those significantly more commonly identified in inflammatory disease: meningeal enhancement and multifocal lesions. 25 The aim of our study was however to evaluate agreement between the two sequences for assessment of several important MRI features, to determine whether a T1W VOL sequence could be used in place of the conventional sequence, rather than to evaluate accuracy for reaching a diagnosis. Additionally, the evaluation of T1W sequences alone, precludes diagnosis; concurrent assessment of other weightings, for example, T2W and FLAIR sequences, would typically be required for comprehensive evaluation of all brain lesions. The collection and inclusion of data regarding the clinical diagnosis for each patient in this study was intended to summarize the characteristics of the study population, rather than permit direct correlation with the study findings.
Based on the findings of this study, the T1W VOL technique provided comparable lesion evaluation and significantly improved GWM differentiation. The T1W VOL provided a suitable alternative T1W sequence for the evaluation of dogs with suspected intracranial disease. Therefore, T1W VOL imaging could replace conventional T1W SE for routine canine brain imaging, providing equivalent lesion appraisal and a reduction in total image acquisition time.