Cerebrospinal fluid flow in small-breed dogs with idiopathic epilepsy observed using time-spatial labeling inversion pulse images: a preliminary study

Abstract

Cerebrospinal fluid (CSF) circulation diseases, such as hydrocephalus and syringomyelia, are common in small-breed dogs. In human patients with CSF circulation diseases, time-spatial labeling inversion pulse (time-SLIP) sequence performed to evaluate CSF flow before and after treatment allows visualization of the restoration of CSF movement. However, studies evaluating CSF flow using the time-SLIP method in small-breed dogs are limited. Therefore, the present study aimed to evaluate intracranial CSF flow on time-SLIP images in small-breed dogs with idiopathic epilepsy, as an alternative model to healthy dogs. Time-SLIP images were obtained at two sites: 1) the mesencephalic aqueduct (MA) area (third ventricle, MA, and brain-base subarachnoid space [SAS]) and 2) the craniocervical junction area (fourth ventricle, brainstem, and cervical spinal cord SAS) to allow subsequent evaluation of the rostral and caudal CSF flow using subjective and objective methods. In total, six dogs were included. Caudal flow at the MA and brain-base SAS and rostral flow in the brainstem SAS were subjectively and objectively observed in all and 5/6 dogs, respectively. Objective evaluation revealed that a significantly smaller movement of the CSF, assessed as the absence of CSF flow by subjective evaluation, could be detected in some areas. In small-breed dogs, the MA, brain-base, and brainstem SAS would be appropriate areas for evaluating CSF movement, either in the rostral or caudal flows on time-SLIP images. In areas where CSF movement cannot detected by subjective methods, an objective evaluation should be conducted.

The “pulsatile flow” theory indicates that cerebrospinal fluid (CSF) flow is generated by cardiac pulsation and respiratory motion [8, 9, 18, 25, 28]. The intracranial blood volume increases because of increased blood flow into the calvarium during cardiac systole or decreased venous return during expiration. The intracranial volume remains constant owing to cranial stiffness. To maintain equilibrium, the CSF flows out of the intracranial subarachnoid space (SAS) into the cervical SAS, compensating for the increase in intracranial blood. Subsequently, the CSF in the ventricular system and spinal SAS generally flows caudally. Conversely, the intracranial blood volume decreases because of the increasing blood flow out of the calvarium during cardiac diastole or increasing venous return during inspiration. Alternatively, CSF can flow into the intracranial SAS from the cervical SAS; subsequently, CSF in the ventricular system and spinal SAS generally flows rostrally. This pulsating CSF movement is termed the “to-and-fro motion [25],” which can be visualized in both humans and dogs using CSF imaging such as phase-contrast (PC) and time-spatial labeling inversion pulse (time-SLIP) sequences on magnetic resonance imaging (MRI) [1, 4, 12,13,14, 18, 20, 25,26,27,28]. PC allows a quantitative assessment of CSF flow, such as the measurement of stroke volume, along with a qualitative assessment of the presence of CSF flow during the cardiac cycle (e.g., 0.5 sec when the heart rate is 120 beats/min) [18, 27]. In contrast, although time-SLIP only enables a qualitative evaluation of CSF flow, it can also enable the visualization of CSF flow for up to 5 sec in humans, thereby providing information about CSF dynamics, even in regions of slow-flowing CSF [27]. The heart rate of dogs is higher than that of humans, while the velocity of the CSF flow in dogs is reportedly lower than that in humans [6]. As such, time-SLIP may be more suitable for CSF flow observation in dogs, as it allows for longer periods of CSF flow observation compared to PC [27].

Time-SLIP CSF imaging is a non-contrast MRI technique used to visualize CSF flow using the CSF itself as an internal tracer [26, 27]. During this technique, all signals in the field of view must first be inverted using a nonselective inversion recovery (IR) pulse, after which they are suppressed. Second, only the signals within an arbitrary user-selected area (tagged area, tag) are inverted using a selective IR pulse (Tag pulse) and restored. Consequently, the signals of the CSF within the tag appear bright, whereas signals of the tissue and CSF outside the tag appear dark. Thereafter, images can be obtained following the inversion time (TI). If the CSF originating in the tag flows out of the Tag during TI (flow-out), the CSF with a bright signal can be easily visualized outside the tag against the dark signal of the background. Conversely, if the CSF originating outside the tag flowed into the Tag during TI (flow-in), the CSF with dark signals appear to flow into the CSF with a bright signal inside the tag. Images can then be repeatedly obtained at different TIs, allowing for the visualization of CSF displacement. For example, the CSF has been shown to flow rostrally and caudally in the mesencephalic aqueduct (MA) and ventral brainstem SAS in healthy humans on mid-sagittal head time-SLIP images [26, 28].

This movement of the CSF in the ventricular system has not been observed in some human patients with hydrocephalus. As such, the presence of CSF flow is confirmed using time-SLIP in human patients with CSF circulation disorders, thereby enabling a better understanding of its pathophysiology [1, 26, 27]. Additionally, some studies used this technique to confirm the restoration of CSF flow following surgery for human patients with hydrocephalus, although the flow was not delineated prior to the procedure [1, 26, 27]. As such, time-SLIP for CSF imaging is expected to be useful in evaluating the effects of treatment. Similar loss and restoration of CSF flow before and after surgical treatment have been reported in a single case report of a dog with congenital hydrocephalus [12], indicating that time-SLIP for CSF imaging may also be useful for the assessment of CSF circulation disorders as well as the evaluation of treatment response in dogs.

To date, only one study and a few isolated case reports evaluating CSF flow using time-SLIP in dogs have been published [4, 12,13,14]. In one of these studies, Cho et al. assessed CSF flow in healthy beagle dogs using midsagittal time-SLIP images, and concluded that the CSF flows caudally in the MA and dorsal and ventral cervical SAS [4]. In the above case report [12], the “to-and-fro motion” of CSF flow in the MA and ventral brainstem SAS have been identified, in addition to the flows demonstrated by Cho et al. As such, we hypothesized that a CSF flow other than that verified by Cho et al. could be identified in dogs, as in humans, using the time-SLIP technique.

Prior to investigating the changes in CSF flow in dogs with CSF circulation disorders using time-SLIP images, it is crucial to understand how CSF flow is visualized in healthy dogs. One prior study on the visualization of CSF flow in healthy dogs included only beagles [4]. However, diseases due to the impairment of CSF circulation, such as congenital hydrocephalus and syringomyelia (SM) associated with Chiari-like malformations (CM), are more common in small-breed dogs [22, 23]. Consequently, small-breed dogs are considered more suitable candidates than beagles for the visualization of CSF flow using time-SLIP in healthy subjects. Healthy dogs were eligible for inclusion in the normal model. However, from the perspective of animal ethics, minimizing the use of laboratory dogs is recommended. In another study in humans, functional MRI showed no difference in CSF movement in the fourth ventricle between epileptic patients and healthy humans [15]. This suggests that the CSF flow in the ventricles and SAS in dogs with idiopathic epilepsy would be the same as that in normal dogs; as such, dogs with idiopathic epilepsy could serve as surrogate models for normal dogs.

In this study, we aimed to visualize the CSF flow in various CSF spaces, including the third and fourth ventricles, MA, ventral brainstem SAS, and dorsal and ventral cervical SAS of small-breed dogs with idiopathic epilepsy, as an alternative model of healthy dogs, using time-SLIP, and to assess visualization of the CSF flow in each CSF space subjectively and objectively.

MATERIALS AND METHODS

All study procedures were reviewed and approved by the Ethics Committee of Nihon University Animal Medical Center (Approval Number 0-013). Written informed consent was obtained from the owners of each dog prior to enrollment in the study.

This study enrolled client-owned, small-breed dogs that presented with epileptic seizures and were diagnosed with idiopathic epilepsy based on neurological examinations and MRI findings between September 2013 and August 2014 at the Nihon University Animal Medical Center. Small-breed dogs were defined as those weighing <10 kg, as in previous reports [5, 21]. For each dog, data on age, breed, sex, body weight, and heart rate and mean arterial pressure during the time-SLIP examination were collected. If abnormalities such as decreased hindlimb proprioception were found on neurological examination, the dog was excluded because of possible impaired CSF circulation [3]. In addition, if abnormal MRI and/or CT findings, including CM, craniocervical junction abnormalities (CJA), and/or SM, were found, the dog was excluded because of the potential for the impairment of CSF circulation [22]. The definitions of CM, CJA, and SM were based on a previous report [16].

All MRI examinations were performed under general anesthesia and in sternal recumbency using a knee coil (QD knee coil, Canon Medical Systems, Ootawara, Japan) and a 1.5-Tesla MRI scanner (EXCELART Vantage, Canon Medical Systems). For anesthesia, the patients were premedicated using atropine (0.04 mg/kg, subcutaneously) (atropine sulfate injection 0.5 mg; NIPRO ES PHARMA, Osaka, Japan) and diazepam (cercine injection 10 mg; Teva Takeda Pharma Ltd., Osaka, Japan) (0.2 mg/kg, intravenously), induced with propofol (6–7.9 mg/kg, intravenously) (propofol intravenous injection 1% “Maruishi”; Maruishi Pharmaceutical Co., Ltd., Osaka, Japan), and maintained with isoflurane (1.5–2%) (ds isoflurane; Bussan Animal Health Co., Ltd., Osaka, Japan).

Sagittal, transverse, and dorsal, T2-weighted (TR, 3.5–4.5 sec; TE, 0.105 sec), T1-weighted (TR, 0.45–0.55 sec, TE, 0.015 sec), and fluid-attenuated inversion recovery (TR, 10 sec; TE 0.108 sec) images of the head and neck including the C2 vertebra were initially obtained. If no structural abnormalities of the brain were confirmed on these images, time-SLIP images at the two sites were subsequently obtained, as follows: 1) the MA time-SLIP images, where the tag was located in the MA area, and 2) the craniocervical junction (CCJ) time-SLIP images, where the tag was located in the CCJ area. The time-SLIP sequence parameters were as follows: TR, 9.6 sec; TE, 0.080 sec; field-of-view, 13 × 13 cm; matrix, 160 × 192; slice thickness, 4 mm; and tag width, 10 mm. The TI was then increased 30 times in increments of 0.050 sec, starting at 1.5 sec. The acquisition time for each time-SLIP sequence was approximately 4 min 30 sec. Figure 1 shows the details of the method used to set the time-SLIP images. Subsequently, the positions of the slices for the time-SLIP images were adjusted using mid-sagittal T2-weighted images. The slice of the time-SLIP image was placed at the same location as mid-sagittal T2-weighted images. The tag slice for the MA time-SLIP image was subsequently set at the plane perpendicular to the MA. Similarly, the tag slice for CCJ time-SLIP images was set at a plane perpendicular to the brainstem. Sagittal, transverse, and dorsal postcontrast T1-weighted images were obtained to confirm the absence of abnormalities after the time-SLIP sequences.

Fig. 1.

Mid-sagittal T2-weighted (A, C) and time-spatial labeling inversion pulse (B, D) images of a 4.3-kg, 7-year-old, Japanese Terrier dog. (A) The mesencephalic aqueduct (MA) time-spatial labeling inversion pulse (time-SLIP) image was set based on mid-sagittal T2-weighted image (A). The plane of the MA time-SLIP image was put in the same location as (A) the mid-sagittal plane. The tagged area (tag) on the MA time-SLIP image (the area surrounded by the yellow open box) was set on the plane perpendicular to the sagittal plane (the point where the caudal edge of tag overlapped the rostral part of MA), at an angle perpendicular to MA (red dotted line). (B) Actual time-SLIP image. On the image, signals outside the tag were suppressed by non-selective inversion pulse (black). Inversely, signals of cerebrospinal fluid (CSF) inside the tag were restored by selective inversion pulse, and were consequently shown as hyperintensity (white). (C) The craniocervical junction (CCJ) time-SLIP image was also set based on mid-sagittal T2-weighted image (C). The plane of the CCJ time-SLIP image was placed in the same location as (C), in the mid-sagittal plane. The tag on the CCJ time-SLIP image (the area surrounded by the green open box) was set at the plane perpendicular to the sagittal plane, the point where the caudal edge of the tag contacted with the most caudal part of the cerebellum, and an angle perpendicular to the red line, the point (closed red circle, a) where the rostral edge of the tag intersects the dorsal edge of the medulla oblongata, and connects with the point (open red circle, b) where the caudal edge of the tag intersects the dorsal edge of the medulla oblongata at the level of the most caudal part of the cerebellum. (D) Actual time-SLIP image. Only CSF signals inside the tag are shown as hyperintensity.

Figure 2 summarizes the details of the observation points. The presence of rostral and caudal CSF flows were evaluated at each observation point. In addition, when both rostral and caudal CSF flow was identified at one observation point, the “to-and-fro motion” of the CSF was confirmed to be present. In general, when evaluating the presence of CSF flow on time-SLIP images, doctors and veterinarians must subjectively determine the presence of CSF flow by grossly observing signal changes inside and outside the tag on the images [26]. Subjective methods require shorter evaluation times than objective methods [2]; however, they may overlook subtle changes, potentially resulting in inter-rater disagreements [2]. Hence, to objectively evaluate the presence of CSF flow on the images, the distance travelled by the CSF on the image was measured as the flow distance, as described in a previous human report [1], while the results of subjective and objective evaluations were compared.

Fig. 2.

(A) Schema for the mid-sagittal image of the same dog as shown Fig. 1. Location of the tagged area (tag) on the mesencephalic aqueduct (MA) time-spatial labeling inversion pulse (time-SLIP) image is presented. Blue and red lines indicate the rostral and caudal edge of the tag, respectively; as such, the tag is the area between the blue and red lines. (B) Enlargement of the area surrounded by the green box in (A). Roman numbers indicate each observation point of cerebrospinal fluid (CSF) flow. At each observation point, the presence of rostral and caudal flows was evaluated. White arrows indicate CSF flowing out of the tag (flow-out), while black arrows indicate CSF flowing into the tag (flow-in). (C) The anatomical locations of each observation point on the MA time-SLIP image are presented. (D) Schema for the mid-sagittal image of the same dog as in Fig. 1. Location of tag on the craniocervical junction (CCJ) time-SLIP image is shown. The tag is shown using the same-colored lines as (A). (E) Enlargement of the area surrounded by the yellow green box in (D). Roman numbers indicate each observation point of the CSF flow. White and black arrows indicate the flow-out or flow-in, respectively, as in (B). (F) Anatomical locations of each observation point on CCJ time-SLIP image are presented. *SAS: subarachnoid space.

To evaluate the presence of CSF flow subjectively, two neurologists (M.K. and D.I.) and a veterinarian (C.I.), all blinded to the dogs’ clinical profiles, independently reviewed all images. A series of 30 SLIP images were sequentially displayed (Supplementary Files 1 and 2). If a white signal appeared more than twice outside the tag or a black signal appeared more than twice inside the tag at the observation point on a total of 30 time-SLIP images, the CSF flow was considered present (Fig. 3, Supplementary Files 1 and 2). If all reviewers confirmed that “CSF flow was observed,” only then was the presence of CSF flow recorded to prevent overestimation of the flow evaluation. An objective evaluation of the CSF flow distance was conducted by a veterinarian (N. S.) blinded to the subjective evaluation results. All images were displayed frame by frame at each observation point, while the rostral and caudal flow distances were measured separately. The CSF flow distance was defined as the distance of the “CSF flow out,” i.e., the distance from the tag edge to the tip of the bright signal outside the tag, and the distance of “CSF flow in,” i.e., the distance from the tag edge to the tip of the dark signal inside the tag (Supplementary File 3). When measuring the “CSF flow out,” the tag was masked, and distances were measured if bright signals appear outside the masked tag. When measuring the “CSF flow in,” the background outside the tag was masked. The distances were measured if dark signals appeared inside the tag. The measurement was repeated thrice, and the mean value was recorded as the flow distance. If the distance was recorded more than twice on 30 images, it was evaluated as the presence of flow. All time-SLIP images were reviewed using a commercially available workstation (Virtual Place; Canon Medical Systems).

Fig. 3.

Sequential series of the mesencephalic aqueduct (A–C) and the craniocervical junction (D–F) time-spatial inversion pulse images of a 2.8-kg, 9-year-old Yorkshire Terrier dog. On each panel, blue and red lines indicate the rostral and caudal edges of the tagged area (tag). (A) Roman numbers indicate the observation points of cerebrospinal fluid (CSF) flow on the mesencephalic aqueduct (MA) time-spatial labeling inversion pulse (time-SLIP) images [I: third ventricle, II: subarachnoid space (SAS) ventral to the frontal lobe of the cerebrum, III: MA, IV: ventral brainstem SAS]. The CSF inside the tag is shown as hyperintensity (white), while those outside the tag (non-tagged area) are shown as hypointense (black). (B, C) CSF outside the tag (indicated by white arrows) is shown as white, compared to A, which indicates that the tagged CSF flows out of the tag during inversion time (TI) (flow-out). The CSF inside the tag (indicated by black arrows) is shown in black compared to A, indicating that non-tagged CSF flows into the tag during TI (flow-in). The signals of the CSF in the remaining observation points do not change compared to A, indicating that the CSF in these areas does not flow during TI. (D) Roman numbers indicate the observation points of CSF flow on the craniocervical junction time-SLIP images [V: fourth ventricle, VI: ventral brainstem SAS, VII: dorsal cervical SAS, VIII: ventral cervical SAS]. (E, F) White or black arrows indicate flow-out or flow-in CSF flow, as (B, C), respectively.

Intraclass correlation coefficients (ICC) were calculated to evaluate the inter-rater agreement among the subjective evaluations by the three reviewers and intra-rater agreement among repeated objective measurements by a single measurer. Inter- and intra-rater reliabilities were further determined based on a 95% confidence interval (CI) of the estimated ICC. As previously described [17], an ICC >0.75 indicated a high level of agreement for the current study. Further, kappa coefficients were calculated to evaluate the agreement between the subjective and objective assessments. As previously described [19], a kappa coefficient >0.61 indicated substantial agreement for the current study. If there was disagreement between the subjective and objective assessments, all flows were classified into six subgroups to probe the causes of disagreement: SP/OP (flow was both subjectively and objectively assessed as present), SP/OA (flow was subjectively assessed as present, but objectively assessed as absent), Sa/OP (flow was subjectively assessed as absent due to disagreement of evaluations among reviewers, while flow was objectively assessed as present), Sa/OA (flow was subjectively assessed as absent due to a disagreement of evaluations among reviewers, and was objectively assessed as absent), SA/OP (flow was subjectively assessed as absent due to agreement of evaluations among reviewers, but objectively assessed as present), and SA/OA (flow was assessed as absent due to agreement among reviewers, and objectively assessed as absent). The mean, maximum, and minimum distances among the SP/OP, Sa/OP, and SA/OP groups were compared using the Kruskal–Wallis test. The SP/OA, Sa/OA, and SA/OA groups, in which flows were objectively assessed as absent, were excluded from the analysis because the flow distances could not be measured. Statistical significance was set at P<0.05. All statistical analyses were performed using commercially available software (SPSS Statistics, version 23.0.0.0, 64-bit edition; IBM Corp., Armonk, NY, USA; and GraphPad Prism 5, GraphPad Software Inc., La Jolla, CA, USA).

RESULTS

During the study period, 146 dogs underwent head MRI examinations, while 116 were excluded for various reasons, such as deviation from the inclusion criteria (e.g., body weight). Although the remaining 30 dogs were recruited for the study, 24 were excluded because of abnormal MRI and/or CT findings. Finally, six dogs were enrolled in the study, including one each of a toy poodle, Miniature Dachshund, Pomeranian, Yorkshire Terrier, Japanese Terrier, and a mixed breed. The study population comprised five males, of which three were intact and two were neutered, and 1 spayed female. The dogs had a median age of 64 months (range, 46–109 months) and weighed a median of 4.3 (1.5–6.5) kg. The median heart rate and mean arterial pressure at acquisition of time-SLIP images were 104.5 (85–153) beats/min and 73 (65–85) mmHg, respectively.

The assessment results of CSF flow at each observation point are summarized in Tables 1 and 2 and Supplementary Files 4 and 5. The third ventricle (observation point I) was excluded from the CSF flow evaluation because all areas of the third ventricle were included in the tag, meaning that the CSF signal change could not be appreciated, in three of the six dogs. Caudal flow in ventral brainstem SAS on MA time-SLIP images was “present” in all dogs both subjectively and objectively. Caudal flow in the MA on MA time-SLIP images and rostral flow in ventral brainstem SAS on CCJ time-SLIP images were “present” both subjectively and objectively in 5 of 6 dogs. CSF flow at these two observation points was not detected in dog 2 (Supplementary File 4). Additionally, rostral flow in the MA and ventral brainstem SAS on MA time-SLIP images and caudal flow in ventral cervical SAS on CCJ time-SLIP images were objectively “present” in 5 of 6 dogs. CSF flow at these three observation points was not detected in dogs nos. 5, 2, and 6 (Supplementary File 4). The “to-and-fro motion” was objectively confirmed in 5 of the 6 dogs only in the ventral brainstem SAS (observation point IV) on MA time-SLIP images (Table 3). In the MA (observation point III) on MA time-SLIP images, the “to-and-fro motion” was confirmed in 4 of the 6 dogs, although rostral or caudal flow in this area was objectively assessed as present in 5 of the 6 dogs, considering that rostral or caudal flow was only observed in 2 different dogs.

Table 1. Subjective and objective assessments of cerebrospinal fluid flow on time-spatial labeling inversion pulse images.

Site of Time-SLIP image	Observation point of the CSF flow a)		Subjective assessment		Objective assessment
			Presence	Absence b)	Presence
MA	I c)	1	NE	NE	NE
		1’	NE	NE	NE
	II	2	0/6	6/6 (0)	0/6
		2’	2/6	4/6 (0)	2/6
	III	3	5/6	1/6 (0)	5/6
		3’	3/6	3/6 (1)	5/6
	IV	4	6/6	0/6 (0)	6/6
		4’	3/6	3/6 (3)	5/6
CCJ	V	5	0/6	6/6 (0)	0/6
		5’	0/6	6/6 (3)	3/6
	VI	6	3/6	3/6 (0)	3/6
		6’	5/6	1/6 (0)	5/6
	VII	7	3/6	3/6 (0)	4/6
		7’	0/6	6/6 (0)	0/6
	VIII	8	4/6	2/6 (0)	5/6
		8’	0/6	6/6 (1)	0/6

a) I: third ventricle; II: SAS ventral to the frontal lobe of the cerebrum; III: MA; IV: ventral brainstem SAS; V: fourth ventricle; VI: ventral brainstem SAS; VII: dorsal cervical SAS; VIII: ventral cervical SAS. At each point, the caudal and rostral flows were observed as Arabic and apostrophized Arabic numbers, respectively. b) Values in parentheses indicate the number of flows assessed as absent due to disagreement of evaluations among reviewers. c) In 3/6 dogs, the presence of CSF flow in the third ventricle (observation point I) was not evaluated because all areas of the third ventricle were included in the tag. Therefore, the third ventricle was excluded from the CSF flow evaluation. CCJ, the craniocervical junction; CSF, cerebrospinal fluid; MA, the mesencephalic aqueduct; NE, not evaluated; SAS, the subarachnoid space; time-SLIP, time-spatial labeling inversion pulse.

Table 2. Cerebrospinal fluid flow distances on time-spatial labeling inversion pulse images.

Site of Time-SLIP image	Observation point of CSF flow a)		Presence	Flow distance (mm) b)
				Mean	Maximum	Minimum
MA	I c)	1	NE	NE	NE	NE
		1’	NE	NE	NE	NE
	II	2	0/6	-	-	-
		2’	2/6	1.47 (0.88, 2.06)	2.01 (0.96, 3.07)	0.66 (0.51, 0.81)
	III	3	5/6	1.24 (0.98, 4.21)	2.22 (1.13, 4.61)	0.86 (0.41, 3.55)
		3’	5/6	1.06 (0.82, 1.95)	1.45 (0.99, 2.14)	0.78 (0.41, 1.81)
	IV	4	6/6	2.43 (1.07, 3.29)	2.96 (1.49, 7.52)	1.47 (0.77, 1.88)
		4’	5/6	1.36 (0.72, 4.07)	3.80 (0.78, 5.74)	0.77 (0.68, 0.88)
CCJ	V	5	0/6	-	-	-
		5’	3/6	0.65 (0.48, 1.51)	0.78 (0.48, 1.69)	0.51 (0.48, 1.33)
	VI	6	3/6	0.64 (0.62, 5.60)	0.88 (0.85, 5.79)	0.47 (0.46, 5.37)
		6’	5/6	2.79 (1.38, 7.60)	4.23 (2.09, 8.52)	1.32 (0.90, 6.43)
	VII	7	4/6	0.71 (0.62, 0.95)	0.81 (0.65, 1.13)	0.62 (0.60, 0.80)
		7’	0/6	-	-	-
	VIII	8	5/6	0.83 (0.70, 1.42)	0.99 (0.72, 1.64)	0.74 (0.49, 1.17)
		8’	0/6	-	-	-

a) I: third ventricle; II: SAS ventral to the frontal lobe of the cerebrum; III: MA; IV: ventral brainstem SAS; V: fourth ventricle; VI: ventral brainstem SAS; VII: dorsal cervical SAS; VIII: ventral cervical SAS. At each point, the caudal and rostral flows were recorded as Arabic and apostrophized Arabic numbers, respectively. b) Data are expressed as median values with minimum and maximum values in parentheses. c) In 3/6 dogs, the presence of CSF flow in the third ventricle (observation point I) was not evaluated because all areas of the third ventricle were included in the tag. Therefore, the third ventricle was excluded from the CSF flow evaluation. CCJ, the craniocervical junction; CSF, cerebrospinal fluid; MA, the mesencephalic aqueduct; NE, not evaluated; SAS, the subarachnoid space; time-SLIP, time-spatial labeling inversion pulse.

Table 3. Assessment of “to-and-fro motion” on time-spatial labeling inversion pulse images.

Site of Time-SLIP	Assessment Method	MA				CCJ
Observation point of CSF flow a)		I b)	II	III	IV	V	VI	VII	VIII
Presence of “to-and-fro motion”	Subjective	NE	0	2	3	0	3	0	0
	Objective	NE	0	4	5	0	3	0	0

a) I: third ventricle; II: SAS ventral to the frontal lobe of the cerebrum; III: MA; IV: ventral brainstem SAS; V: fourth ventricle; VI: ventral brainstem SAS; VII: dorsal cervical SAS; VIII: ventral cervical SAS. b) In 3/6 dogs, the presence of CSF flow in the third ventricle (observation point I) was not evaluated because all areas of the third ventricle were included in the tag. Therefore, the third ventricle was excluded from the CSF flow evaluation. CCJ, the craniocervical junction; CSF, cerebrospinal fluid; MA, the mesencephalic aqueduct; NE, not evaluated; SAS, the subarachnoid space; time-SLIP, time-spatial labeling inversion pulse.

In the nine flows, the subjective and objective assessments did not align (Table 4A). Details of CSF flow assessments in these subgroups are provided in Table 4B. Among the nine flows with discordant evaluations, four were classified into the Sa/OP group (flow was subjectively assessed as absent owing to disagreement of evaluations among reviewers, while flow was objectively assessed as present), and five were classified into the SA/OP group (flow was subjectively assessed as absent owing to agreement of evaluation among reviewers, while flow was objectively assessed as present). The ICC for the inter- or intra-rater reliability of the subjective or objective CSF flow evaluation was 0.885 or 0.999 (95% CI: 0.843–0.918 or 0.999–0.999, respectively), indicating good reliability of both the subjective and objective evaluation of CSF flow. The kappa coefficient was 0.787, indicating a substantial agreement between the subjective and objective evaluations. The mean, maximum, and minimum CSF flow distances in the three groups (SP/OP, Sa/OP, and SA/OP) are summarized in Fig. 4. Overall, significant differences were observed in the mean and maximum flow distances among the three groups (P=0.02, 0.01 respectively). The mean and maximum flow distances in the SP/OP group were significantly larger than those in the SA/OP group (P<0.05, 0.001 respectively). The maximum flow distance in the Sa/OP group was also significantly greater than that in the SA/OP group (P<0.05).

Table 4. Agreement between subjective and objective assessments.

(A) Whole
		Objective assessment
		Presence	Absence
Subjective assessment	Presence	34	0
	Absence	9	41
(B) Subgroup
		Objective assessment
		OP	OA
Subjective assessment	SP	34	0
	Sa	4	3
	SA	5	38

OA: flow was objectively assessed as absent, OP: flow was objectively assessed as present, SA: flow was subjectively assessed as absent due to agreement of evaluations among reviewers, Sa: flow was subjectively assessed as absent due to disagreement of evaluations among reviewers, SP: flow was subjectively assessed as present.

Fig. 4.

Mean (A), maximum (B), and minimum (C) flow distance of cerebrospinal fluid on time-spatial labeling inversion pulse images. The mean and maximum flow distances were significantly greater in the SP/OP group than those in the SA/OP group. The maximum flow distance was significantly greater in the Sa/OP group than that in the SA/OP group. The flow was subjectively assessed as absent due to agreement of the evaluations among reviewers, while flow was objectively assessed as present in the SA/OP group; the flow was subjectively assessed as absent due to disagreement of evaluations among reviewers, while flow was objectively assessed as present in the Sa/OP group; and the flow was subjectively and objectively assessed as present in the SP/OP group. *: P<0.05, **: P<0.01.

DISCUSSION

Caudal flow in the MA and ventral brainstem SAS on MA time-SLIP images and rostral flow in the ventral brainstem SAS on CCJ time-SLIP images were subjectively and objectively identified in all and 5/6 dogs, respectively. These results indicate that the time-SLIP method may be useful for evaluating CSF flow in these areas in small-breed dogs. Rostral flows in the MA and ventral brainstem SAS on MA time-SLIP images and caudal flow in the ventral cervical SAS on CCJ time-SLIP images were objectively identified in 5/6 dogs, although the flows were subjectively evaluated as present in 3/6, 3/6, or 4/6 dogs, respectively. Statistical analysis further revealed that subjective methods cause errors in the evaluation of CSF flow when the movements of the CSF are smaller than those of the CSF that are subjectively and objectively identified. Thus, if the movements of the CSF cannot be clearly identified subjectively, then the flow should be evaluated using an objective method. Our results also indicate that the ventral brainstem SAS (observation point IV) may be suitable for observing a “to-and-fro motion” of CSF on time-SLIP images in small-breed dogs, as this motion was identified in 5/6 dogs. Caudal flow in the SAS ventral to the frontal lobe of the cerebrum on MA time-SLIP images and caudal flow in the fourth ventricle and rostral flows in the dorsal and ventral cervical SAS on CCJ time-SLIP images could not be subjectively or objectively identified in any dog. As such, visualizing CSF flow in these regions in small-breed dogs using the time-SLIP method in the current study may be difficult.

The agreement between the subjective and objective evaluations of CSF flow was comparatively high. The subjective evaluation would be considered reliable when the flow was subjectively assessed to be “present,” as all flows identified subjectively were confirmed objectively. Conversely, some CSF flows were subjectively assessed as absent, but objectively assessed as present. The mean and maximum flow distances in the SA/OP group were overall significantly shorter than those in the SP/OP group. The small movement of the CSF in the SA/OP group could not be grossly confirmed, and was subjectively evaluated as the absence of flow. The maximum flow distance was significantly greater in the Sa/OP group than in the SA/OP group, although the mean flow distance showed no significant difference between the two groups. The reviewers who recognized maximum movement could have assessed this as “CSF is flowing,” whereas those who did not recognized such movement could have assessed this as “CSF is not flowing,” thereby resulting in disagreements in evaluation among reviewers, as subtle movements may be missed by subjective assessment [2]. In the Sa/OA group, there were three points where one of the reviewers subjectively assessed CSF flow, although the flow was objectively assessed as absent. As shown in Supplementary Files 1 and 2, the head position moved slightly when a series of 30 time-SLIP images were displayed sequentially. This movement of the head would be generated by thoracic motion during respiration, and the reviewers may have assessed this slight oscillation as a CSF movement. As such, there is a possibility that presence of CSF flow could be misjudged when using a subjective method. In the current study, subjective evaluation by consensus among multiple reviewers may have higher reliability than evaluation by a single reviewer; however, to ensure a more accurate evaluation of CSF flow on time-SLIP images, the flow should be objectively evaluated in situations in which the tip of the CSF movement on time-SLIP images is not apparent during subjective evaluation.

The mean value of the maximum distance of caudal flow in MA in the small-breed dogs in this study (2.55 mm) was smaller than the mean value of maximum flow distance from MA to the fourth ventricle in healthy humans, as identified in a prior report (22.8 mm) [1]. The intracranial volume of humans is approximately 1,130–1,400 cm³ [7, 24], while the mean intracranial volume of the 6 dogs in this study, as calculated from CT head images, was 52.11 cm³. The flow distance/intracranial volume was calculated as the relative flow distance, with values of 0.016–0.020 for humans and 0.049 for dogs, indicating a larger value for dogs than for humans. The CSF flow distance in dogs is only diminished compared to that in humans due to the small size of the canine brain. A higher spatial resolution is necessary for visualizing CSF flow in a small space. However, further improvements have been constrained by equipment limitations. Equipment with a higher magnetic field may enable researchers to observe CSF flow at more observation points, as this could provide a higher spatial resolution.

Based on the results of a human functional MRI study [15], in the present study, we only included dogs with idiopathic epilepsy as an alternative model of healthy dogs without impairment of CSF circulation. However, unlike in human patients, CSF circulation in these dogs could not be normal. Although the age at initial epileptic seizure onset in dogs with idiopathic epilepsy is commonly less than 6 years [11], the two dogs with epilepsy in this study were over 6 years old. The onset of epileptic seizures in one of the two dogs was at <6 years of age. And in both dogs epileptic seizures were successfully controlled using long-term antiepileptic medication. Approximately 32% of dogs diagnosed with idiopathic epilepsy are over 6 years old [10]. These support the diagnosis of idiopathic epilepsy in both 2 dogs.

CM, CJA, and SM were present in approximately half of the dogs recruited for the current study, resulting in the inclusion of a small number of dogs. In a previous study on Chihuahuas [16], CM was observed in all included dogs, regardless of whether they had associated symptoms. Thus, small-breed dogs without CM, CJA, or SM may be less common than expected.

CSF flow is generated by cardiac pulsation and respiratory motion [8, 9, 18, 25, 28]; however, the time-SLIP used in the present study did not synchronize with the motions for acquiring data. If it becomes possible to synchronize with the motions and acquire data during the maximum point of CSF movement, it is conceivable that larger CSF movements could be visualized, thereby increasing the number of points at which the CSF flow could be visualized.

Caudal flow in the MA and ventral brainstem SAS on MA time-SLIP images and rostral flow in the ventral brainstem SAS on CCJ time-SLIP images were subjectively and objectively identified in all or most of the dogs. Considering the results of the statistical analysis, the movements of the flows in these areas were larger than those of flows that were evaluated as absent by the subjective method but present by the objective method. As such, in small-breed dogs, flows in these areas could be clearly and reliably identified using a subjective method. However, CSF flow should be evaluated using objective methods if CSF movement on time-SLIP images cannot be clearly identified by subjective evaluation.

CONFLICTS OF INTEREST

None of the authors have any conflicts of interest to declare.