Impact of Hybrid Subtraction CTA

Recently, many bone removal techniques, including direct subtraction CTA, and MMBE CTA, were developed for diagnosing cerebral aneurysms. In the algorithm of MMBE CTA, the transosseous or intraosseous vessels will be masked and become poorly visualized because partial volume effect and the dilation operation could change the vessel pixels (which touching the bone) into mask during processing. MMBE CTA

3

could obliterate minor mis-registration artifacts within the bone area. If registration was not perfect, the result of MMBE CTA would look better than subtraction CTA. By using direct subtraction CTA, one can obtain an image of the existence of contrast material (similar to DSA) that can be used for diagnosis. However, the contrast-to-noise ratio decreases, especially when a low-dose precontrast scan is used for masking.

MMBE CTA removes bone, as well as adjacent tissues, which are frequently affected by certain vascular disorders. Vessels traversing in or on the bones may also be eliminated. Our hybrid technique can be used to perform subtraction in the masked area and can keep postcontrast image data out of the masked area. (Figure 1-4, Figure 1-5)

Hybrid CTA has the advantage of demonstrating the transosseous or intraosseous vessels. (Figure 1-6) This makes diagnosis of vascular diseases in or adjacent to the bones possible. Hybrid CTA can provide a rapid diagnosis in many clinical situations because of its quick acquisition and non-invasive nature.

doi:10.6342/NTU201603776 Figure 1-1: Hybrid bone subtraction without registration (a) and after registration (b).

Mis-registration artifacts decreased significantly.

5

Figure 1-2: Thick maximal intensity projection (MIP) of subtraction CTA (a) without bone registration (b) with bone registration. The image quality improves after bone registration.

doi:10.6342/NTU201603776 Figure 1-3: Generation of hybrid bone subtraction CTA after registration. (a)

Postcontrast image (CTA). (b) Precontrast image. (c) Bone mask generated from precontrast image by threshold (>150HU) and 1-pixel dilatation. (d) MMBE image from CTA with soft tissue value (40HU) in bone mask. (e) Hybrid bone subtraction CTA with simple subtraction in the mask, resulting in lower density in the bone. (f) Hybrid bone subtraction CTA with soft tissue compensation to achieve an even background.

7

Figure 1-4: Advantage of hybrid bone subtraction CTA (hybrid CTA) (a) unprocessed CTA image showed early enhancing isolated left transverse sinus (large arrow) and multiple engorged medullary veins in left temporal lobe and left cerebellum (surrounded by small arrows). (b) subtraction CTA increased noises in the brain parenchyma, the engorged medullary veins were masked by noise. (c) MMBE CTA masked the information in and adjacent to bones. (d) hybrid CTA preserved the information in brain parenchyma, and added information in the bones (large arrow).

doi:10.6342/NTU201603776 Figure 1-5: Images in 60-year-old man with pulsatile tinnitus. Images were obtained

with (a) MMBE, (b) direct subtraction, and (c) hybrid subtraction (hybrid CTA). (b, c) Arrow = prominent transosseous arteries supplying the arteriovenous fistula. Tiny channels in the bone can be seen; these arteries are not seen on (a). Higher image noise can be seen on (b). In this case, hybrid CTA on (c) outlined the abnormal vessels best.

9

Figure 1-6: Maximal intensity projection (MIP) of hybrid CTA in frontal projection (a) and lateral projection (b). Lateral projection of right carotid angiography at early arterial phase (c) and late arterial phase (d). Black arrow indicated the occlusion point of proximal internal carotid artery (ICA). White arrows indicated reconstitution of petrous segment of ICA, which was clearly seen on hybrid CTA.

doi:10.6342/NTU201603776 Table 1-1: Acquisition Protocols of CT scanners

CT scanner Collimation (mm)

Pitch Rotation (sec)

KVp (kV)

Tube Current (mA) C-/C+

GE Lightspeed VCT

(before 2010) 32x0.625 0.969 0.4 100 200/400

GE Lightspeed VCT

(after 2010) 64x0.625 0.984 0.4 100 200/400

Siemens Sensation 64 64x0.6 0.8 0.33 100 243/484

Philips Ingenuity CT 64x0.625 1.015 0.4 100 100/250 KVp: Kilovoltage peak. C-: precontrast. C+: postcontrast.

11

Chapter 2

Hybrid CTA in Diagnosis and Treatment Planning of Dural Arteriovenous Fistulas

2.1 Introduction

Dural arteriovenous fistulas (AVFs) occur at dural sinuses, which are proximate to or inside the skull bones. Because of the overlapping bone structures, it is difficult to demonstrate the detailed vascular pattern and to make a diagnosis of dural AVF with conventional CTA, especially in small lesions. Magnetic resonance (MR) imaging and MR angiography were thought to be more useful in the diagnosis of dural AVF [12-14], but they cannot be used in patients with a non-MR-compatible pacemaker or in uncooperative patients. To our knowledge, the use of CTA for the diagnosis of dural AVF remains uncommonly reported. A hybrid CTA can help demonstrate vessels close to or inside the skull bones that makes diagnosis of dural AVF possible. The purpose of this study was to analyze the diagnostic effectiveness and application of CTA by using hybrid CTA algorithm in dural AVFs.

2.2 Materials and Methods

2.2.1 Patient and Control Populations

This study was approved by our institutional review board. In this study, the comparison between hybrid CTA and DSA, the determination of the presence of specific imaging signs for diagnosis of dural AVFs at hybrid CTA, and the test of inter-observer agreement were performed retrospectively. However, we used hybrid

doi:10.6342/NTU201603776 CTA in treatment planning before the patients underwent endovascular treatment. From

January 1, 2008, to March 31, 2009, 167 patients under-went both cerebral CTA and DSA in National Taiwan University Hospital (Taipei, Taiwan). We divided the patients into two groups: patients with a dural AVF (n=22, 10 women and 12 men; age range, 41–77 years; mean age, 60.3 years ± 11.3) and patients without a dural AVF, the control group. The control group (n=14, six female and eight male subjects; age range, 17–77 years; mean age, 61.9 years ± 17.1) included patients who did not have any cerebrovascular abnormalities noted at DSA; patients excluded from the control group included those with cerebral arteriovenous malformation (n=3), those with aneurysm (n=59), those with direct carotid artery–cavernous fistula (n=6), those with cerebral space-occupying lesions (n=2), those with dural sinus thrombosis (n=3), those with moyamoya disease (n=6), those with a large stroke (n=48), and those who had undergone external carotid artery–internal carotid artery bypass surgery (n=4).

2.2.2 Image Evaluation

Identification information was removed from the hybrid CT angiographic source images and maximum intensity projection images, and the images were evaluated independently by two neuroradiologists who were blinded to the clinical data and results of diagnostic CT and DSA. The methods of interpretation included multiplanar reformation (MPR), maximum intensity projection (MIP), and volume rendering. The comparison between hybrid CTA and DSA, the determination of presence of imaging signs for diagnosis of dural AVFs, and the test of inter-observer agreement were performed retrospectively with high-brightness liquid crystal display monitors (3 megapixels, 1536 3 2048 native resolution, monochrome). From the final post-processed images, the readers recorded the presence and location (left side, right

13

side, or midline) of dural AVFs and the presence of imaging signs, including engorged arteries, transosseous enhanced vessels, engorged extracranial veins, engorged cortical veins, asymmetric sinus enhancement, and the associated dural sinus occlusion. We assigned grades to the dural AVFs by using the system proposed by Cognard et al [15].

DSA images were used as the reference standard.

2.2.3 Treatment Planning

The treatment was planned according to the grade of dural AVF. In our institution, transvenous embolization was the method of choice for treating dural AVFs if there was no normal cortical vein draining into the lesion sinus and the lesion sinus was accessible.

We used all available imaging data for treatment planning, including images from hybrid CTA, DSA, and MR angiography. Treatment planning with hybrid CTA was performed before the patients received endovascular treatment. We applied the imaging data to analyze the best route to approach the point of the fistula. We usually designed two routes for the transvenous approach for each case. In those patients with an occluded sinus, we always tried to cross the occluded access site. If the transvenous approach failed or preservation of normal cortical venous drainage was technically difficult, we changed the treatment to transarterial embolization (eg, occlusion of the feeder vessels and fistulas with liquid embolic material or applying coils to the fistula after accessing the fistula from the arterial route).

2.2.4 Statistical Analysis

We compared the imaging findings and grades of dural AVF obtained with hybrid CTA with those obtained with DSA. Sensitivity, specificity, positive predictive value, and negative predictive value were calculated for the six CT angiographic imaging

doi:10.6342/NTU201603776 findings (engorged artery, transosseous enhanced vessels, engorged extracranial vein,

engorged cortical vein, asymmetric sinus enhancement, and dural sinus occlusion) individually and overall. The adjusted Wald method was used to estimate the 95%

confidence intervals (CIs) of observed rates. The Cohen un-weighted κ statistic was used to assess the level of inter-observer agreement with regard to imaging signs for diagnosis of dural AVFs seen with hybrid CTA. A value of less than 0.20 implied poor agreement; values of 0.21–0.40, fair agreement; values of 0.41–0.60, moderate agreement; values of 0.61–0.80, substantial agreement; and values of 0.81–1.00, almost perfect agreement [16]. The 95% CI and Cohen κ were computed with online calculators [17,18].

2.3 Results

2.3.1 Patients

Of 22 patients, two (patients 8 and 10) had two dural AVFs each; thus, there were a total of 24 dural AVFs. Table 2-1 lists the demographic information, clinical manifestations, lesion locations, treatments, and outcomes for the 22 patients with 24 dural AVFs. There was no significant difference in age distribution (P=0.579) and birth sex (P=0.728) between the control subjects and patients with dural AVF.

2.3.2 Imaging Analysis

All 24 dural AVFs were examined with DSA and hybrid CTA. In all 24 lesions, at least two of the six imaging signs were present (Figure 1-5, Figure 2-1). Overall, the most common imaging finding was asymmetric sinus enhancement; this was not present in only the two patients with a dural AVF in the superior sagittal sinus. In the control

15

group, three subjects had a false-positive finding (two patients had a transosseous enhanced vessel, one patient had asymmetric sinus enhancement) and two had a false-negative finding (both had asymmetric sinus enhancement). The sensitivity, specificity, positive predictive value, and negative predictive value of the two readers in consensus are listed in Table 2-2. The κ test (Table 2-3) revealed a high level of interobserver agreement in the reading of the imaging signs. The best agreement was in regard to sinus occlusion and asymmetric sinus enhancement, followed by transosseous enhanced vessels.

As compared with the grade assigned with DSA (Table 2-4), the observed agreement between DSA and the readers was 100% for cavernous sinus, hypoglossal, and clival lesions and 78%–89% for lesions in the transverse sigmoid sinus.

2.3.3 Treatment Planning

Treatment planning by using hybrid CTA was performed before the patients received endovascular treatment. Hybrid CTA can provide additional information in identifying the best route of approach to treat the dural AVF. In osteodural type dural AVFs, the pathway between the drained sinus and the fistula is very tortuous and small.

Hybrid CTA can provide the road map about the best route to approach the lesion (Figure 2-2). In those dural AVFs that involved only the dura mater, hybrid CTA can localize exactly the fistula site inside the dura mater and its relationship to adjacent normal cerebral venous drainage and sinus (Figure 2-3). Hybrid CTA can also provide a high-resolution birds-eye view of the dural AVFs (Figure 2-4), which cannot be obtained with DSA. Although the drained sinus of the dural AVF was occluded, hybrid CTA can provide the information about localization and pathway for the trial to cross the occluded sinuses. In this series, nine patients were treated with the transvenous

doi:10.6342/NTU201603776 approach, seven patients who were originally to be treated with the transvenous

approach were treated with the transarterial approach, seven patients made a decision not to receive any treatment during the study period, and one patient was treated with stereotactic radiosurgery (Table 2-1). All nine patients in whom the transvenous approach was used were treated by using a route planned according to the results of hybrid CTA. In the patients who finally were treated with the transarterial approach (Figure 2-4), either treatment failed with the transvenous route or use of this route was not safe.

2.4 Discussion

The difficulty of bone removal limits the application of CTA in the diagnosis of dural AVF. In one re-cent report [19], CTA had a sensitivity of 15.4% in the diagnosis of dural AVF. In our study, we proved that hybrid CTA is a promising tool for the diagnosis of dural AVF. DSA can help in the diagnosis of dural AVF, can depict the supplying arteries, and can aid characterization of the draining system of dural AVF.

The latter determines the clinical presentation, grade, and therapeutic strategy in patients with a dural AVF [15,20,21]. Our results suggest that hybrid CTA could be used to determine the diagnosis in all types and grades of dural AVFs, to demonstrate the feeder artery, to aid localization of the lesion, and to show the pattern of venous drainage. One patient in the control group had a false-positive sign of asymmetric early venous enhancement. From DSA, we found that the patient had dominant drainage of the superior sagittal sinus to the right transverse sinus, which was less enhanced compared with the left side at hybrid CTA. We speculate that the vein of Labbé was diluted more at the dominant right transverse sinus owing to nonopaque blood flow coming from the superior sagittal sinus. Asymmetric opacification of the transverse and

17

sigmoid sinuses during CTA is a rather common finding, likely secondary to the preferential direction of opacified blood flow to the dominant sinus at the early time of arterial scanning. This asymmetry disappears when late-phase venous scans are obtained. Two patients had a false-positive sign of transosseous enhanced vessels in the occipital bone near the sigmoid sinus. We speculate that these were emissary veins. We were not able to identify the direction of the vessels inside the cranium with hybrid CTA. We need to emphasize that finding at least two imaging signs present is the key to making the right diagnosis in dural AVFs.

Two patients with a dural AVF in the superior sagittal sinus did not have asymmetric sinus enhancement. In general, the vertex superior sagittal sinus becomes densely enhanced near the end of the scan. As such, enhancement asymmetry may not be apparent in a patient with a dural AVF at or close to the vertex superior sagittal sinus.

However, other signs allowed diagnosis of dural AVF even in these patients with dural AVF of the superior sagittal sinus. Dural AVFs can be divided into two major groups according to their locations: cavernous sinus and non-cavernous lesions. In cavernous sinus dural AVFs, frequent findings include asymmetric early sinus enhancement and engorgement of the extracranial veins (frequently superior ophthalmic veins). Engorged arteries are infrequently seen in cavernous sinus dural AVFs because of their relatively small size. In non-cavernous dural AVFs, engorged arteries are more frequently seen.

Engorgement of cortical veins and extracranial veins depend on the size and the drainage patterns of the dural AVF. On the basis of our results, hybrid CTA is highly sensitive for different types of dural AVFs, especially in patients with the osteodural-type lesion. Hybrid CTA has the advantage of demonstrating the transosseous vessels and also the fistula inside the bone structures.

doi:10.6342/NTU201603776 On the basis of our study results, hybrid CTA may also provide sufficient

information for treatment planning. Hybrid CTA can provide additional information for identifying the route of approach to treat dural AVFs. Identification of the draining venous pattern is important for the treatment of dural AVFs [22,23]. Pretreatment planning with hybrid CTA was feasible in our series. Transvenous embolization is one of the treatments of choice for most dural AVFs. Hybrid CTA can be used to eliminate the bone structure, while the image noise in non-bone areas is not increased, and to obtain better three-dimensional volume data.

Our study had limitations. Because this was a retrospective study, the tube current for precontrast scanning could only be reduced by half. The radiation dose for precontrast scanning was slightly higher than that in the scanning protocol recommended for MMBE CTA, in which one-quarter tube current is suggested for a precontrast scan. The effective tube current–time product used in our hybrid CT angiographic scanning protocol was similar to that used in a study on the use of MMBE study [24]; however, the tube voltage is 100 kVp rather than 120 kVp in the proposed hybrid CTA protocol. For example, with the same machine (scanner B in our study) and same irradiated length, the irradiation dose (estimated CT dose index volume) was 9.4 mGy for the precontrast scans and about 18.8 mGy for the post-contrast scans with the hybrid CTA protocol and 7.8 mGy and 28.1 mGy, respectively, in the other MMBE protocols [24]. So, the total irradiation dose is lower with hybrid CTA than with other MMBE methods. A lower peak kilovolt is not used because of the concern about contrast-to-noise ratio on the images.

19

Another limitation of our study was the small number of cases. We could not conclude that hybrid CTA can be used for the diagnosis in all patients with a dural AVF.

False-positive and false-negative rates may be under-estimated because it is difficult to have a large number of patients who undergo DSA without positive hybrid CT angiographic, MR imaging, or MR angiographic findings. In addition, we used subjective findings that have not been well established for diagnosing retro-grade cortical veins. These need to be better defined with further research. In conclusion, we found that hybrid CTA is a valuable tool for the diagnosis of dural AVF. It can provide the key information necessary for treatment planning. Further studies in larger series of patients are warranted.

Note:

This part was published on Radiology 2010.

Lee CW, Huang A, Wang YH, Yang CY, Chen YF, Liu HM. Intracranial Dural Arteriovenous Fistulas: Diagnosis and Evaluation with 64-Dector Row CT Angiography.

Radiology. 2010; 256: 219-228.

doi:10.6342/NTU201603776 Figure 2-1: Patient 8. (a) Lateral DSA image after left external carotid injection shows

isolated left transverse sigmoid sinus dural AVF. (b) Coronal hybrid CT angiogram shows asymmetric sinus enhancement (large arrows), engorged cortical veins (small arrows), and incidental small right proximal transverse sinus dural AVF (arrowheads).

Images from (c) hybrid CTA and (d) MMBE CTA show isolated, occluded transverse sinus. Transosseous enhanced vessels (arrows on c) could be demonstrated only on hybrid CT angiogram.

21

Figure 2-2: Patient 15. Images in 48-year-old man with pulsatile tinnitus and dizziness.

(a) Maximum intensity projection image from hybrid CTA shows dural AVF (arrow) with early enhancement and engorged drainage vein at the left clivus to the paraspinal area. (b) hybrid CTA source image shows fistula in the left clivus (arrow). (c) DSA image of left external carotid artery (L ECA) demonstrates a fistula (arrow) at the left hypoglossal-clival area with paraspinal drainage. The lesion was treated with the transvenous approach according to findings from hybrid CTA through the route of the anterior internal spinal venous plexus. (d) Precontrast image obtained after embolization shows coils inside left clivus.

doi:10.6342/NTU201603776 Figure 2-3:Patient 13. Images in 60-year-old man with pulsatile tinnitus. (a) Lateral

projection of left external carotid artery demonstrates a type IIA dural AVF (Cognard et al [4] classification) at the left transverse-sigmoid sinus area. (b) Left common carotid artery angiogram obtained in the venous phase shows the left vein of Labbé (arrows) also draining into the left transverse sinus. (c) Hybrid CT angiogram obtained with volume rendering shows the fistula (red arrows) actually located at the dura next to the sigmoid sinus. The vein of Labbé (open arrows) is next to the fistula and drains into the transverse sinus.

23

Figure 2-4: Images in 56-year-old woman with pulsatile tinnitus. (a, b) DSA images of bilateral external carotid arteries demonstrate the fistula, engorged cortical vein, and superior sagittal sinus. The relationship between the fistula and the superior sagittal sinus is not clearly shown. (c) Hybrid CT angiogram obtained with volume rendering depicts the fistula inside the bilateral paramedian posterior frontal bones (red), the engorged cortical drainage vein (red), and patent superior sagittal sinus. After transarterial approach with coils deploying in the venous side of the fistula, complete

Figure 2-4: Images in 56-year-old woman with pulsatile tinnitus. (a, b) DSA images of bilateral external carotid arteries demonstrate the fistula, engorged cortical vein, and superior sagittal sinus. The relationship between the fistula and the superior sagittal sinus is not clearly shown. (c) Hybrid CT angiogram obtained with volume rendering depicts the fistula inside the bilateral paramedian posterior frontal bones (red), the engorged cortical drainage vein (red), and patent superior sagittal sinus. After transarterial approach with coils deploying in the venous side of the fistula, complete