Investigation of the need for an x-ray scatter-reduction grid during neurointerventional procedures

Abstract

X-ray guided neurointerventions are catheter-based treatments for cerebrovascular diseases such as strokes and aneurysms. During such procedures visualization of treatment devices is the primary imaging task. In this work we investigate the necessity of x-ray scatter-reduction grids in performing those tasks.

Various endovascular treatment devices such as stents, coils and catheters along with a low contrast blood vessel phantom were placed on a head-equivalent phantom. Images of the objects were acquired with and without a grid (15:1 grid ratio, 80 lines/cm and Al interspace). The x-ray field was set to the full 8 x 8 inch FOV to allow for realistic scatter generation. The detector was positioned close to the phantom to investigate maximal scatter conditions. Contrast and Contrast to Noise (CNR) ratios of the catheter tip and the blood vessel phantom were measured and compared for images obtained with and without the grid. The x-ray technique parameters were kept constant for all acquisitions.

For the catheter tip there was a 43% reduction in contrast with the removal of the grid due to increased scatter reaching the detector. However, due to increased primary there was a 18% increase in CNR. For the blood vessel phantom, there was a 33% reduction in contrast, whereas a 17% increase in CNR. All the devices and the blood vessels in the phantom were still visible even with the increased scatter without the grid. The results of the study indicate the use of grids during neurointervention procedures might not be necessary to perform the intervention.

Introduction :

Endovascular neuro-interventions are catheter-based treatment procedures for cerebrovascular diseases such as strokes and aneurysms. Catheter systems are introduced into the human vasculature through a minimal insertion typically into the femoral artery and then guided into the treatment area using x-ray imaging. Under the same x-ray guidance, the treatment devices such as stents and coils are then deployed to treat the disease. The main anatomy of interest during these procedures are the blood vessels. In x-ray images, blood vessels are often indistinguishable from the surrounding tissue and can only be visualized using roadmap and digital subtraction angiography techniques with iodine contrast agent. The primary and often times the only imaging task is to visualize the treatment catheters and devices, and not the tissue or the bone anatomy.

During these procedures, anti-scatter grids are used to reduce the scatter reaching the detector. This improves the contrast in the image. However, to compensate for the grid attenuation, the exposure is typically increased by the Bucky factor resulting in increased exposure to the patient. X-ray radiation is ionizing in nature and thus can cause patient injury. X-ray induced damage can vary from long term effects such as cancer and DNA modification, or short-term damage such as erythema and epilation. It is the responsibility of all the individuals using ionizing radiation in medicine to achieve an optimal balance between its clinical utility and its harmful effects to the patient. As per ALARA principle the use of x-ray radiation must be just enough to achieve the medical goals of the procedure.

In this work we investigated the necessity of an x-ray scatter reduction grid for imaging tasks described above and whether the visualization will be affected by removing the anti-scatter grid.

Methods and Materials:

Model Setup:

To simulate the head attenuation conditions, a modified AAPM head-equivalent phantom consisting of a stack of 37.5 x 37.5 square cm plates of 2.5 cm thick acrylic; 2 mm thick Al; 10 cm thick acrylic; 1 mm thick Al and 2.5 cm thick acrylic was used [1]. To investigate the effect of scatter on visualization of treatment devices, commonly used devices such as guide catheter, microcatheter, stents and guide wires were placed on this modified AAPM head equivalent phantom. Figure 1 shows the experimental setup.

Figure 1.

Experimental Setup to study effect of scatter on visualization of treatment devices. The devices are placed in cardboard tray with minimal x-ray attenuation on the head equivalent phantom. Head Equivalent phantom composition:

1: One 2.5 cm thick 37.5 x 37.5 sq cm acrylic block ;

2: One 1 mm thick 37.5 x 37.5 sq cm Al block;

3: Three, 2.5 cm thick 37.5 x 37.5 sq cm acrylic blocks;

4: One 2 mm thick 37.5 x 37.5 sq cm acrylic block;

5: One 2.5 cm thick 37.5 x 37.5 sq cm acrylic block

The images were acquired with and without the grid using the 7.5 x 7.5 sq cm FOV, Hi-Def mode of the dual resolution detector. However to simulate maximum scatter conditions, using manual programming the x-ray exposure field was kept at 20 x 20 sq cm at the detector entrance

To investigate the effect of scatter on visualization of blood vessels, one of the 2.5 cm acrylic blocks was replaced with a 2.5 cm acrylic block containing an iodine-filled low contrast artery [2]. Figure 2 shows the experimental setup.

Figure 2.

Experimental Setup to study effect of scatter on visualization of artery phantom.

Head equivalent phantom composition:

1: One 2.5 cm thick 37.5 x 37.5 sq cm acrylic block ;

2: One 1 mm thick 37.5 x 37.5 sq cm acrylic block;

3: One , 2.5 cm thick 37.5 x 37.5 sq cm acrylic artery phantom;

4: Two 2.5 cm thick 37.5 x 37.5 sq cm acrylic block;

5: One 2 mm thick 37.5 x 37.5 sq cm acrylic block

The images were acquired with and without the grid using the 20 x 20 sq cm FOV, FPD mode of the dual resolution detector. To simulate maximum scatter conditions, using manual programming the x-ray exposure field was kept at 20 x 20 sq cm at the detector entrance

To maximize the effect of scatter, the x-ray exposure field was opened to a large 8 inch x 8 inch FOV at the image receptor and the air gap between the object and the detector was kept minimum. Images of the devices and the artery phantom were acquired with and without the standard manufacturer-installed grid (15:1 grid ratio, 80 lines/cm and Al interspace). The x-ray exposure technique factors were kept at 78 kVp, 160 mA and 8.0 ms for all images shown here.

Image Detector:

To image the objects a new dual resolution detector from Canon Medical Systems Corporation, Tochigi, Japan [3] was used. The detector has the following two imaging modes

• Flat-panel detector (FPD) mode with Conventional FPD pixel array

  • Pixel size 194 μm
  • FOV : 30 × 30 sq cm to 15 × 15 sq cm
• HiDef mode with Complementary metal oxide semiconductor array

  • Pixel size 76 μm
  • FOV: 7.5 × 7.5 sq cm to 3.8 × 3.8 sq cm

The mode desired can be selected instantly using an electronic switch, without lengthening the procedure time. In this work the HiDef mode ( 7.5 × 7.5 sq cm) was used to acquire images of the treatment devices and the FPD mode (20 x 20 sq cm) for the artery phantom. Using manual programming the collimator blades were kept open to 20 x 20 sq cm FOV at the detector entrance to allow for maximum scatter generation.

Results:

Figure 3 shows the high-resolution images of commonly-used interventional devices with the grid (figure 3A) and without the grid (figure 3B) acquired using the Hi-Def mode of the detector. It can be seen that due to more scatter reaching the detector the brightness in figure 3B is higher than figure 3A, but the contrast in figure 3B is lower than figure 3A. Using image processing the display contrast in the image can be restored (Figure 3C) to levels similar to the image with the grid (Figure 3A). Table 1 shows the percent contrast, background signal to background noise measurements, and the contrast to background noise (normalized to the signal in the background) ratio for the two regions shown in figure 3B. For the catheter tip there was a 43% reduction in contrast with the removal of the grid due to increased scatter reaching the detector. However, due to increased primary there was a 18% increase in CNR.

Figure 3:

High Resolution images of regularly-used interventional devices, placed on a head equivalent phantom acquired using the Hi-def mode of the detector . The x-ray FOV was opened to 20 x 20 sq cm at the detector to allow for maximum scatter conditions.

Figure 3A : Grid on.

Figure 3B : Grid off.

Figure 3C : Figure 3B with display contrast adjusted.

The two regions (R1 and R2) shown in Figure 3B are for contrast measurements presented in table 1.

Table 1:

Contrast and Contrast to noise (CNR) calculated between the two regions (guide catheter tip and background as shown in figure 2B) for both grid (Figure 2A) and no grid conditions (Figure 2 B). Each measurement was repeated for a set of 6 consecutive images and presented in the table as mean ± std deviation.

		Grid	No Grid	% difference between no grid and grid
Region 1 (R1)	Signal (Avg. Grey Level)	98.2 ± 0.8	375.9 ± 1.3
Region 2 (R2)	Signal (Avg. Grey Level)	189.3 ± 0.3	515.4 ± 0.8
	Noise (Std. Dev.)	11.9 ± 0.2	15.4 ± 0.2
	SNR (Signal to Noise)	15.9 ± 0.2	33.5 ± 0.5
% Contrast	(1-(R1/R2))*100	48.1 ± 0.4	27.1 ± 0.2	43.6% decrease
CNR	(R2-R1)/Noise_R2	7.7 ± 0.1	9.1 ± 0.1	18% increase

Figure 4 shows the images of the low contrast artery phantom, with the grid (figure 4A) and without the grid (figure 4B) acquired using the FPD mode of the detector. Table 2 shows the contrast and contrast to noise measurements for the two regions shown in figure 4B. For the blood vessel phantom, there was a 33% reduction in contrast, whereas a 17% increase in CNR.

Figure 4:

Images of the low contrast artery phantom acquired using the FPD mode of the detector. The x-ray FOV was opened to 20 x 20 sq cm at the detector to allow for maximum scatter conditions.

Figure 4A : Grid on.

Figure 4B: Grid off.

Figure 4C: Figure 4B display contrast adjusted. The two regions (R1 and R2) shown in Figure 3B are for contrast measurements presented in table 2.

Table 2:

Contrast and Contrast to noise (CNR) calculated between the two regions (as shown in figure 3B) for both grid (Figure 3A) and no grid conditions (Figure 3B). Each measurement was repeated for a set of 6 consecutive images and presented in the table as mean ± std deviation. Even though the x-ray exposure technique factors are the same between figures 2 and 3, the detector acquisition modes are different, resulting in the difference in measured signal and noise levels.

		Grid	No Grid	% difference between no grid and grid
Region 1 (R1)	Signal (Avg. Grey Level)	1549 ± 3.6	4392 ± 1.9
Region 2 (R2)	Signal (Avg. Grey Level)	1592 ± 2.2	4474 ± 3.5
	Noise (Std. Dev.)	50 ± 0.4	83.4 ± 0.9
	SNR (Signal to noise)	32 ± 0.3	54 ± 0.6
% Contrast	(1-(R1/R2))*100	2.7 ± 0.18	1.8 ± 0.16	33% decrease
CNR	(R2-R1)/Noise_R2	84 ± 5.9	98 ± 6.8	17% increase

Discussion:

Two critical elements during intervention procedures are real time imaging and x-ray radiation dose. Since x-rays are ionizing radiation, the imaging systems utilize the ALARA principles[4, 5], which recommends use of just enough radiation to get acceptable image quality to perform the procedure successfully.

The main anatomy of interest during neurointervention procedures is the cerebrovasculature and not the tissue or the bone anatomy. In x-ray images, blood vessels are often indistinguishable from the surrounding tissue and can only be visualized using iodinated contrast media with techniques such as roadmapping and digital subtraction angiography. The primary, and often times only, imaging task is to visualize the treatment catheters and devices. This is different from the field of mammography where x-rays are used to image breast tissue or bone radiography where x-rays are used to image bone anatomy.

Druing neurointervention procedures the use of x-ray scatter reduction grids is a universally-followed clinical practice for any x-ray angiographic system. While they do effectively prevent the scatter from reaching the detector, they also reduce the primary radiation. To compensate for the grid attenuation, the exposure is typically increased by the Bucky factor resulting in increased exposure to the patient.

In this work we investigated

1. The effect of scatter on the main imaging task which is the real-time visualization of treatment catheters and devices.

2. Need for an x-ray scatter-reduction grid to accomplish this task.

Commonly used actual treatment devices such as stents, catheters and wires were placed on head equivalent phantoms. Iodine filled low contrast artery phantom was used to simulate blood vessel visualization using iodine contrast injection. To maximize the effect of scatter, the x-ray exposure field was opened to a large 20 x 20 sq cm FOV at the image receptor and the air gap between the objects and the detector was kept minimum. Images of the devices and the artery phantom were acquired with and without the standard manufacturer-installed grid (15:1 grid ratio, 80 lines/cm and Al interspace).

The measurement results presented in table 1 and 2 show that when the grid is removed there is more scatter reaching the detector resulting in a decrease in contrast. However due to increased primary quanta reaching the detector with the removal of the grid, the contrast to noise ratio is improved. The treatment devices and the vessel anatomy are visible in both grid (figure 3A,4A) and no grid (figure 3B,4B) images. Furthermore, using image processing the display contrast in the no grid images can be restored to similar levels as in the grid images (figure 3C,4C). The results in the tables and the figures suggest that even with increased scatter present without the grid, the primary imaging task is still achieved.

Conclusions:

The results of the study indicate the use of grids during neurointervention procedures might not be necessary to successfully achieve the clinical task for the intervention.