Human-Compatible Multi-Contrast Mammographic Prototype System

Abstract

In the past decade, grating-based x-ray multi-contrast imaging has demonstrated potential advantages for breast imaging, including reduced anatomical noise, sharper tumor boundary and improved visibility of microcalcifications. However, most of the studies have been performed on benchtop-based systems. The experimental conditions including the dose, scanning time and system geometry may not meet clinical standards. Therefore, to evaluate true clinical benefits of grating-based multi-contrast breast imaging, in-vivo imaging should be performed, which requires a human-compatible system. The purpose of this paper is to report the development of a human-compatible prototype multi-contrast imaging system. In particular, this work focuses on several key challenges in building the prototype system. Regarding the challenge of patient safety, the mean glandular dose (MGD) and the scatter radiation were evaluated for the prototype system. Regarding the challenge of the limited field-of-view (FOV), the origin of the problem and corresponding technical solutions are presented. Finally, imaging results of several test phantoms are presented and strategies to improve the image quality are discussed.

1. INTRODUCTION

The propagation of x-rays through the breast tissue is often accompanied by a series of physical interactions. Among a variety of possible interaction mechanisms, conventional mammography primarily utilizes photoelectric absorptions and incoherent Compton scatterings to generate image contrast. Tissue information encoded via other types of physical mechanisms such as small-angle scattering (SAS) and refraction is unfortunately wasted in the attenuation-contrast mammography that solely explores the particle nature of x-rays. The use of a Talbot-Lau interferometer allows an x-ray imaging system to pick up not only attenuation signal but also information carried on the wave nature side of x-rays, enabling simultaneous generation of attenuation, phase, and dark-field contrast images from a single data acquisition process.^1–3 This so-called multi-contrast x-ray imaging method has demonstrated potential in breast cancer imaging, including noninvasive classification of microcalcifications,⁴ improved visibility of microcalcifications,⁵ better visualization of spiculation and abnormal fibrous structures in highly dense breasts,⁶ etc.

Despite these promising applications indicated in previous studies, an intriguing question to be answered is whether the multi-contrast x-ray imaging method can provide additional clinical benefits to conventional mammography in a clinically-compatible setting. The question remains open since the vast majority of previous studies were performed at exposure levels, acquisition times, or system geometries that far exceeded those deemed clinically acceptable. For benchtop-based studies, the imaging conditions were quite idealized compared with a clinical system. For example, the use of the optical table significantly alleviated the impacts of tube/gantry vibration to the interferometer system. Finally, some works that demonstrated improved imaging sensitivity to breast cancer masses and microcalcifications were performed using formalin-fixed tissue specimens or very thin specimens, which may not accurately represent the signal and noise level of in-vivo imaging.

To fully evaluate the clinical benefits of multi-contrast breast imaging, we have constructed a prototype system based on a clinical full-field digital mammography (FFDM) unit.⁷ The purpose of this paper is to report the development of the prototype system towards human-compatible imaging. In particular, the work focuses on addressing some key challenges in constructing the prototype system: (1) patient safety concerns; (2) limited field-of-view (FOV) of grating-based interferometer system; (3) compact system geometry. The next section will elaborate on these challenges and the corresponding scientific approaches to tackle these challenges.

2. METHODS

2.1. Patient safety

Patient safety during in-vivo imaging is one of the foremost considerations in developing the human-compatible prototype system. With the added gratings and mechanical components, preventives actions should be taken to protect patients from potential safety hazard such as scattered radiation to the patient body and skin cut by the sharp components of the grating interferometer. In addition, the system should withstand a breast compression force up to 200 N (45 lbs). Since major components of the interferometer are placed in between the breast and the detector, a collapse of the breast support may lead to serious patient injury.

As shown in Fig. 1, a customized breast support was developed to (1) conceal the sharp components of the interferometer from the patient; (2) prevent x-rays scattered by the interferometer from reaching the patient chest wall; (3) withstand strong compression force but with minimal x-ray attenuation loss. The side walls of the device are made of lead-lined solid wood panel to absorb scattered x-rays. The exterior of the side walls was carefully sanded and sealed to improve patient comfort and prevent skin cut. The top surface was made of very thin sheet of plexiglass (0.1”) and was strengthened by T-slot aluminum extrusions mounted outside the primary field. Equipped with this device, the prototype system passed safety tests conducted by board-certified medical physicist and mammography technologist.

Figure 1.

(a) Constructed prototype system. The breast support in (b) conceals interferometer components in (c).

To measure the scattered radiation, a mammographic accreditation phantom (Gammex 156, Gammex Inc., Middleton) was placed in the imaging position and the scattered radiation on the plane close to the chest wall was measured. Comparison of the measured scattered radiation between the original mammography unit and the modified system is shown in Fig. 2. Results show that the scattered radiation is comparable to that of the original mammography system when their mean glandular doses are matched.

Figure 2.

Measured scattered radiation maps.

To estimate the mean glandular dose for a five-step phase stepping acquisition, the incident air kerma was measured to be 4.9 mGy under the following exposure setting: 36 kVp, Rh/Rh, 500 mAs. The half value layer was measured to be 0.79 mm aluminum. Based on these measurements, the mean glandular dose was estimated to be 2.4 mGy, using methods by Dance et al.^8–10

2.2. Limited FOV

The compact system geometry may introduce the so-called vignetting effect, which is a loss of fringe visibility on the peripheral region of the grating due to the reason shown in Fig. 3. Revol et al. showed that the critical angle (ϕ_c) beyond which the visibility will be completely lost is given by¹¹

$${\varphi }_c={\text{tan}}^{-1}\left(\frac{p}{h}\right).$$ (1)

Figure 3.

Cause of the vignetting effect.

For a Talbot-Lau setup, the vignetting effect is mostly caused by the analyzer grating, the usable spatial area is determined by the length L_c covered within ±ϕ_c:

$$L_c=\text{SDD}\times \left(2\text{tan}{\varphi }_c\right)=\frac{2p}{h}\text{SDD},$$ (2)

where SDD is the source-to-detector distance. Compared with a non-compact benchtop system, both p and SDD of the prototype system are much smaller, which reduced the usable grating area and increased the severity of the vignetting effect. As shown in Fig. 4, when using flat analyzer grating, the center of the FOV has a visibility of 28%, however, the visibility quickly drops to almost zero in the peripheral region. As a result, the noise level is greatly amplified in the differential phase contrast image.

Figure 4.

Visibility maps (top row) and phase contrast images (bottom row) acquired with conventional flat grating and bent grating.

To address this issue, we have designed a frame to bend the analyzer grating. As shown by the fringe visibility map in Fig. 4, by bending the grating using our in-house method, the undesirable vignetting effect was effectively mitigated. The whole area of the analyzer grating has almost uniform fringe visibility and uniform noise property in the differential phase contrast image.

In addition to the vignetting effect, another factor that often limits the FOV of multi-contrast imaging system is the relatively small area of grating. Even fabricated with a 6” silicon wafer, the active area of the grating is limited to about 10 × 10 cm². In comparison, the FOV of a full field digital mammography unit can be as large as 29 × 24 cm². Apparently, multiple gratings need to be tiled together to meet the FOV requirement of human breast imaging. As shown in Fig. 5, two gratings can be titled to enlarge the FOV, with minimal signal loss in the space between two tiles.

Figure 5.

Images acquired with two pieces of analyzer gratings tiled together. The gap in-between the two pieces is barely perceivable in the phase and dark-field image. The attenuation image shows a single pixel-width bright gap because of reduced beam hardening effect.

3. RESULTS AND DISCUSSION

To test the capability of the prototype system in generating multi-contrast images, a slice of kiwi fruit and a slice of pork belly were imaged. The total entrance dose of the phase stepping acquisition was 4.9 mGy. Results are shown in Fig. 6. The dark-field image of the kiwi fruit shows superior contrast with respect to the attenuation image. The differential phase contrast image of the pork belly sample shows sharp edges while the dark-field image shows reduced anatomical background with improved visibility of microcalcifications. Note that microcalcification simulants can also be observed in the attenuation image, howwever, one can hardly distinguish between microcalcifications and muscles due to their very similar contrast levels.

Figure 6.

Multi-contrast images of a kiwi fruit (top row) and a slice of pork belly with microcalcification simulants (bottom row).

While the performance of the current system seems to be sufficient for these test phantoms, for breast imaging, the sensitivity of the system may need further improvement due to the lower signal level and stronger attenuation of the breast specimen. To further improve the performance of the system for breast imaging, several strategies will be applied in the future work: (1) using smaller period gratings and increasing the intergrating distance while keeping the source-to-detector distance fixed; (2) using less absorbing grating substrate material.

4. CONCLUSION

To evaluate the true clinical utility of x-ray phase contrast and dark-field imaging in breast cancer diagnosis, a prototype multi-contrast breast imaging system that is compatible with in-vivo human imaging was constructed based on a clinical digital mammography unit. Challenges introduced by the compact system geometry, limited FOV due to the small grating area, and strict patient safety regulations in breast imaging were addressed via technical innovations presented in this work. Imaging results of several test phantoms have demonstrated the capability of the prototype system in generating multi-contrast images with clinically acceptable dose level and imaging time, which cleared the path for initiating the in-vivo human subjects study in the upcoming phase of the project.