Chinese Journal of Chemical Physics  2018, Vol. 31 Issue (4): 529-536

#### The article information

Guohua Cao

Biomedical X-ray Imaging Enabled by Carbon Nanotube X-ray Sources

Chinese Journal of Chemical Physics, 2018, 31(4): 529-536

http://dx.doi.org/10.1063/1674-0068/31/cjcp1806132

### Article history

Received on: June 5, 2018
Accepted on: June 19, 2018
Biomedical X-ray Imaging Enabled by Carbon Nanotube X-ray Sources
Guohua Cao
Dated: Received on June 5, 2018; Accepted on June 19, 2018
Virginia Tech-Wake Forest University School of Biomedical Engineering and Sciences, Virginia Polytechnic Institute and State University(Virginia Tech), Blacksburg, VA 24061, USA
*Author to whom correspondence should be addressed. Guohua Cao, E-mail: ghcao@vt.edu
Abstract: Although discovered more than 100 years ago, X-ray source technology has evolved rather slowly. The recent invention of the carbon nanotube (CNT) X-ray source technology holds great promise to revolutionize the field of biomedical X-ray imaging. CNT X-ray sources have been successfully adapted to several biomedical imaging applications including dynamic micro-CT of small animals and stationary breast tomosynthesis of breast cancers. Yet their more important biomedical imaging applications still lie ahead in the future, with the development of stationary multi-source CT as a noteworthy example.
Key words: Biomedical imaging    X-ray imaging    X-ray source    Carbon nanotube X-ray source    Carbon nanotube    Field emission    Computed tomography    Breast tomosynthesis
Ⅰ. INTRODUCTION

Among many available and emerging biomedical imaging technologies, X-ray imaging is by far the dominant form today for the reasons of penetration depth, resolution, signal-to-noise ratio, as well as cost-effectiveness. Currently, at least ten X-ray images are acquired for each of the other imaging modalities [1]. Medical X-ray market size was valued at USD 9.9 billion in 2016 and is projected to reach USD 15 billion 2024 [2].

All X-ray imaging technologies are based on the physical principles that govern the interactions between matters and X-rays, which include photoelectric absorption, Compton scattering, and Rayleigh scattering, K-edge absorption, X-ray fluorescence. One of the key elements for all X-ray imaging-the X-ray-was discovered more than a century ago in 1895 [3]. Since its discovery, however, the way of generating X-rays has essentially not changed in the past century.

A. Conventional X-ray sources

X-ray was discovered by Dr. Roentgen when he was experimenting with a Crookes tube, a device used to study fluorescence at that time by applying a voltage potential between a cathode and an anode in a gas-filled or partially-evacuated tube. In the early 1900s, in order to improve the X-ray output, Coolidge modified the X-ray tube by heating the cathode [4]. Even until today, in almost all X-ray sources approved for X-ray imaging in patient care, their cathode filaments need to be heated to around 1000 ℃ (hence called 'hot' cathodes). This high temperature causes the charge carriers (electrons) in the cathode material to obtain a sufficient thermal energy to overcome a potential-energy barrier (work function of the material), so that the charge carriers (electrons) in the filament can escape into the vacuum. This process is called the thermionic emission. An empirical equation for thermionic electron emission can be written as

 $\begin{eqnarray} J = {A_G}{T^2}{{\rm{e}}^{ - W/kT}} \end{eqnarray}$ (1)

where $A_G$ is a constant, $J$, $T$, and $W$ represent current density, temperature, and work function, respectively. Electrons released from the thermionic emission process are accelerated toward a metal anode by a potential voltage applied between the cathode and the anode. At 100 kV voltage, electrons can obtain a relativistic speed of up to 55% of $c$, where $c$ is the speed of light in vacuum. When electrons of such high speed strike the anode, the deceleration (i.e. braking) of the electrons produces X-rays from a process known as Bremsstrahlung. Our long experience in using the thermionic emission electron sources as the method of X-ray generation has produced reliable X-ray sources for many applications including clinical diagnosis and treatment, industrial inspection, transportation security, forensic investigation, and material analysis. However, the thermionic approach has several intrinsic limitations on X-ray sources including slow switching, high operating temperature, bulky size, and complex electromagnetic optics for electron-beam focusing [5, 6].

B. Field emission cathodes

To overcome the limitations of the conventional X-ray sources due to the thermionic emission, other cathode materials that employ a different electron emission mechanism, particularly field electron emission, were explored. Field emission is a quantum phenomenon very different from thermionic emission, in which electrons in a cathode material are able to 'tunnel' through the potential energy barrier under the influence of an electric field. Field electron emission is governed by the following equation

 $\begin{eqnarray} J = a{V^2}{{\rm{e}}^{ - b{\phi ^{3/2}}/\beta V}} \end{eqnarray}$ (2)

where $a$ and $b$ are constants, $J$, $V$, $\phi$, and $\beta$ represent current density, applied voltage, work function, and field enhancement factor, respectively. Field emission cathodes require no heating (hence called 'cold' cathodes), work at room temperature, and therefore can release electron almost instantaneously with application of an electrical field.

The quests to achieve a field-emission based X-ray source date back to the 1950s by Dyke [7]. Following the work of Dyke, in 1968 Spindt [8] published a new method of fabricating field emission arrays based on Mo tips (i.e. Spindt tip emitters). Although the Spindt tip emitters have found applications in many electron emission systems such as field emission displays [9], the Spindt tip emitters can only offer relatively low aspect ratios (the ratios of length to width), which prevents the Spindt tip emitters to be used in some more demanding application such as X-ray sources. Cathode materials in X-ray sources often require high current densities with low turn-on voltages. New cathode materials with suitably high current densities and low turn-on voltages were required. With the advent of nanotechnology, nanowires and nanotubes were recognized for their high aspect ratios and sharp tips, which can concentrate the applied electric field to their tips (i.e. field enhancement factor). Furthermore, self-assembly via chemical vapor deposition enabled the nanomaterials to be fabricated over large areas. Therefore, nanowires and nanotubes were long considered good candidates for cathode materials in X-ray sources [10].

Carbon nanotubes (CNTs) are such nanomaterials that have been investigated extensively in the past decades for their use in X-ray generation. With their discovery in the early 1990s [11], CNTs were recognized for their potential to achieve aspect ratios of about 10$^4$, high thermal conductivity, and high chemical stability, which quickly led to their applications as field emission materials [10]. CNT field emitters have been constructed using both single-walled CNTs (SWNTs) and multi-walled CNTs (MWNTs) [12]. However, because SWNTs are more reactive and thermally less stable than MWNTs, MWNTs are the most commonly used as electron field emitters in X-ray sources used in biomedical imaging, in which high current densities and long-term stability under high-voltage and in a non-ideal vacuum environment are needed [13].

C. Carbon nanotube X-ray Sources

A CNT X-ray tube employing CNT field emitters as the cathode is shown in FIG. 1. A typical CNT based X-ray tube has a triode-type structure with a CNT cathode, a gate electrode, a focusing electrodes, and a metal target housed in a vacuum tube with an X-ray window. X-ray tube current is generated when the electron beam from the CNT cathode is focused to a small area on the anode. The X-ray tube current increases exponentially with the applied electric field on the CNT cathode, a characteristic of all field emission X-ray sources. X-ray radiation can be turned on/off instantaneously by switching on/off the extraction voltage on the gate electrode. A spatially distributed CNT X-ray source can be built using a multi-pixel CNT cathode where each CNT pixel is equivalent to the thermionic cathode in a traditional single-beam X-ray source. Each CNT pixel can be individually controlled by programming the voltage on the corresponding gate electrode, and once activated, sends an electron beam to a distinct focal spot on the X-ray anode to generate X-rays [14]. By programming the voltages on all gate electrodes, a scanning X-ray beam can be generated in various sequences. One example is a scanning beam generated sequentially from different locations on the X-ray anode one at a time, which illuminates the object from different viewing angles to provide tomography imaging without mechanical motion. The second is multiplexing, where a subset of the X-ray beams is activated at a given time according to a predetermined scheme [15].

 FIG. 1 (a) A prototype CNT X-ray source. The ion pump is at the right and the X-ray window is at the left, from which the X-rays exit. (b) A schematic of a CNT X-ray source. CNT emitters are at the bottom. A gate electrode is placed on top of the CNT cathode. When a sufficient voltage is applied at the gate, electrons are released from the CNT cathode and accelerated toward the anode to generate X-rays. A SEM image of the CNT cathode is also shown.

There are several advantages for CNT X-ray sources. First, the 'cold' field emission approach in CNT X-ray sources solves the heat problem on the cathode, which could allow for the tight packaging of multiple CNT cathodes at a very small pitch distance. Second, CNT X-ray sources enable ultrafast switching on and off X-rays [16], leading to a higher temporal resolution [17]. Lastly, CNT X-ray sources allow an array of X-ray sources to be densely populated inside the same vacuum chamber [18]. The result is a compact distributed X-ray source that can be controlled with great precision, both in space and in time.

Ⅱ. CURRENT BIOMEDICAL IMAGING APPLICATIONS OF CNT X-RAY SOURCES

The unique advantages of CNT X-ray sources in modulating X-ray emissions in space and time have allowed some interesting biomedical imaging applications. In this section, we present two example applications, one utilizing the fast temporal switching advantage of CNT X-ray sources, and the other utilizing the spatial modulation advantage of CNT X-ray sources.

A. Preclinical micro-CT imaging

The fast switching advantage of CNT X-ray sources has been demonstrated in small animal micro-CT imaging. Preclinical small animal micro-CT imaging has a critical role in phenotyping, drug discovery, and understanding the pathophysiology of disease [19, 20]. Because of the development of gene knockout and transgenic technologies, small animal models, particularly genetically engineered mice, have become widely used in basic and preclinical studies of cancer and cardiovascular disorders [21, 22]. Nowadays, small animal imaging constitutes an integral part of testing new pharmaceutical agents prior to commercial translation to clinical practice. Small animal imaging is thus crucial in a wide range of biomedical investigations, including the majority of new drug discovery, phenotyping of transgenic animals, profiling of new disease models, pharmacological and pharmacokinetic analysis for target identification, and safety evaluation of new biotechnologies and biomaterials [23, 24].

Micro-CT imaging the thorax of the mouse, however, has proven to be challenging, primarily due to the motion blur resulting from the rapid cardiac (up to 600 beats per minute) and respiratory rates (about 120 breaths per minute) in mice. Continuous imaging and retrospectively selecting the images according to the recorded respiratory and cardiac cycles allows for nearly motion-free micro-CT imaging, but this approach involves a high radiation dose, often at a substantial fraction of the lethal dose for a small animal [25]. The ability to perform longitudinal imaging is thus limited. Prospectively gated micro-CT imaging can be performed on animals after the animal's breathing was controlled with intubation and mechanical ventilation, however, this invasive approach affects lung morphology and physiology [26, 27]. The instantaneous and flexible switching of the CNT X-ray sources offered a solution to imaging anesthetized but free-breathing animals, at both high spatial and temporal resolutions, with high dose efficiency, and without need for intubation and ventilation.

CNT-enabled X-ray sources capable of biomedical imaging have been incorporated into in vivo micro-computed tomography (micro-CT) scanners for dynamic micro-CT imaging of small animals [28, 29]. A CNT X-ray source powered dynamic micro-CT scanner has been developed. X-ray image acquisition in this scanner can be precisely timed to the desired physiological phase in the respiratory and/or cardiac cycles of a mouse. The respiratory and cardiac cycles in live animals can be recorded, in real-time speed, by pneumatic or optical sensors of the chest expansion and by electrocardiogram electrodes, respectively. Based on the respiratory and cardiac waveforms from the sensors and electrodes, a physiological trigger signal can be generated from the desired respiratory and cardiac phases, which was sent to the CNT dynamic micro-CT scanner to trigger X-ray pulse exposure and image data acquisition. After one X-ray pulse is fired and one projection image is acquired, the gantry rotates and the scanner awaits the same physiological state to acquire the next projection image. This process is repeated until a sufficient number of projection images are collected. A inside view of a CNT dynamic micro-CT scanner is shown in FIG. 2, along with the representative dynamic micro-CT images of a beating mouse heart acquired at the diastolic (heart expansion) and systolic (heart contraction) phases. The typical scan time is about 10 min for the acquisition of a total of 400 projection images. The scan time depends on the animal's heart and respiratory rates but is primarily limited by the slow gantry rotation speed. As shown in FIG. 2, image quality is excellent, which is a result of the small focal spot size ($<$100 μm), narrow X-ray pulse width ($\sim$10 ms), and nearly instantaneous switching on/off the X-rays by the CNT X-ray source. Furthermore, such high quality micro-CT scan only requires radiation dose of about 10 cGy per scan, which easily allows repeated longitudinal imaging on the same animal. The radiation risk can be further reduced by up to 85% if a region-of-interest oriented scanning mode (i.e. interior tomography) is implemented [30, 31]. To date, the CNT micro-CT technology has been used in experiments to measure heart and lung functions [32, 33], detect and treat heart disease [34, 35], and guide therapy for cancers [36].

 FIG. 2 An inner view of a CNT dynamic micro-CT scanner. The scanner is capable of dynamic 4D (3D space+time) imaging of the free-breathing lung and fast-beating heart in a mouse. Micro-CT images shown here are from two micro-CT scans of the same mouse heart. (a) and (c) CT slice images were acquired at the diastolic phase, and (b) and (d) CT slice images were acquired at the systolic phase.
B. Stationary breast tomosynthesis

The spatial modulation advantage of CNT X-ray sources has been readily demonstrated in a CNT X-ray source based stationary breast tomosynthesis system. Tomosynthesis (Tomo) is an imaging approach in which multiple radiographic projections are obtained from different view angles relative to the object [37]. It is essentially a limited-angle three-dimensional (3D) CT, in which a relatively high resolution can be obtained in the axial plane but only a coarse resolution can be obtained in the third (depth) direction. Tomo was first introduced in the late 1930s and represents our earliest attempt to overcome the limitation of planar imaging, mainly the superposition of overlying detail. Even after the invention of the CT (which is a truly 3D imaging device) in the late 1970s, Tomo still underwent many developments and it is now adapted to many clinical settings, such as angiography [38].

Tomosynthesis imaging systems today employ a digital X-ray detector and a conventional X-ray source mounted on a rotation gantry. During image acquisition, the X-ray source is rotated by the gantry in a continuous or stop-and-shoot mode, to collect the series of projection images from different view angles ($<$80$^\circ$) [39]. Rotating a bulky thermionic X-ray source by a rotation gantry in a tomosynthesis system is slow, and additionally, a significant amount of blur will be resulted, either from the source motion or from the patient motion due to the slow process. Blur leads to degraded spatial resolution, which is the most important factor in characterizing micro-calcifications in early breast cancers. As a result, digital breast tomosynthesis (DBT) today is approved for use only in combination with the standard mammography for breast cancer screening.

Because of the 'cold' cathodes in CNT X-ray sources, CNT X-ray technology allows many X-ray sources to be closely spaced together and aligned in an array [40]. Moreover, each X-ray source in the array can be individually controlled using an external electronic circuit. Given the fast switching speed of the CNT X-ray source technology, a scanning X-ray beam can be easily generated from a CNT X-ray source array. Through programming the voltages on all gate electrodes, a scanning X-ray beam can be generated either sequentially, or a selected pattern such as multiplexing [15]. In a CNT-enabled stationary breast tomosynthesis shown in FIG. 3, acquisition of the 31 projection views can be achieved via sequentially turning on and off the individual CNT X-ray sources in the CNT X-ray source array, in equivalence to "moving" a conventional X-ray source from one end to the other but without the actual mechanical rotation. The improvement in image resolution is obvious when comparing the tomosynthesis images of the same breast phantom acquired from a conventional DBT with moving a conventional X-ray source and a stationary DBT with a linear CNT X-ray source array, as shown in FIG. 3 (a) and (b) [41, 42, 43]. The stationary DBT technology for breast cancer screening is currently under clinical trials [44].

 FIG. 3 A prototype stationary breast tomosynthesis system based on a linear CNT X-ray source array. The improved spatial resolution from the stationary breast tomosynthesis system can be easily appreciated by visually comparing the two sets of tomosynthesis images acquired from the same breast phantom: (a) images from conventional DBT with moving a conventional X-ray source, and (b) images from the stationary DBT with a linear CNT X-ray source array.
Ⅲ. OUTLOOK ON CNT ENABLED BIOMEDICAL X-RAY IMAGING

As shown in the Section Ⅱ, the novel CNT X-ray source technology has already enabled some unprecedented biomedical imaging applications, such as small animal dynamic micro-CT and stationary breast tomosynthesis. Looking forward, there exist quite some biomedical imaging applications where the CNT X-ray source could make strong impacts, with potential for safer, faster, and more flexible biomedical X-ray imaging technologies. One interesting possibility is the development of a stationary multi-source CT.

A. History of multi-source CT

CT is the leading imaging modality in the modern diagnosis. Ever since Hounsfield's pioneering work on the first CT prototype [45], increasing imaging speed has received the highest priority and was the driving force behind CT developments. Among the many technological breakthroughs that contribute to this ever faster imaging speed, the ability to increase the gantry rotation speed has been essential [46]. However, this rotation system architecture is fast approaching its physical limit (the $g$-force on gantry now exceeds 30 g) [47]; an attractive alternative to faster CT is multi-source systems.

CT systems with multiple X-ray sources have been discussed in the 1970s. The dynamic spatial reconstructor (DSR) in the 1980s is the first real multi-source CT prototype [48]. Although a fast scan speed ($\sim$17 ms) has been demonstrated, the DSR had to employ a large rotation gantry ($\sim$5 m in diameter) to fit $\sim$30 X-ray sources [48, 49]. It was prohibitive for wide-spread application due to its size and cost, and thus was dismantled. The state-of-the-art in the medical CT field is the dual-source CT (DSCT) scanner which utilizes a regular slip-ring gantry and two X-ray sources [50]. While well received, its scan speed (up to about 80 ms) is still not sufficient to image patients of fast or irregular heartbeats. With the rotation speed of the gantry (0.25 s per revolution) reached an engineering upper limit, the physical obstacle for using more conventional sources in the current CT architecture is the limited gantry space [51]. On the other hand, respective simulations showed that doubling the number of source-detector chains on a given gantry is more efficient to reduce CT scan time than reducing the gantry rotation time by a factor of 2 [52].

Realizing the limit of the rotating-source architecture, the concept of stationary-source CT using spatially distributed X-ray sources has been proposed and some have been developed including the electron-beam CT (EBCT) [53] and scanning beam digital X-ray (SBDX) [54, 55], as illustrated in FIG. 4. Both the EBCT scanner and the SBDX tube use an electromagnetic force to steer the electron beam to different spots on a large X-ray anode to produce a scanning X-ray beam, very much like the design of a cathode ray television tube. Such X-ray tubes are in general large and have a limited range of viewing angles due to the difficulty in steering the high energy electron beam. Recently a multi-source inverse-geometry CT (IGCT) is being developed with a high power (up to 60 kW) distributed X-ray source [56, 57]. Unlike the sources in EBCT or SBDX where the electron beam is originated from the same cathode and steered toward different spots on the X-ray anode, the source in IGCT produces X-ray radiation by extracting electron beams from multiple cathodes and sending each electron beam to a distinct focal spot on the X-ray anode one at a time. However, the distributed X-ray source in the IGCT still needs to be rotated, thus it will be subject to the same $g$ force as seen in the current rotating-source architecture.

 FIG. 4 (a) Electron-beam CT scanner, (b) scanning beam digital X-ray tube, and (c) multi-source inverse-geometry CT.
B. CNT based multi-source CT

From the example application of CNT X-ray source in stationary breast tomosynthesis described above, it is natural to envision a CNT based multi-source CT illustrated in FIG. 5. Similar to the stationary breast tomosynthesis system, the stationary linear CNT X-ray source arrays can generate scanning beams without rotation, which could lead to faster image acquisition. Furthermore, the three source-detector imaging chains could simultaneously acquire three projections, which will further increase the imaging speed. Imaging speed for the CNT multi-source CT was estimated to reach about 30 ms [58, 59], which will potentially allow all-phase cardiac imaging for all patients regardless of patients' pre-existing conditions [60, 61]. Preliminary research about the feasibility [62, 63] of such CNT CT system, as well as the image reconstruction algorithms [64-66], has been carried out by our group.

 FIG. 5 Concept of a stationary source CT with three stationary and linearly distributed CNT X-ray source arrays (red color), along with three small X-ray detectors.
Ⅳ. CONCLUSION

CNT X-ray source is a new X-ray source technology enabled by the advances in nanotechnology in the past decades, particularly in the growth of MWNTs with suitable morphology and assembly of such MWNTs into field emission cathodes with high current density and long-term stability. Compared to the conventional thermionic X-ray sources, CNT X-ray sources have the key advantages in allowing flexible and instantaneous modulation of X-ray exposures in space and in time. As a result, CNT X-ray sources have been successfully adapted into dynamic micro-CT imaging of small animal models of human disease, and stationary breast tomosynthesis screening of breast cancers. In the future, CNT X-ray sources may enable some other biomedical imaging technologies that are safer and faster, with one example being the CNT based stationary multi-source CT.

Ⅴ. ACKNOWLEDGEMENTS

The work was supported by Dr. Guohua Cao's CAREER award from the U.S. National Science Foundation (CBET 1351936).

Reference