Low Field Mri a Review of the Literature and Our Experience in Upper Extremity Imaging

1 Introduction

Magnetic Resonance Imaging (MRI) is i of the safest and most versatile biomedical imaging methods available. Higher field strength systems ( $>$ one.five tesla) can produce increased indicate-to-noise pushing voxel size as depression every bit hundreds of micrometers [ane]. Still, this capability comes at a cost of at least $1M per tesla [2] making these high-field systems inaccessible to those in depression-to-eye income countries (LMIC). The unequal distribution of MRI and other medical technologies throughout the earth has been well documented [3]. Even when technology such equally MRI is introduced into LMIC through donation or purchase, more than half typically falls into disrepair and more than a quarter may never even be operational [iv]. This is due to the lack of spare parts, consumables, trained technical staff, or reliable power in addition to the high initial cost of acquisition [five]. A sustainable solution is necessary if diagnostic imaging is to be a useful tool for clinical do in LMIC.

The term "sustainability" has been used in many different contexts, even as it relates to medical intendance. Post-obit a review of the uses of "sustainability" in medical research in which the authors attempted to aggregate the most common definitions used, this piece of work most closely matches the category of "Continued program activities" [6]. Hither we invoke "sustainability" in the most literal sense—nosotros need technology that tin can operate in LMIC in the long term.

A status representing high clinical need in LMIC is childhood hydrocephalus. Globally at that place are an estimated 400,000 new cases of pediatric hydrocephalus per yr with over ninety% in LMIC [vii]. In sub-Saharan Africa where postal service-infectious hydrocephalus is most common, there are approximately 180,000 new cases per year [8]. These infants require treatment to survive. In Hydrocephalus there is a buildup of cerebrospinal fluid around the brain and within the ventricles, creating excess intracranial pressure level. It is treated with either ventriculoperitoneal shunting [ix, x] or endoscopic third ventriculostomy (with or without choroid plexus cauterization, ETV-CPC) [11, 12]. For handling planning, computed tomography is oftentimes employed in LMIC since it provides practiced spatial resolution and contrast between brain and CSF, and is less plush than high-field MRI, all the same the ionizing radiation associated with this technology has been shown to be a high take a chance particularly for infants [13]. Ultrasound can exist a useful tool, but merely before the first year of life when the skull begins to fuse [fourteen]. Fortunately, the imaging needs for hydrocephalus handling planning are relatively straightforward—clinicians must visually split CSF from brain with plenty conviction to program surgery. While typical loftier-field images tin provide sub-millimeter voxel spacing, we accept suggested that considerably larger voxels (3 × iii × 10 mm³) could exist sufficient for treatment planning [15] opening the door for the apply of low-field MRI engineering science.

In that location has been recently renewed involvement in clinical low-field MRI equally an affordable and portable alternative to loftier-field imaging. Permanent magnet systems with main field strengths of 50 milli-tesla [sixteen], fourscore milli-tesla [17], and 64 milli-tesla [18] have all shown capability for clinically relevant brain imaging at varying levels of portability and toll. The system described by O'Reilley et al. (2020) has been fabricated available in an open up source forum for an estimated cost of x,000 euros (https://world wide web.opensourceimaging.org/project/halbach-array-magnet-for-in-vivo-imaging/). Coil-based low-field MRI offers another potential road to sustainable clinical imaging at fifty-fifty lower field force. Piece of work by Ref. [19] has shown impressive images of brain using a large Helmoltz roll blueprint (220 cm diameter) with field strength of 6.five milli-tesla. Although this system produces quality human brain images with a coil-based low-field MRI, information technology is not portable and consumes more iv kW of energy to generate the principal, static field. Previous work from our group has described a pre-polarized depression-field MRI (PMRI) specifically designed to assist with treatment planning for babe hydrocephalus in Uganda [15]. Details of this specific magnet design are discussed in Ref. [15] and in further general detail in Refs [20–22].

While the work in Ref. [fifteen] described the magnet pattern and testing for the PMRI system, the present work focuses on the radio frequency chain (RF chain) of electronics required to operate the PMRI arrangement. The motivation behind this piece of work was to pattern a rugged and low ability option for sustainable operation in low resource countries similar Republic of uganda and provide documentation that tin serve every bit a basic tutorial on how to implement the RF organisation. The RF chain is powered using two 12 volt tractor-style pb-acid batteries. We have advantage of the relatively low operating frequency of 180 kHz to allow components in our RF chain to remain unmatched to 50 ohm impedance. This allows the transmit (Tx) ringlet to be tuned to a minimum impedance, maximizing the current delivered to the coil. In improver, we utilize usually available creature enclosure fencing (craven wire) as a Faraday shield and compare the effectiveness over more traditional shielding options. Finally, nosotros testify imaging capability using basic geometrical phantoms filled with water.

2 Materials and Methods

The Drive-L spectrometer developed by PureDevices was used for our PMRI organisation. Figure 1A shows the cake diagram of the RF concatenation which includes a Transmit (Tx) amplifier, a Tx coil, a tuning network for the Tx roll, a receive coil (Rx), a tuning network for the Rx coil, a low noise amplifier for the Rx coil (LNA), and a shield surrounding the system. All analog-to-digital conversion takes place within the spectrometer.Figure 1B shows the RF characteristics of the Tx amplifier and LNA. The main magnet in this system generates a center field of iv.23 milli-tesla which corresponds to a Larmor frequency of 180.1 kHz.

www.frontiersin.org

Figure 1. (A) Block diagram of the RF chain in the PMRI organization; (B) RF characteristics of the Tx amplifier and LNA.

ii.1 Transmit and Receive Coils

Ii carve up RF coils were used for transmit and receive (Tx and Rx respectively). A saddle design was chosen for both since these coils are straightforward to build with copper wire and take a relatively uniform field. Table one shows the parameters of the two coils used. An initial scale step is necessary to ensure the Tx and Rx roll are decoupled. For a two-roll system, decoupling can be achieved past rotating the coils and then that their field directions are orthogonal to each other. This was done past connecting the Rx coil to an oscilloscope and passing a 180 kHz point through the Tx coil at 5 mV _pp with a function generator. When the coils are strongly coupled the indicate amplitude volition approach 5 mV _pp equally measured past the oscilloscope. Decoupling is achieved by rotating the Rx coil until a signal amplitude less than a microvolt is visible on the oscilloscope.

www.frontiersin.org

Table 1. RF coil properties.

Litz wire has been suggested to be a more efficient wire for low frequency RF signals since it reduces the loss in current density along the wire produced past the peel result. It has also has been shown to exist useful in low field MRI applications [23]. In a round conductor, current density is pushed toward the exterior edge every bit frequency increases due to the changing internal magnetic fields caused past the alternate signal. This increases the effective resistance of the usher. The optimal wire diameter should be twice the skin-depth of the wire material at the frequency of operation. A smaller wire also has increasingly small electric current carrying chapters, just if many small wires are soldered in parallel, the current can be divide between them reducing the impedance. The Litz wire used in our PMRI organization has 100 strands of AWG 36 wire.

2.2 Tuning Circuits

Both the Tx and Rx coil tin can be modeled as a resonant resistor-capacitor-inductor (RCL) circuit. For PMRI, the gyre should be resonant at the Larmor frequency in order to detect the small signals (μV) resulting from low static magnetic field. A tuning capacitor is added in parallel with the coil to shift the natural resonant frequency to the Larmor frequency, equally shown in Figure 2. A matching capacitor is typically added in series to match the impedance of the coil to 50 ohms. Add-on of this capacitor alters the tuned frequency and one must use an iterative process to tune and match. In the low-field arrangement at 180 kHz, matching is non a requirement and it may be desirable to take minimum impedance in the Tx coil while maintaining a college impedance in the Rx ringlet.

www.frontiersin.org

FIGURE 2. RCL circuit. Typical tuning of the RF coil uses a tuning capacitor in parallel (c _tune) and a matching capacitor in serial (c _match). In the low-field blueprint, Tx can exist tuned in series, providing minimum impedance, and Rx tin can be tuned in parallel, providing a higher Q.

It is a well-established principle in RF engineering that an impedance mismatch between input and output ports of two continued RF components leads to reduction in power manual efficiency. It has also been shown that matching impedance of source and load between each component in a chain of RF electronics provides the nearly optimal power transfer efficiency [24]. For historical reasons, most RF electronic components and cables are impedance matched to 50 ohms.

In that location are specialized cases where ports and cables are non matched to 50 ohms. Examination equipment that is sensitive to pocket-sized voltages may have high impedance (100 Mohm or more) so that voltage drops preferentially on the input port of the measurement device. In some audio/video applications 75 ohm transmission cables and impedance matching is used. For nearly mid- to short-wave frequency applications, fifty ohm matching is the standard.

The ability attenuation from impedance mismatch comes from the reflection of the transmitted wave at the input back to the source or to the surrounding space. When the reflected wave combines with the incident wave to create standing waves, peaks and nodes occur along the transmission line. The measurement of this miracle is frequently called the Standing Moving ridge Ratio (SWR) [25].

The result of SWR on efficiency is dependent on transmission frequency and the length of the transmission cable [25]. Significant reflections tin can be avoided if the cable is much less than one-quarter of the frequency wavelength. This dominion of pollex makes impedance mismatch possible at depression RF frequencies without sacrificing much efficiency. For example, at our typical operating frequency of 180 kHz, the wavelength is around one,600 m. Since the cables used in the PMRI system rarely exceed 1 yard they are not considered "long," and therefore free energy loss at the load due to reflection at this frequency is virtually non-existent. This means an impedance mismatch will non lead to significant power inefficiencies.

Relaxation of the impedance matching principle provides more flexibility in designing electronics that are cost-effective and low power. For example, if the transmit amplifier is not driving fifty ohms, it could be low voltage and still evangelize a useful corporeality of electric current. Since the impedance of a series RCL excursion has a frequency dependent minimum at resonance, a serial capacitor is used to tune the Tx coil, as shown in Figure 2. Conversely, the impedance of a parallel RCL circuit has a frequency dependent maximum at resonance, then the parallel resonant excursion is used to tune the Rx coil as shown in Figure two. Past creating a series resonant RCL excursion at the desired frequency, impedance tin be shifted to a minimum approaching the DC resistance of the Tx roll. Inefficiencies in passive electronic components and balance skin effect in the wire of the Tx coil make the impedance higher than DC resistance, but we were able to tune our scroll to 180 kHz at an impedance of five ohms, providing a $>$ 5X electric current increment per signal voltage over the fifty ohm, impedance matched case.

In our low-field system, a one milli-second pulse with a ane A_pp electric current delivers the desired π/2 flip angle. For a curlicue tuned to 180 kHz with an impedance of five ohms, this requires a 5 V_pp signal, or 25 W, compared to the 50 ohm impedance matched case, which would crave 25 V_pp and over 125 W. In addition to an increment in power, the required voltages and increased ability dissipation does deport meaning design implications for the Tx amplifier—namely number of standard op-amps required to drive the load and number and type of required ability sources for the op-amps.

2.3 Transmit Amplifier

The Tx amplifier was designed to operate with a minimum impedance Tx curl. The key aspect in designing an RF amplifier is choice of the right operational amplifier (op-amp). For the specific aims of this project the op-amp should meet the voltage and current requirements of the application, maintain signal integrity, be equally readily bachelor throughout the world as possible, be of the appropriate size to work without specialized equipment, and be affordable in depression quantities. The circuit design should also exist every bit unproblematic as possible to let for fewer components and low-tech construction or repair.

Texas Instruments (TI) op-amp OPA-549 was chosen to drive the voltage amplifier. The op-amp data sheet is available on the Texas Instruments website (www.ti.com) and the circuit design for the Tx amplifier can be found in the supplemental data. The op-amp is capable of 8 A continuous output and can be driven by upward to ±thirty volt. Open loop proceeds is linear at the frequency of intended use and noise voltage density is at a minimum of 70 nV/ $\sqrt{H z}$ .

Ability supply requirements are of import to consider in terms of rut dissipation and rails-to-rail voltage. Consider the example where one might want to drive a 50 ohm impedance matched RF coil at 180 kHz with an OPA-549 amplifier. If we want to bulldoze 0.5 A of current we need a 50 V_pp swing (25 volts in each direction). The runway-to-rail voltage of a differential amplifier (i.e. the voltage limit in the positive and negative direction before the point is clipped) is typically slightly less than the supply voltage, and is often specified in the op-amp documentation. This ways, for a single OPA-549, nosotros could supply this desired signal amplification with a ±thirty volt power supply, but nosotros would demand to consider a skilful cooling strategy for the op-amp. We would too exist utilizing a very pocket-sized amount of the available current output for the maximum voltage supply.

Alternatively, nosotros chose to build a bridged amplifier with two OPA-549 op-amps—ane per supply pole. This allows united states of america to divide the oestrus dissipation beyond two op-amps and reduces our need for cooling. It also maintains the current driving capability while substantially doubling the rail-to-rail voltage of the amplifier. This means that we could supply 60 volts to each side (roughly 120 volts rail-to-rail) and even drive the coil with 1 A of electric current. If more electric current is desired, op-amps tin exist added in series to each polar side of the pattern. This design meets the amplification needs of our system without requiring forced cooling strategies.

Since the Tx coil in our system was tuned to 5 ohms instead of 50 ohms the ability supply strategy can be simplified to two 12 volt supplies. We can supply the OPA-549 with ±12 volt to produce a runway-to-rail of 20 V_pp, or nosotros tin can bridge two amplifiers and supply each with 12 volt and drive a little over i A of current without excessive heat dissipation. This method has a few other advantages. Many other electronic components in the system are driven with 12 volt power supplies, which ways we tin couple them all to ii 12 volt linear ability supplies, or power the entire RF chain using 2 12 volt car or tractor batteries. This has many sustainability advantages when it comes to use in LMIC.

The Tx amplifier is capable of 8 A continuous (x A acme) output with a rail-to-rail voltage of around 20 volts before clipping. The Tx coil was tuned to five ohm impedance which allows for a maximum of 4 A delivered to the coil. For the spin-echo imaging used in the PMRI organization, minimum pulse widths of between 0.v ms and 1 ms are typically used for the π/2 pulse with 1 A electric current. shorter pulse widths could be desirable, in which example supply voltage tin can be increased or the amplifier can be bridged to increase the runway-to-rail voltage.

2.four Low Noise Amplifier

One of the most important components of the RF chain is the low noise amplifier (LNA). The LNA is used to dilate the modest indicate coming from the receive coil, which tin be as low as μV for PMRI. A proficient LNA design volition add the appropriate amount of gain over the bandwidth of interest without adding a meaning amount of noise. Unfortunately, it is difficult to tune all of these parameters simultaneously. A modest amount of noise is unavoidable, since these are agile electronic devices constructed using transistors. For RF signals, transistor performance depends on source impedance, frequency, and amplifier gain [26]. In guild to optimize noise performance, matching networks are often used to transform the source impedance from 50 ohms to the optimal impedance at the target frequency, often at the expense of gain. For most applications, the target source impedance is much more 50 ohms and has been suggested to be closer to 1 kohm [26].

Placement of the LNA within the organisation tin as well be of import. Long cables connecting the coil to the input of the LNA can attenuate signal, add thermal noise, and introduce other types of noise artifact such as interference from other systems or movement of wires. The LNA is often placed every bit close to the coil as possible, begetting in mind that the potent magnetic fields tin can induce unwanted currents through the Hall outcome [27].

Lastly, in many receive chains there are other components after the LNA which could also contribute to the corporeality of noise added to betoken after reception. This is often referred to as the dissonance figure (NF) which is a measure of the racket added to the signal by a circuit element. Every bit the starting time electronic component in the chain, the NF of the LNA dominates that of the other components and so greatest care must be taken in its design [27]. For our depression-field system, many of these pattern restrictions can be relaxed. For case, at 180 kHz a 1 m cable connecting the Rx coil to the LNA will add very petty attenuation and minimal resistance. Lower frequency besides reduces the likelihood of parasitic capacitance within the board layout, and so a more relaxed structure approach can exist realized. Unfortunately, most out-of-the-box LNAs are commonly optimized for higher frequency applications and can be quite expensive.

Nosotros designed an LNA to meet the specifications outlined in Table 2. The bodily build specifications are likewise listed for comparison. The circuit diagram of the LNA tin can be found in Figure 3A. The LNA is an instrumentation amplifier with a 60 kHz bandwidth Butterworth filter centered at 180 kHz on the output.

www.frontiersin.org

TABLE 2. LNA Blueprint Spec vs. Actual.

www.frontiersin.org

FIGURE 3. (A) The circuit diagram for the custom LNA design using TINA software from Texas Instruments. (B) The excursion diagram for the 4th order butterworth filter used at the output of the LNA.

The choice of op-amp for the LNA is the most important attribute of the pattern compages. A single-ended op-amp has a single input and output, both referenced to ground. This creates a unproblematic compages, merely does not pass up common mode signals arising from interference on signal lines. A differential amplifier has differential input and output. The output becomes the gain times the subtraction of the two input lines. This means any point common to both lines will exist removed from the output, which is useful in high external noise scenarios [28].

Instrumentation amplifiers have all the benefits of a differential amplifier with the added do good of unmarried-ended output and, ofttimes, lower dissonance [29]. An instrumentation amplifier op-amp (INA-103) from Texas Instruments (TI) was chosen for the LNA design. The instrumentation amplifier has 3 internal op-amps. The 2 input amplifiers act as differential input rejecting common way at a mutual-mode rejection ratio (CMRR) of 100 dB. The third stage combines these signals and delivers a unmarried concluded output referenced to ground. Another useful feature is that gain is set by adding a single external resistor at pin 14, while the internal feedback resistors are laser-trimmed (3mΩ ± 0.1%) which provides excellent gain balance betwixt the internal differential stages. INA-103 has excellent noise performance for source impedance below 10kΩ. In addition, the amplifier acts as low-pass filter for frequencies in a higher place 1 MHz. This is useful for eliminating high frequency dissonance, but with an operating frequency at 180 kHz nosotros chose to add a bandpass filter as well.

An active fourth-order Butterworth bandpass filter with two stages was designed and added to the output of the LNA, as tin be seen in the excursion diagram in Figure 3B. At high frequencies, passive filters are oftentimes designed using reactive elements such as inductors and capacitors. At low frequency, the inductor values can become too large to be practical, adding unwanted time delays in the circuit, and so active circuits are preferred. The bandpass filter was constructed with ii OPA-656 op-amps. It has a unity gain bandwidth of 500 MHz, vii $due north V / \sqrt{H z}$ input noise voltage, and a 60 kHz bandwidth around the three dB attenuation points. Table iii shows the specifications of each stage in the bandpass filter. While the 500 MHz bandwidth was non an important aspect of the blueprint consideration and is considerably wider than required, unity gain stability in the bandpass filter avoids amplification of dissonance added in the filter stage.

www.frontiersin.org

Tabular array iii. Bandpass filter specifications.

This amplifier was designed to replace a loftier end LNA made past Stanford Enquiry Systems (SR560). While the SR560 is an excellent LNA, it has a high NF (24 dB) at the frequency and source impedance of the PMRI application. An experiment was designed to compare the LNA designed in this piece of work to the SR560. A ane mV_rms signal was sent to the Tx ringlet, received past the Rx ringlet tuned at 180 kHz, and amplified with both the LNA and SR560. The output of the LNA or the SR560 was measured by a spectrum analyzer and the noise floor effectually the signal was recorded. Figure 4 shows the comparison. The LNA shows an comeback of 24 dBm over the SR560 in this application. This is an illustration of how simple, purpose built electronics tin can outperform demote-top solutions at a fraction of the cost. The LNA developed in this work costs around $100, while the SR560 costs around $5,000. It should exist noted that the SR560 was non designed to be a low noise amplifier at frequencies or source impedances as depression equally this application requires and we should not expect loftier performance for our awarding. This example is intended to illustrate the toll-savings that can be accomplished through simple, custom built designs.

www.frontiersin.org

Figure 4. A comparison of the noise floor and signal level between the custom LNA and the SR560. The custom LNA shows a 20 dB improvement in dissonance floor over the SR560. Data points were extracted and plotted from an analog spectrum analyzer.

The LNA is powered by ii 12 volt linear power supplies or two 12 volt batteries. An internal excursion is used to pace down the power voltage to the filter op-amps to ±v volt.

ii.5 Shielding

The final component of the RF chain to discuss is the shield for the system. For high-field MRI systems, a well shielded room is an expensive and necessary component in guild to accomplish the high SNR these systems are designed to exhibit. Although shielding for low-field systems tin can be much simpler and more affordable, it is however an important component of maximizing bachelor SNR that must be addressed.

The first consideration is the size of the shield. An excellent solution for reducing price with good shielding effect is to build an aluminum Faraday cage around the arrangement alone, leaving the room unshielded. This also allows for portability of the organization, an important component to sustainable blueprint. This design was used in [16] to great effect, producing good quality low-field brain images in vivo. I potential risk with this method is that i side of the shield must be left open for the patient to enter the machine. The patient also acts equally an antenna every bit he or she couples with the RF receive gyre. In [16], a conductive aluminum cloth grounded to the shield was used to cover nearly of the patient's trunk, which reduced dissonance by a factor of 10. The downside of such a shield is that the conductive cloth is quite expensive and could be hard to acquire in the developing globe. The small shielded enclosure is too hard to see inside and may not exist ergonomically designed for baby imaging.

Another selection is to construct a room sized enclosure, large enough to seat at least two adults (parent and technician) and image an infant. One potential advantage of such a system is that, with low-field MRI, ferric material that is relatively far from the arrangement will not distort the magnetic field or turn into a dangerous projectile. This means there is the potential to use ferric metal as the shield cloth, such as an blend of steel. In our feel, steel is typically less expensive and more readily available (especially in LMIC) than aluminum or copper and would exist a preferred option for sustainability reasons. A drawback of this pattern is that it is not portable, unless built around a truck-bed or in a trailer, and the larger size could incur more cost. The use of inexpensive steel materials may be able to offset the cost of larger size.

Another fortuitous aspect of shielding at low frequency, such as 180 kHz, is that it is much easier to construct an ergonomic enclosure with a focus on patient comfort without embedding the shield in the walls of the room. This tin be achieved by using perforated material as a shield. If designed properly, perforated material tin be every bit effective equally solid material and it allows for good airflow and reduces claustrophobia.

In electromagnetic interference (EMI) shielding theory, in that location are iii main aspects of the shield design that should exist considered as information technology relates to the target frequencies to exist shielded: 1) the shield material, ii) the thickness of the shield, and iii) the size of any gaps in the shield. In some applications, shields are designed to be effective for electric and magnetic fields, still for the PMRI system nosotros focused on shielding electric fields. Electrical interference interacts with the shield by reflection, absorption, and transmission [30].

Figure five shows a cantankerous-section of an EMI shield interacting with a dissonance bespeak exterior the shield. Depending on the shield cloth and frequency of the noise point, some of the incident moving ridge volition be reflected and some will exist absorbed. The corporeality of the incident moving ridge absorbed depends on the shield material and frequency of the wave, but also on the thickness of the shield material. To empathise this, nosotros need to look at how Air-conditioning current is carried in conductors.

www.frontiersin.org

FIGURE 5. (A) Cross-section of an EMI shield. Interference signals are shielded by external reflection or absorption of RF free energy. Depending on the efficiency of the shield, an attenuated interference bespeak is re-transmitted into the shielded space. (B) Shield exam setup. The coil was placed within each shield with coil cable grounded to the shield and continued to input A of the SR 560 low dissonance amplifier. A 50 ohm resistor was connected to input B of the amplifier. Output was measured by a spectrum analyzer.

A DC current uses the entire cross-section of a conductor to move charge. As frequency increases in an Air conditioning electric current, a back-EMF is generated the by the alternating charge through increasing magnetic field at the center of the conductor. This increased magnetic field looks similar higher and higher impedance to the current, and charge density is pushed toward the edge of the conductor. This issue besides increases the resistance of the conductor as frequency increases. This phenomenon is called the skin-result. The cantankerous-exclusive surface area in which more than than 37% of the accuse is being carried is chosen the skin-depth, every bit measured from the surface of the conductor. The equation for skin depth is [31].

where ω is the angular frequency of the Air conditioning signal, μ _o is the permeability of a vacuum, and σ is the conductivity of the material. From Eq. 1 nosotros meet that every bit frequency increases pare depth decreases. This is of import for the design of the thickness of the shield because information technology will decide the upper limit frequency that can be finer absorbed by the shield. If the shield thickness approaches the skin depth of the target frequency, the shield will non behave as an effective conductor and most of the free energy that is not reflected volition exist re-transmitted into the enclosure. Equally a rule of thumb, the shield should be more than five times the thickness of the skin depth [30]. Comparisons between frequency, material, and thickness tin be fabricated using the shielding effectiveness nomograms printed in Ref. [30].

For the PMRI signal, nosotros want to shield frequencies around 180 kHz. Since the Low Noise Amplifier has a 60 kHz bandwidth filter on the output, nosotros at least must shield frequencies that will be passed by this filter. In practise, this is an easy example to design for. If nosotros have the highest important frequency to be 240 kHz (a full filter bandwidth higher up centre), the wavelength of this frequency is over 1,200 1000. This means that, although the all-time case scenario shield is a solid box with no gaps, we should be able to apply a perforated material to allow for better airflow. When choosing material, an obvious choice is aluminum or copper. The skin depth of 240 kHz in these materials is δ _Al = 0.iimm and δ _Cu = 0.15mm. Since the PMRI system has such a low field and we plan to build an enclosure that is much larger than the five Gauss line, we might also consider steel with a peel depth of δ _Steel = 1mm.

Iii pocket-sized enclosures were constructed to test the all-time case to worst case scenario for shielding for comparison. Each box was ane × i × 1 ft^three in size. Box ane was made of x mm thick solid aluminum, box 2 was made of 10 mm thick perforated aluminum with x mm diameter perforations, and box iii was fabricated of steel chicken wire with a 12.7 mm foursquare grid design.

The tests were performed using a 50 ohm impedance matched single-ended saddle curlicue resonating at 180 kHz. Coil location in the room was consistent across tests. The coil was placed in each shield with the coaxial cablevision grounded to the shield. The SR 560 low noise amplifier made past Stanford Enquiry Systems was used to dilate the signal with a gain of 100x. The racket spectrum was analyzed with a spectrum analyzer. Each shield case was compared to the baseline noise of a l ohm resistor continued to port B of the amplifier. A 4th scenario was tested without any shield to plant the maximum racket picked upwardly by the coil in the environment.

It tin can be seen from Table iv that perforated and solid aluminum are the best performers, at least to the noise flooring of the spectrum analyzer, since they both match the dissonance power of the 50 ohm resistor. Interestingly, the chicken wire shows a 10 dBm improvement over no shield, and only lags the aluminum options by 5 dBm. If nosotros consider patient comfort, then the perforated aluminum option is better than the solid aluminum since it allows for air menstruum. The cost of the perforated aluminum is roughly x times that of the chicken wire and so chicken wire could exist an excellent option for a shielded enclosure on a sustainable system targeted to the developing world.

www.frontiersin.org

Table 4. Shield effectiveness comparing.

The shielded enclosure for the PMRI system was designed to be large enough for two adults to freely move inside the enclosure during imaging. Figure half dozen shows the footprint of the shield enclosure. In that location is at least 1 m betwixt the PMRI system and the shield to allow for a technician to access any function of the system with ample space. The enclosure is 188 cm tall—large enough to accommodate a tall adult. The shield frame is made from mutual lumber used for framing small structures (in this instance, 2 × four inch lumber) and brass screws. The shield material is craven wire as described above. The total material cost of the shield was $300 USD, a 10X cost reduction over a shielded enclosure with perforated aluminum.

www.frontiersin.org

FIGURE 6. Footprint of the shield enclosure (left) with epitome of constructed shield (correct). The shield was constructed using ii × iv southward and craven wire.

three Results

The PMRI arrangement can accomplish an SNR of ten for the cylindrical water phantom with a 64 × 64 resolution FOV of 25 cm^two and using 5 averages with 50 mT Bp field (20 Amp, 75 volt battery ability). Effigy 7 shows the cylindrical water phantom measured with 5 averages (A) and xx averages (B). Although the SNR roughly doubles by using twenty averages instead of 5 averages, the imaging time increases by iv. Effigy 8 shows images of a kiwi (A) and three modest h2o bottles bundled in a triangle (B).

www.frontiersin.org

FIGURE 7. Comparison between 5 averages (SNR = 23) (A) and an 20 averages (SNR = 39) (B) for a turbo spin echo sequence of image the cylindrical water phantom.

www.frontiersin.org

FIGURE viii. Images of a kiwi (A) and three pocket-sized bottles of water arranged in a triangle (B).

Imaging experiments were performed in order to testify the capability of the designed electronics. To determine SNR capability, a phantom was constructed using a plastic cylinder 5 cm long and 6.5 cm in diameter and filled with distilled water. This phantom is ideal for testing since information technology fills well-nigh of the Rx gyre described in this piece of work and is non long in the z direction. This helps to minimize inhomogeneity furnishings caused by unwanted Bm gradients in the z direction just still allows for a significant amount of bespeak for imaging in the 10-y plane. A turbo spin-repeat (TSE) sequence was used with a turbo factor of sixteen on a 64 × 64 matrix prototype. The turbo gene defines the number of TR blocks that volition be used in the TSE sequence. A turbo cistron of sixteen for a 64 × 64 prototype means at that place volition be four TR blocks. The Bp was on for 4 due south prior to imaging and in that location was a 550 ms expect time betwixt turning off the Bp and starting the imaging sequence. This await time causes a 27% loss in SNR due to T1 decay. The receiver bandwidth was 50 Hz per pixel with a π/ii pulse of 1 ms. The effective echo time was 80 ms.

Although this basic phantom is useful for troubleshooting, more circuitous phantoms were also imaged. To show adequacy of imaging a phantom with more complex internal structure, a kiwi was imaged using a TSE sequence with a turbo gene of 32 and with echo time of 50 ms, and Bp duration of iii southward with a wait time after Bp before imaging of 450 ms. The Bp was powered to fifty mT past batteries. Image resolution was gear up to 32 × 32 with a 17 cm² square FOV. A 75 Hz/pixel bandwidth was used with a one ms π/ii pulse. The image of the kiwi tin be found in Effigy eight.

In addition, three small water-filled tubes arranged in a triangle were besides imaged as tin be seen in Figure 8. The water tubes were each 5 cm long with a bore of ane cm. The triangle was imaged with the l mT Bp field and twenty averages using the same imaging sequence as the cylindrical phantom.

4 Discussion

Components of the RF chained designed in this work were successfully paired without the demand for l ohm impedance matching. Not merely did this reduce the complication of the excursion design and construction, it more importantly reduced the energy required to power these circuits. Our goal was to design an electronic system that could exist paired with the magnet described in [15] and which had the potential for utilise in hydrocephalus imaging. At the time of this writing, our colleagues at the Mbararra Academy of Science and Engineering science in Mbararra, Uganda have reproduced the coil-based PMRI system and electronics.

Despite this success, at that place is a drawback to the united nations-matched RF chain. When matching to 50 ohms, a component such as a ringlet tin can exist replaced or inverse without altering the experimental parameters or ability efficiency of the RF chain. When the RF chain is unmatched, adding of a new component with different RF characteristics from the old component will require adjustment and calibration. For example, switching a Tx or Rx curl in the organization will require recalibration of Rx pulse width and aamplitude to achieve the optimal flip bending. This could be a meaning drawback considering SNR is improved when scroll size closely matches the object it is measuring. Farther investigation should explore whether the ability and simplicity gained in the unmatched case prove beneficial fifty-fifty in systems that are permanent magnet based. These systems already take a ability advantage over coil based systems and an SNR advantage since they tend to produce larger static fields.

The imaging experiments in Figures seven, 8 demonstrate the capability of this affordable and low-power MRI system. The water phantoms show that with proper shielding and common post-processing techniques fifty-fifty depression-voltage signals take enough SNR to accurately epitome collections of water. The epitome of the kiwi in Figure 8A further illustrates the ability of this organisation to show contrast between regions of the kiwi with high water content and low water content. Other details, such as seeds, would be hard to come across at this resolution. Figure 8B shows that our organisation accurately captures the spacing and size of the water bottles, suggesting the possibility of 4 mm resolution with a 64 × 64 image.

In the context of hydrocephalus treatment planning, these are promising results, where we are near interested in accurately imaging large fluid collections and distinguishing them from brain. The water triangle phantom in Figure 8B demonstrates resolution accuracy that is on the verge of our target resolution for hydrocephalus imaging, as discussed in [fifteen]. While the Rx coil used in this written report is too small for infant head imaging, the residuum of the organisation is accordingly sized and future work could likely provide similar results with a larger coil.

Another of import aspect of this organization is that these images were generated using very affordable and common electronic components such as tractor batteries, hand soldered electronics, and chicken wire. We demonstrate that low-field MRI opens the door to system design that does not require the strict tolerancing and high-finish fabrication techniques common to the field of MRI engineering science. In this style, systems that tin can exist congenital and maintained in low-resource regions of the globe have the potential to help in clinical conclusion making equally with the instance of infant hydrocephalus handling planning in countries like Republic of uganda. To that end, the MRI system employing both the PMRI coil described in [15] and accompanying electronics are actively being reproduced by our collaborators at the Mbarara Academy of Science and Technology in Mbarara, Uganda.

5 Conclusion

RF concatenation pattern is an important attribute of an MRI system, especially at low magnetic field where SNR is also typically low. Despite this, uncomplicated electronics and affordable shield materials, such every bit chicken wire, can provide high enough SNR to visualize internal structures at a low resolution. By leaving RF chain components unmatched to 50 ohms, we farther simply the blueprint and reduce RF power requirements during transmit. The pattern strategy for the RF chain compliments our efforts to develop a sustainable low-field MRI for employ in the developing world.

While advanced low-field MRI solutions are becoming available that offering heady clinical possibilities for a variety of diseases, we accept demonstrated the capability of a low-powered RF chain for our portable PMRI system which was designed for treatment planning of hydrocephalus in low-resource settings. The power efficiency and simplicity, and low cost of the RF chain design offering the potential for a arrangement that tin be sustained in LMIC with primarily local resources. Further development and adoption of this engineering science will allow for access to diagnostics in rural populations where in that location are few current options that could have substantial clinical touch on throughout the globe.

Data Availability Statement

The raw data supporting the determination of this article volition exist made available by the authors, without undue reservation.

Writer Contributions

JH—Pattern, construction, and testing of the electronics and experiments described in this work. Master writer of the text. CZ—Blueprint, construction, and testing of the LNA and authorship of the section on the LNA.

FK—Pattern, structure, and testing of the LNA and authorship of the section on the LNA.

IM—Structure and testing of select electronics and experiments.

JM—Advisor to CZ and FK during design and construction of LNA.

JO—Design, construction, and testing of whorl-based magnet used in this work.

SS—Advisor to primary writer during this work.

Funding

Supported by Us National Institutes of Wellness grant R01HD085853. ClinicalTrials.gov registration number NCT01936272. Design and construction of the LNA supported by CONACYT-Paraguay.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed equally a potential conflict of involvement.

Publisher'southward Note

All claims expressed in this article are solely those of the authors and do non necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any production that may be evaluated in this article, or merits that may be made by its manufacturer, is not guaranteed or endorsed past the publisher.

Acknowledgments

We would like to admit the fabrication back up provided by the Learning Mill at Penn State and Bob Crable at the Electronics Research Instrumentation Facility at Penn Land. We also acknowledge Rod Kreuter for his design, fabrication, and testing support on the electronics outlined in this piece of work. Finally, Kristen Crable who worked on fabrication of electronics and shielding.

Supplementary Material

The Supplementary Material for this commodity tin exist found online at: https://world wide web.frontiersin.org/articles/10.3389/fphy.2021.727536/full#supplementary-textile

Supplementary Figure S1 | The circuit diagram for the Transmit amplifier.

References

1. Rutland JW, Delman BN, Gill CM, Zhu C, Shrivastava RK, Balchandani P. Emerging Use of Ultra-high-field 7t Mri in the Study of Intracranial Vascularity: State of the Field and Future Directions. AJNR Am J Neuroradiol (2020) 41:2–9. doi:10.3174/ajnr.a6344

PubMed Abstruse | CrossRef Total Text | Google Scholar

two. Klein HM. Clinical Low Field Force Magnetic Resonance Imaging: A Practical Guide to Accessible MRI. Cham, Heidelberg, New York, Dordrecht, London: Springer (2015).

Google Scholar

3.World Health Arrangement. Baseline Country Survey on Medical Devices 2010. Tech. rep. Geneva: World Wellness Organization (2011).

Google Scholar

half-dozen. Lennox L, Maher L, Reed J. Navigating the Sustainability Landscape: a Systematic Review of Sustainability Approaches in Healthcare. Implement Sci (2018) 13:27–17. doi:10.1186/s13012-017-0707-4

PubMed Abstract | CrossRef Full Text | Google Scholar

7. Dewan MC, Rattani A, Mekary R, Glancz LJ, Yunusa I, Baticulon RE, et al. Global Hydrocephalus Epidemiology and Incidence: Systematic Review and Meta-Analysis. J Neurosurg (2018) 130:i–15. doi:x.3171/2017.10.JNS17439