Following our previous analysis of Nanox Imaging's patents, in this blog post we consider recent claims Nanox makes for their technology, leading up to and during the Radiological Society of North America conference. At Nanox's 'Our Tech' page, they list a series of purported advantages of their cold cathode Xray tube. In the following we address these in order, after which we look at a few more issues of interest.
Rapid time switching
"In a hot cathode, the acceleration of the electrons towards the Anode (thereby creating X-rays) is done by activating and switching a high voltage supply. This process takes time (on the order of milliseconds). In a cold cathode, one can set the high voltage and switch only the gate voltages – a process that can take only microseconds."
It is not clear how this is relevant to medical imaging; millisecond turn-on and turn-off times are quite sufficient even for the frame rates of a full body CT, which reaches at most tens of frames per second. Blur can be a problem due to the rapidly moving CT mechanism but it appears the limiting factor here is not the turnon/turnoff time but rather the amount of illumination possible to deliver in the short timeframes available (eg. 50ms less the turnon/turnoff time for a 20fps device).
NASA has developed a sub-nanosecond 100keV Xray source [2] thousands of times faster than the Nanox device.
All the other cold cathode vendors will enjoy the same (but quite possibly irrelevant) fast turnon and turnoff.
Rapid intensity change / independence of current and voltage
"The Field Emitter current depends on the applied voltage, and the X-ray intensity depends on the Field Emitter current. With cold cathode technology, it is possible to uncouple these two parameters, i.e., the current's power is independent of the voltage. Thus, the X-ray intensity can be rapidly controlled by increasing the speed of switching or by creating short pulses."
This is a somewhat confusing statement given that the cold cathode technology uses field emitters while the 'hot cathode' uses thermionic emission. If the claim is that for field emission intensity depends on emission current, which depends on applied voltage, then these parameters are by definition dependent - but then Nanox goes on to claim these are independent. So let's assume for the moment that the statement should actually claim that thermionic current depends on applied voltage, and X-ray intensity depends on thermionic current, while for field emission these are independent.
Lastly, the title of the section, 'Rapid intensity change', seems to have nothing to do with current-voltage independence, which this sections analyzes.
The graph from Nanox's site shows that Ia and Ig currents are independent of acceleration voltage past about 10kV. If we assume that Ia is anode current (which is proportional to emission current and Xray intensity) and Ig is gate current then indeed the field emitter current is independent of acceleration voltage (Note - it seems that whoever produced this graph forgot to use a negative exponent for his current units.)
However this is largely the case for thermionic emission as well, see graph below. The graph shows the anode current Ia as it depends on acceleration voltage Ua, for a series of different filament currents If.
Note that the Nanox graph shows a current of Ia=0.01A (10mA or 10^-2A) which on the graph below lies squarely in the region of constant anode current with varying acceleration voltage.
Even if this were not the case, the fact remains that with control over both filament current and acceleration voltage, any cathode-anode current (proportional to X-ray intensity) can be reached for any given acceleration voltage. One can think of it like this : the X-ray operator has two knobs (acceleration voltage and filament current) and by spinning these knobs, can reach any point in the Ia-Ua graph.
Thus it appears that:
Thermionic sources also have anode current independent of acceleration voltage (at low currents).
Even if anode current and acceleration voltage were not independent, it does not appear relevant since any intensity can be reached at a given acceleration voltage for thermionic sources, by changing filament current.
Voltage-current independence does not seem to have anything to do with 'rapid intensity change'.
Colder Mechanism
"Electrons are extracted from the metal cathode by an applied electric field, while the emitter temperature is significantly lower than that of Hot Cathode (thermionic emission) filaments. Hot cathode temperature is over 2000 degrees Celsius while a Cold Cathode temperature is that of room temperature." [11]
"The typical technology requires a massive machine, because it has to manufacture a lot of heat (up to 2,000 degrees celsius), and then cool itself down. The Nano-X device... distributes this task of creating electrons across 100 million digitally controlled nanocones. That's what makes this X-ray a lot smaller, and a lot cheaper to manufacture and maintain." [9]
It doesn't appear that Nanox has in fact found a way around heat production at the anode; they do avoid heat production at the cathode since there is no heated filament. But the heated filament of a conventional tube is a relatively minor energy input (30-150W [3]) into the entire heat load (of ~100KW in a CT tube; for low power/intermittent use tubes, these heat loads are inconsequential.) A Nanox device of a given power must still reject the same amount of heat as a conventional tube, since it generates X-rays by the same inefficient process of braking radiation. As is clear from their figure [11], the entirety of the process beyond the cathode is identical.
Longevity/Lifetime improvement/Filament Burn
The hot cathode's longevity is of thousands of patients' lifetime, whereas that of the Cold Cathode is > 1M patients' lifetime. [11]
If true, this would be great, although it is somewhat offset by the fact that the tubes are also supposed to be cheap. Replacing $100 tubes a few times a year vs. a few times a decade would hardly make a difference to the overall operating costs.
Current CT tubes, which have the most demanding requirements (being high-power devices) have a lifetime expectation of 10,000-40,000 hours operation, enough for hundreds of thousands of patients. In any case the Nanox tube is in a different ballpark entirely, being a low power (~100W) device and therefore competing with other devices of similar or lower cost (see below).
Cost
A key Nanox claim is that their $100 tube replaces a $150,000 CT tube. As detailed in part I of this 2-part series, this is an apples-to-oranges (or perhaps grapes to watermelons) comparison of the ~100W Nanox tube to a ~100kW CT tube. The conclusion that the Nanox tube ~100W is based on images such as the following (where 40KV can be multipled by 2.5mA to find 100W).
A conventional 1000W tube suitable for dental Xrays can be obtained for ~100$, as shown here:
Note that Nanox hasn't developed a detector, only a source. The Nanox.Arc appears to use the Exprimer DRTECH detector below the table in their latest videos, thus the entire device will have to be at least this price. The Exprimer sells for tens of thousands of dollars. One radiology company CEO estimated that detector, bed, computer, software, and arc bill of materials would be at minimum $40,000.
Efficiency
'High efficiency due to sturdy gate design'.
Its unclear what this might mean; Nanox relies on the same highly-inefficient process of braking radiation as regular tubes, with 99% of the energy going to heat and 1% going to X-ray production.
Uniformity and Stability
These are minimum requirements for a usable device, and do not appear to signify improvements over extant technology, which is also uniform and stable.
Power Management
"Supply of less than 50V is required to activate electron ejections from the chip and enable voltage-independent current."
The control electronics for a 50V gate are indeed somewhat simpler than that required to switch a 50kV source, but suitable AC-DC power supplies were available since the 1950's, and a later generation of high-speed switching has been available since the 1980's.
The independence of current and voltage was treated above.
(In)advisability of yearly scans for early detection
"One screening for each symptomatic patient every year"
Using Xray for early detection has a drawback, namely that Xrays carry some degree of cumulative exposure risk. The more scans you have, the more likely you will get cancer.
"If you start at age 45, and have them [full body CT] annually until you are 75, you are talking about a one-in-50 chance of radiation-induced cancer, which is a huge risk. Until the benefit is clear, there is not much of an advantage to having routine body scans yearly or even every two years." [1] Obviously, routine screening for risk groups is a different matter.
Intellectual Property
"As of July 8, 2020, we had three issued patents in the United States and eight provisional or pending U.S. patent applications. We also had three patents issued in each of Israel, Japan and China, three pending patent applications in the European Patent Office, three pending patent applications in Korea and six pending Patent Cooperation Treaty patent applications, which are the counterparts of our U.S. patent applications...We intend to continue filing for patents on new technologies as they are developed and to actively pursue any infringement upon our patents. We believe that our know-how and trade secrets represent de facto barriers to potential competition." [10]
From a detailed analysis of the Nanox portfolio it does not appear that they are in a strong position in terms of patents. None of the 13 patents studied show any particular breakthrough in dealing with known problems of nano Spindt emitters (current uniformity, lifetime), and none show any experimental work beyond simulations. Several other companies operate in the digital X-ray source space as detailed below.
Nanox pioneered the digital Xray source
The Spindt array was patented in 1970, and a number of companies have developed variations thereof for X-ray imaging. There are number of systems currently for sale using digital sources either using small numbers of macroscopic Spindt emitters, or carbon nanotubes as emitters. Carestream [6], Vec Imaging [7], Micro-X [16] and Adaptix [8] are all active in the digital Xray source space.
Multispectral Magic
Nanox Ceo Ran Poliakine has asserted in at least one interview [12] that “because it’s digital, it’s multispectral. You don’t need different machines to do different kinds of imaging.” X-rays, CT scans, mammograms not requiring breast compression, angiograms and fluoroscopies have all been claimed as within the purview of the Nanox Arc due to this multispectral operation.
However the spectrum (range of energies) of X-rays produced (by both cold cathode and hot cathode sources) is almost entirely determined by the 'braking radiation' process wherein high energy electrons hit the metal (usually tungsten) anode and release X-rays. The spectrum has a characteristic form as shown below.
Here the intensity vs. energy (aka the spectrum) is shown, with a broad braking-radiation (aka 'brehmsstrahlung') curve and several peaks due to atomic transitions (aka 'characteristic radiation'). The point is that there is a wide spectrum produced by braking radiation, which is the process occurring in regular 'hot cathode' tubes as well as the Nanox 'cold cathode' tubes - so there is no difference between the Nanox spectrum and a regular hot cathode spectrum. In any case the ability to discriminate between different energies doesn't depend on the source (unless it has been filtered to produce only one energy) but rather upon the detector. This ability is very useful (allowing for instance identification of different atomic components of a sample in an EDX-equipped electron microscope) but has nothing to do with the source.
A transmission anode with field emitter was shown to be useful for medical imaging [13] but Nanox uses a standard anode, not a transmission anode.
FDA
Nanox submitted a 510-k application for a single-source version of the Nanox.ARC under the (the FDA's 510-k Third Party Review Program in January 2020. This submission was denied for the lack of information concerning safety, efficacy and durability of their product when compared to an FDA approved product. The submission is being advised upon by Dr. Daniel G. Schultz, former FDA director of the Center for Devices and Radiological Health, who at one point approved a device over the unanimous objection of his staff at the FDA [14] and now is part of an FDA regulatory consulting firm[15].
Nanox plans to submit an additional 510-k for the multiple-source Nanox.ARC during Q4 2020. As of August, they wrote "...we have not obtained feedback from the FDA regarding our regulatory strategy."
CEO's record
Ran Poliakine was sued for mismanagement of his previous startup Powermat:
“The dreadful financial situation of Israeli company Powermat...is a result of failing management at the company headed by founder and CEO Ran Poliakine, who manages the company improperly and unlawfully, in breach of his duty as an executive officer, by treating the company as his own, providing favors to his close associates, mixing together various businesses and companies he controls and the company, and taking advantage of company resources and the company's employees” reads part of the complaint brought by US billionaire J. Christopher Burch, an active investor, fundraiser and entrepreneur, who has been on Powermat’s board of directors since 2007, and is a shareholder. [4]
The suit stated that Poliakine approved a $54 million acquisition of HoMedics, in which Poliakine and entities affiliated with board members had stakes not disclosed to the broader board. The suit held that this was “to enable them to collect personal profits, at the expense of the company and its shareholders’ wellbeing...causing significant financial losses to the company.” Powermat allegedly wrote down 95% of HoMedics’ inventory within three months.
Burch claimed he invested on the basis of forecasts that Poliakine made of up to $400+ million in annual sales based on “the vast business potential of the company”, while actual revenue ended up being less than 10% what Poliakine had claimed. Powermat “wrote off at least $200 million in investments” and failed to make a profit. [5]
.