The Summary of the 510(k) submission by Nanox for its Nanox.Cart device was published last week. Here are some observations, in no particular order of importance (yet).
The name of the predicate device is wrong
The name of the predicate device, cleared under K021016, is AMX-4 Plus Mobile X-Ray System, not AMX-4 Mobile X-Ray System as claimed by the Summary. What else is incorrect, if Nanox cannot even get the name of the predicate device right? The "Plus" system is the upgraded model. The predicate of the Plus model is AMX-3 Mobile X-ray System, K802047, another GE system. The chair of Nanox Advisory Board is a former GE executive.
A micro-controller Arduino Mega 256 does not exist
Table 1 claims that Nanox.Cart uses a micro-controller Arduino Mega 256 that "controls the Nanox Cart X-ray System's functionality and GUI display." No such micro-controller exists. There is an Arduino Mega 2560 micro-controller board designed for hobbyists that uses the old and cheap ATmega2560 micro-controller released more than 15 years ago. Quite novel.
The target angle of 0 degrees in Table 1 is non-sensical (a typo) and contradicts the 16 degrees value in Table 2
The target, or anode, angle is a very important characteristic of an x-ray tube, as it determines focal size and beam width, strength and composition. At zero degrees, the tube will be completely unusable. It is one mistake that Nanox should not have made, if its "X-ray source technology [were] the basis of [its] business" (page 9, annual report).
The maximum tube voltage for the predicate device in Table 1 is incorrect
Table 1 claims that the maximum tube voltage for the predicate device is 100 kV, which clearly contradicts the 125 kV value from the "kV range" section in the same table. The actual value is 130 kV (from the tech specs of the HRT09 tube).
The power output of the reference device is annoying
Table 2 states that the power output of the reference device is 4.8 kW @ 104 msec, which is incorrect (a typo) and it should be "@100 msec," which is the standard (for example, IEC 60613:2010).
The x-ray source used by Nanox.Cart is still a mystery
There is no mention in the summary of any of the non-sensical descriptions that Nanox typically uses for its proposed x-ray source - digital, MEMs, silicon, semi-conductor, novel, etc. Table 2 claims that the "Nanox Tube" is similar to "Xinray CNT Tube," but that is incorrect based on the data in Table 2, as the CNT tube is 60x as powerful (4.8kW vs 0.08kW), capable of substantially higher tube voltage (110kVp vs 40kVp) and current. Table 1 mentions that Nanox.Cart uses a "Nano-x's
Cold Cathode tube" in the system description, but the tube type/model in both Table 1 and 2 is given as "Nanox Tube" (no cold-cathode here) and there is no tube model (Nanox' web site shows at least 4 completely different and incompatible "Nanox tubes" that look remarkably similar to regular industrial/dental hot-cathode tubes).
The mention of an x-ray source in the intended use is non-sensical
The description of the device's intended use begins with the non-sensical statement
The product is intended as an X-Ray source for diagnosis.
The product is a mobile x-ray system - FDA product code IZL - not a x-ray source (which almost exclusively means an x-ray tube in the context of modern diagnostic equipment - other sources could be radioactive isotopes, synchrotrons, etc). The product is supposed to include many more components other than an x-ray tube, as confirmed by the "system components" section in Table 1, for example, It appears this statement was intentionally inserted by Nanox to confuse investors and possibly subvert the 510(k) clearance process.
The single-source Nanox device is cleared only for hands, wrists, and fingers, on adult patients only
Both Table 1 and Table 2 claim that the intended use of the device is similar to that of the predicate and reference devices. But that is incorrect and contradicts the actual description of the intended use, as the device is cleared for a very limited subset of examinations, while both the predicate and reference devices can do all general purpose X-ray diagnostic procedures. In fact, the limitation for use explicitly states:
This device is not intended for general radiographic X-Ray examinations other than the indicated use...
So much for Nanox curing cancer.
The Nanox device is cleared to work with only one detector model, which appears unsuitable and has to be purchased separately
There is a bit of problem with the tech specs of the detector that Nanox has chosen to work with its device. The summary states:
The Nanox Cart is specified and designed to operate only with a Flat Panel Digital X-ray Detector Model EVS3643, manufactured by DRTECH Inc.
The summary of the detector clearance specifies that the X-ray system using it must have tube voltage equal or higher to 40 kVp, so Nanox.Cart barely complies (its tube voltage is fixed at 40 kVp per Table 1 and 2). What is more troubling is that the generator "mA Range" used in the detector clearance is specified as "10mA ~ 1000mA," which Nanox Cart fails to meet, as it cannot deliver more than 2mA (implied by 0.08 kW power output and 40 kVp tube voltage).
More importantly, this detector cannot be used for diagnostic purposes on a live subject by the proposed multi-source Nanox.Arc device, as it is too slow and takes about 5 seconds to capture and transfer an image. A 45-image tomosynthesis of a wrist, for example, would take at least 4 minutes, if the RSNA 2020 demo were anywhere close to reality.
Finally, the lowest quote for this detector, obtained in the gray market - new, but from unauthorized distributors and without warranty - is about $20,000. So much for being "cheap."
Many of the images supposedly made with the single-source device in the annual report and in investor presentations are likely fake
According to the annual report,
[Nanox has] generated the images below with the Nanox.ARC using a single X-ray tube on an imaging phantom (page 61).
However, none of these images were generated by the device that received clearance, as the device tube voltage is limited to 40 kVp (so the 50 kVp tube voltage in the images is impossible). Moreover, the device is not cleared for ankle/foot examinations.
Here is another image, from Nanox investor presentations, that is impossible to create by the device that got cleared.
First, the device is not cleared for shoulder examinations. Second, the 2.5mA reading exceeds the maximum device tube current of 2mA.
The mobility of the device is questionable
The device is cleared under the IZL product code, but it is not truly mobile/transportable. The device description states:
The system facilitates X-ray examinations in situations where it is not possible or feasible to transport the patient to a ward with fixed equipment
But the device has no battery, unlike its predicate - it is as mobile as the length of the cord (less mobile than a regular vacuum cleaner).
The device is "similar" to the predicate device, except that it is not
The section "Substantial Equivalence Discussion" is somewhat confusing. The section argues that the device is equivalent except that it is not.
The technical characteristics of the System are not different from the predicate device except for the fixed Source-to-image Distance, Field of view, aperture, focal spot size, and the fixed tube voltage and reduced maximum exposure current-time product.
Virtually all technical characteristics of the two devices are significantly different, and, it can be argued, raise many questions of effectiveness. Table 1, for example, confusingly states that the fixed tube voltage and current exposure time product (or charge) are similar to the significantly wider ranges that are needed in practice and can be obtained from the predicate device. For example, typical "technique charts" for digital detectors stipulate tube voltages of least 46 kVp for the intended use (adult fingers/wrist/hands), above the 40 kVp limit of the device.
The device requires cooling fluid
This must be surprising to Nanox investors who are led to believe by the CEO that a cold-cathode tube, even if real, runs somehow cooler than a regular hot-cathode tube of the same power.
The intended use contradicts the disclosures in the SEC filings
Nanox implies in its SEC filings that the device will not be commercialized, and so the statement that the indented use is to perform diagnostic radiographic examinations is misleading.
Specifically, Nanox states in its SEC filings:
the multiple-source Nanox.ARC [rather than this cleared Nanox.Cart device] ... will be our commercial imaging system (page 2, Prospectus).
Nanox has further revealed that, while not intending commercial distribution of the cleared device, it is using the 510(k) submission as part of its regulatory strategy, a step in
a multi-step approach to the regulatory clearance process (page 1, Prospectus),
where the apparent ultimate goal is to induce the FDA to clear the "the multiple-source Nanox.ARC" device by first creating a predicate out of the Nanox.Cart.
Therefore, any statements by Nanox about "indications for use" or intended use or intent to market the cleared device, other than an admission that the device is not intended to be marketed and the submission is simply a step in Nanox regulatory strategy, are problematic.
Update: Here is a cheap (dental) tube, Toshiba/Canon D-081B, that is used in other devices cleared under the IZL product code, that is smaller but much more powerful and much more useful than the proposed "Nanox Tube."
Update: Replaced "implied by 2mAs and 1 second" with "implied by 0.08kW power output and 40 kVp tube voltage" as it is the correct derivation for max tube current (sustained for 0.1s) - in this case, both derivations result in 2mA tube current.
Update: Nanox predecessor claimed in 2016 that the chip that forms its cold cathode can do 2.5A/cm2 (Nanox CEO was a Chief Strategy Officer at the time). If Nanox had made no improvements since, it means that the active area of its "chip" is now 0.0008 cm2 or a square of about 0.3 mm x 0.3mm. So why do the chips shown in Nanox annual report (page 66) and in a March 2021 tweet look much larger, at least 10 mm x 10 mm? Each covering an area that is at least 1000x the supposed area claimed in 2016 ...
Of course, as discussed elsewhere on this blog, Nanox has been unable to manufacture such a chip commercially (and so the proposed Nanox Tube is almost certainly not using any chip or any cold cathode), contrary to claims in its annual report (the University of Tokyo labs, which Nanox claims to rent, prohibit commercial use).
Update May 11, 2021: Replaced the image of the wafer from the annual report with an image from a tweet that shows the chip next to a ruler.
Update May 14,2021: Garage Blitz TV @Youtube makes a great point about the reference device using a CNT tube, which according to the Nanox annual report cannot work. Moreover, the Nanox "founder"
claimed in November 2019 that no such device exists, to the best of Nanox "knowledge."
