validating a LabVIEW application for medical devices

[hotshot]

In this case I actually do need to validate LabVIEW. Thanks for all of your help! I really appreciate it.

[PaulieQ]

---

@rolfk wrote:

Why stop at VC++, Word or whatever and not require the validation of Windows and the hardware you run your OS on?

In short you don't validate the tools in itself but your application and of course the results it produces. A proper design process starts with a good (and validated) requirements specification. From that you create your application. Then you have a validation specification that describes all the specific steps to verify and validate that your application actually does what it is supposed to do. Here you work with well known devices and DUT's so you have control over the results you expect. And of course everything needs to be documented to the last point. All this specification documents and according validation and reports is what makes development for medical devices more involved and expensive than in some other applications.

---

Are you sure that you don't have to validate also the LabView? Finally the product runs in the LabView Environment.

I'm not sure, it's only a personal opinion, but I think everything from hardware to software requires validation.

I'll give you an example. For validating an electronic product for a certain safety level (lower requirements as medical application) every single component of it must be validated: starting from FPGA, VHDL code, Architecture and so on. I repeat: the FPGA must be validated!!

If I extrapolate it, in this case the PC, LabView and Application require validation. Finally LabView is not a tool but a part of the final application. A tool is the Editor.

[RavensFan]

---

@PaulieQ wrote:

Are you sure that you don't have to validate also the LabView? Finally the product runs in the LabView Environment.

....

If I extrapolate it, in this case the PC, LabView and Application require validation. Finally LabView is not a tool but a part of the final application. A tool is the Editor.

---

There is no way you can run your application in the LabVIEW development environment and claim the application is validated.  You've just allowed the user access to the source code that could be easily modified, either intentionally or unintentionally, thus invalidating and validation you've done to that point.

You have to ultimately compile and build your application into a LabVIEW executable.  That is what you need to validate, your application in the runtime environment as it exists in a given operating system.

LabVIEW IS the editor tool that gets you to your built application.  It IS NOT a part of the final application.

[PaulieQ]

---

@Ravens Fan wrote:

---

You have to ultimately compile and build your application into a LabVIEW executable.  That is what you need to validate, your application in the runtime environment as it exists in a given operating system.

LabVIEW IS the editor tool that gets you to your built application.  It IS NOT a part of the final application.

---

Oky. But as I said: A LabView executable needs LabView Runtime-Engine or Runtime Environment (or how do you like to call it).

Thus, NI Software provides an Editor and an Engine/Environment/SW-Support for running the executable LabView application. The application which has to be validated is made up from: User Code + LabView Runtime-Engine so both parts of the software requires validation.

[rolfk]

Obviosly! But you can't separate the two at all. There is no way in validating the runtime engine on itself. How to do that? Proof that there is no possible way to make it crash? Proof that it works in 2012 after the end of the days?

You simply test your application under all use cases you can come up with, and validate every measurement and report it produces with what you have defined in the requirements spec. If it still fails, you either have not made up a good requirements spec or validation spec.

[RavensFan]

And if you wrote the program in C, or Visual C, or Visual Basic, they all have runtime engines and a load of .dll's running as well.  Do you validate all of the them?  Do you validate the kernal and the .dll libraries of whatever operating system you are running them on?

[PaulieQ]

I don't say that I find normal to validate everything and I don't know for sure what does it mean to validate for medical purposes. 

But I know what means to validate something (electronic device) for a certain safety level: you have to validate everything. From FPGA or ASIC architecture, all electronic components, communication interfaces and VHDL code (each module separate). If you see which are the requirements for a it (and it's not that restrictive as medical) you will say that it's a completely lost of time. If I had to extrapolate to a more complex system I would say that it must be validated the PC and the entire software no matter what.

Maybe someone who knows more about this issue could clear things up.

[rolfk]

What you describe may be alright when validating a device that is actually used for direct application at a person in some ways, but if you plan to use LabVIEW on such a device you are on your own and most likely out of luck. NI specifically states in just about every document that their products are not meant to be used in such applications.

Using LabVIEW for testing such devices is however a very different thing, and that is what I, and the others in this thread were actually refering to.