IEC 62304 & FDA: Software Development Methodologies for Class II- DICOM device

[Taimoore]

Hello Everyone,

we are trying to decide which Software Development Methology/ies or a mix of them would be most suitable for a CLASS-II Software medical device (Moderate Level of Concern) such as DICOM and PACS.

I know there are many methodologies available i.e. Agile, SCRUM, V, Waterfall etc but which one or mixture would be most suitable for our product development?

Anyone from the Class-II (moderate level of concern) want to emphasize? Bare in mind that we will be applying for FDA-510(k) and IEC 62304 certification for our DICOM/PACS medical software product.

Please provide your input and suggestions....

Thank you

best regards
Tim

[Stijloor]

Any suggestions for Tim?

Thank you!!

Stijloor.

[Marc]

'Bump". If anyone can help with this, I apreciate it! My Thanks in advance!

[sagai]

When you submit traditional 510k, you only shall comply with 21CFR820 (and some more CFRs you will be notified at the last 2 paragraph of you clearance letter from FDA).
In case you submit abbreviated, you shall comply with 62304 (you shall declare).
62304 does NOT set your development methodology, regardless it describes the implementation 62304 into waterfall, your method "only" shall comply with the processes/activities/tasks declared in 62304.
Be aware, in case you declare conformance with 62304, only fully conformance acceptable as it is clearly defined in the standard inself.
br
Sz.

[sagai]

moreover there is no such think like 62304 certification.

[yodon]

Agile and SCRUM are methods that are intended to solicit frequent feedback during development. They require developers and team leads that understand the methods and a management team that understands what they're getting. If your development team is not well versed in the methodology, I don't think it would be wise to try it out on a medical device.

If your development team is experienced in waterfall but you need to increase in-process visibility, you can take an iterative (or spiral) approach.

So I'd say it depends as much on your existing developmetn team as it does your product. As Sagai pointed out, in the end, you have a product that is required to be compliant with FDA regs.

[ErikaP - 2010]

We are a small software company developing a Class IIa (CE) MD.

As already mentioned in a previous post IEC 62304 suggests the waterfall modell as the best way to implement medical device software. This modell is also strongly recommended by our notified body.

However, our programmers is not found of the waterfall modell as a programming method as such. What we have done is to implement a waterfallish process with several phases: "Prototype" - "Requirement" - "Design" - "Implementation" - "Validation/Verification" - "Production" - "Maintenance". Throughout development we freely iterate between the first 5 phases. When coming closer to a product release we practice the waterfall modell by having specified toll gates for each phase in our process. These tollgates are verified before we moving on to the next phase.

Note: We have no FDA-appoval so I have no knowledge about the requirements of FDA.

Good Luck!

[Jerome]

We've experimented with some models such as waterfall, V-model, also more iterative ones (re-occurring V) but found all a bit rigid.
Via the ISPE(.org) I found that the medical and pharma industry is moving towards more dynamic models based on ASTM E2500-7.
(have a peek at: http://www.qbd-dtc.com/wp-content/up...rification.jpg (found via google))
This kind of model is more appealing to me as it integrates more easely with my standard QMS and gives more freedom to theengineer in the design stage.
For a full version of this guide go and get it at ASTM for less then $40,- (I'm not affiliated to ASTM in any way).
Hope it helps