Journal: Proposal for Open Microscopy Initiative

My research utilizes custom-written software for automated time-lapse microscopy. It is a very specialized system designed specifically for Leica microscopes through the serial port interface. I have learned recently that the new generation upright series (DM6000) has removed the serial port interface, opting instead for a specialized USB port and protocol. There is a "virtual driver" that should be able to convert serial commands to the USB protocol, but I do not see that as a long-term solution. Of course, if my lab had all the time and money in the world we would dedicate the resources to write new code specifically for the USB port, but in my estimation it would take several man-months, if not a year. So, why is the subject of this message "Open Microscopy Initiative"? It is quite obvious to me that the reason Leica has abandoned the serial port interface - which is supported by many commercial software packages (OpenLab, Compix, Metamorph, etc) in addition to custom- written software like ours - is that they are basically going down the road of forcing us to use Leica proprietary software. Ok, so you ask what is the problem with this? Well, for one, their software absolutely cannot do what we need it to do. My research would stop dead in its tracks if we did not have the ability to use our own software to control the Leica scope. I don't think this is just a Leica problem. Zeiss, Nikon, and Olympus appear to be going down similar paths. What is my answer? Well, this might sound crazy, but I would propose that the US government - through the NIH, NSF, etc - mandate that any public research funding that is spent on microscopy equipment must support "open hardware standards". In other words, if Leica wants me to buy their microscopes, they would required to make their hardware open. Let me state the advantages, and then I'll address the problems. The advantages are as follows: 1) Open standards are simply more transparent and less resistant to rapid change. Some may disagree with the benefits of slow change, but if you have "mission-critical" software - it is a nightmare if the proprietary protocol you depend on changes from one hardware version to the next. 2) Software reuse. The advantage of software reuse is self-evident. Why should different labs re-create the wheel for every different brand and model of microscope. There are few things that these microscopes actually do differently (open/close shutters, change cubes, move stage), yet each company has its own proprietary way of controlling these functions. With a common standard, we could write the software once, and any hardware changes - as long as they aren't functional - wouldn't require changes to our code. To give an analogy, look at the HTML standard. Browsers have come and gone over the years, but for the most part, if you wrote HTML code 5 or even 10 years ago, it is still readable. It is inconceivable to me that this is not the case for equipment that can easily cost hundreds of thousands of dollars. Another advantage of software reuse is that if I switch from Leica to Olympus or Nikon to Zeiss, I don't have to re-write thousands or tens of thousands of lines of code. 3) Cost. It should be clear that all the man-months that go into writing/ maintaining code costs money. Lots of money. Not only that, but for every computer programmer who is hired by a research lab, that is one less grad student or postdoc. Or even worse, students/postdocs have to do it themselves, which wastes valuable time that could be used for science training. A biologist should not have to know C++. For all these reasons and more, open standards would be good for science. So, what are the problems with my approach? Clearly, one can argue that this type of regulation stifles competition. Well, I don't see it that way. Open standards would actually create more competition - to develop better software. The companies know they can make a lot of money on the software, but they have little incentive to improve the software, because most people buy the scopes based on the optics, not the software features. Open standards would enable open source initiatives (for example, the micro-manager project at UCSF) to fluorish, because these programs to could spend more time improving the features, rather than spending the bulk of the time re-writing the same piece of code for the differernt proprietary platforms. If there were better open-source projects, the microscope manufacturers and the commercial software companies would have to improve their own products. Look at the competition Microsoft is dealing with from Linux and other open source projects. They have to improve their own product (forgetting about Vista for the moment) or risk losing market share. I don't know if the NIH/NSF are big enough to really affect the market for microscopy, but probably combined with other governments in Europe and Asia who fund research, this type of approach could be made to work. It would benefit us all in the long run.