I am developing a microscope system which generates a lot of data, e.g. 200 images totalling 1.5 Gb which I want to stack using ImageJ. Currently, ImageJ stacks only about half of the images and stops, with an ‘out of memory’ message. This happens even if I increase the available memory on my laptop. Soon, I will be dealing with datasets perhaps 5x larger than this. What specifications should I seek when buying a new computer to overcome this problem? Please note: I am a biologist/ optical designer with only a basic knowledge of computer hardware but I would really appreciate a steer before I buy a fancy system.
Did you try allotting more memory to ImageJ and see if the stack opens faster? You can find this under Edit>> Options>>Memory and threds.
Also what is the currently configuration you have on your laptop?
Before you invest a lot of money for new hardware I would also expore the possibilities of Imagej and ImageJ2 to load and access massive datasets (disk resident):
which is using ImgLib2 behind the scenes.
Which can be used within Scifio (a flexible, extensible framework for image I/O):
Related (opening images as Virtual Stacks):
Firstly, I agree with @Bio7 that it is worth exploring the more recent and scalable efforts which are being actively developed. There is substantial desire and effort in the ImageJ developer community to improve support for big data, so these plugins will certainly get better and better over time.
That said, the current ImageJ certainly works best if you have sufficient RAM to load your entire stack into memory. You say you will have ~1000 images totaling ~7.5 GB? A mid-to-high end workstation these days should be able to handle this quite easily. My main suggestion would be to ensure you purchase a machine with 32GB or more of RAM.
Thank you for your suggestions. I did allot more memory as you suggested and got more images loaded into the stack but there seemed to be a limit to the amount of memory ImageJ could use and I could not load the whole stack of images however high I set the memory size.
I think this may be because my operating system is 32 bit and I can try upgrading to 64. My laptop is a Samsung R530 with 3 Gb RAM and is 64-bit capable.
Thank you Ctrueden,
Your message is extremely useful. I will purchase a workstation with much more RAM and also enquire about upgrading my existing laptop to provide a portable display system for the smaller datasets. My current 3 Gb is clearly totally inadequate.
Thank you Bio7!
Your message is tremendously useful. Briefly exploring the links you gave has shown me that a great deal of effort has already gone into this and I feel sure that our problems have already been solved. In case you are interested, our application is a new type of microscope objective called the Mesolens which covers a field 6mm in diameter but with a much higher numerical aperture than any previous design with a comparable field size. When used as a camera lens, it generates images which are barely Nyquist sampled by a 150 megapixel ( chip-shifting) camera and are even larger in scanning confocal mode. This means that we can collect even bigger datasets of biomedical specimens such as embryos and tissues than those currently obtained with the lightsheet microscopy mentioned in your links. Our work will soon be published in a paper by McConnell et al. in eLife which is going through minor revisions just now.