5 April 2019

Together with CPR we Strengthen our Image Data Integrity

Image Data Integrity

Rapid advances in microscopy and scientific journals’ increasing requirements for image data can challenge even experienced researchers. Therefore, CPR and DanStem held a theme day on the topic with leading international experts. Here is their advice.

mikro.jpg

A lot has happened within microscopy in recent years: Researchers now have more and better opportunities to create images containing much more detail; more researchers use microscopy (to a greater extent) in their research, and the journals demand better documentation of the integrity of the images. 

Therefore, it is important to structure your work with image data right from the beginning of a project, to prevent issues concerning documentation and thus publication. This was the focus of the theme day, “Data Integrity in Image Analysis" held by CPR and DanStem last week.

The event consisted of a seminar and workshop visited by three leading microscopy and image analysis experts, namely Bernd Pulverer, Chief Editor of The EMBO Journal, Kota Miura, Freelance Image Analyst, and Christian Tischer, IT Engineer and Image Analyst from EMBL in Heidelberg. See the programme.

The Experts' Three Most Important Pieces of Advice
According to Jes Dreier, Image Analysis Specialist at CPR's and DanStem's Imaging Platforms, who helped organise the workshop, the day was characterised by lively debate on good scientific practice within microscopy and image analysis.

'Of course, the general rules for good scientific practice also apply to image data. It is about not using data to say something that is not true. But I think the participants had some 'a-ha' experiences concerning the practical requirements for managing image integrity. Documentation is so much more than keeping your microscopy data in a folder on your computer. You have to organise it carefully from the outset', says Jes Dreier. 

Jes Dreier and the other organisers of the day, Professor Claudia Lukas, Chief Operating Officer Gretchen Repasky and Microscopy Specialist Jutta Bulkescher, have summarised the experts' most important advice to researchers using microscopy images and other image data in their research: 

  1. Make sure to have your raw data in order. Always only analyse copies thereof. This way you can always present the original images taken on the microscope, in case of doubt.
  2. Document everything you do at every step of the image data processing process and use software that helps you do this, e.g. Fiji.
  3. Spend a lot of time during analysis to understand in detail how you process the images, and make sure you do not add anything unwanted. This requires insight and experience, and it is something the imaging platforms at DanStem and CPR can teach and help you do. 

Journals Check Images 
According to the organisers of the project day, the scientific journals tend to go to great lengths to ensure that images submitted have been processed fairly. During the day, Chief Editor of The EMBO Journal Bernd Pulverer explained that journal editors screen images for the most common errors. And they often discover breaches of image integrity requiring clarification. In 99 per cent of the cases, such breaches are merely human, accidental errors caused by e.g. a mix-up or something that has been 'lost in translation' during data processing. Such cases are usually solved by going through the raw data.

'Just before publication is not the best time for a bump in the road, but if you have documented your process carefully, it is easy to answer questions about your results or method – and otherwise you risk not being able to recall what you did. And the probability of mistakes in the first place is also lower if your data is well-structured, says Jes Dreier and adds:
'If you are working on a project where microscopy is one of the cornerstones, you will be working with a lot of images. You may have 10-20 different samples each with 10-20 different images, and then the “what, how and from where" is not something you can just know by heart'.​

See an interview with Jes Dreier