Yes use ImageJ or so to measure the graph precisely. There are also specific tools available that do the calculations from pixel distance to the scale on the graph for you. I usually calculate this myself. Because you know the actual axis you simply express the pixels in units per X and thus you can extract the "raw" data. Best! 

Agree that you can use digitizer software (freely downloadable from various sources) which imports the graphs onto a grid, which you can then use to calculate numerical values from. In meta-analysis, this is most useful when studies do not report on the Odds ratio/Relative risk and standard error which are required for meta-analysis (for instance using the generic inverse variance method), but present the data in graphical form only. The relevant reference for this is: Parmar MK, Torri V, Stewart L. Extracting summary statistics to perform meta-analyses of the published literature for survival endpoints. Stat Med 1998;17:2815-34. Additionally, in some instances when the required data is missing but other forms of data is available, you can use additional software which can calculate the required values from the available data.

I'm sorry I couldn't add anything to this conversation but thank you so much for some very helpful information. Kind regards,

Nicola

Our experience in Cochrane Schizophrenia Group shows that WebPlotDigitizer is the best option.

In the recent paper titled “Systematic review automation technologies” [doi:10.1186/2046-4053-3-74], the authors reported three tools as follow:

1. http://plotdigitizer.sourceforge.net/

2. http://digitizer.sourceforge.net/

3. http://arohatgi.info/WebPlotDigitizer/

http://arohatgi.info/WebPlotDigitizer/

http://digitizer.sourceforge.net/

http://plotdigitizer.sourceforge.net/

dx.doi.org/10.1186/2046-4053-3-74

The following options have also been mentioned by the contributors to the topic:

• ImageJ
• Getdata graph digitizer
• Measuring Tool in Adobe Acrobat
• dcsDigitiser Graph Digitizer

The most recent paper (2019) introduces a good program developed by people who have been involved in systematic reviewing:

Article The development and evaluation of an online application to a...

How to use "Digitizer software" ?

Farhad lists several digitiser software; I've used the Engauge digitiser software (http://digitizer.sourceforge.net/) which is very intuitive and easy to use. The instructions (along with FAQs) are on the website.

It's also possible to measure distances on a graph directly in Adobe Acrobat - I have attached a protocol (it depends on your version but it is a useful method as measurements can be saved with the pdf and accessed later)

I have made the same question a time ago. The response is... YES

Please, take a look in:

Silva V, Carvalho APV, Grande AJ, et al. Can data extraction from figures perform a meta-analysis? Cochrane Database Syst Rev 2012;(Suppl):45. DOI:10.1002/14651858.CD201200. In: 20th Cochrane Colloquium, 2012, Auckland. 20th Cochrane Colloquium, 2012.

1. Available in: http://2012.colloquium.cochrane.org/abstracts/can-data-extraction-figures-perform-meta-analysis

2. Available in: http://www.cochranelibrary.com/dotAsset/d5356a06-cddf-42c5-bf24-af7a6edbff99.pdf

Just as a curiosity, I think I was the first to have a published study on this matter. Please, find the attached paper.

I've found Plot Digitizer to be the friendliest software for converting graphs into data. You can simply load in the image file, select the base and maximum of the x and y axes, and then click the data points you want to convert into x,y coordinates. The data can then be copied into Excel or another program for further analysis.

@ Rosie Morland, Daniel Becker and All, thank you very much for your answers and comments.

I've found Adobe Acrobat, measuring tool, very easy for converting graphs into data. Plot Digitizer is also a friendly software for converting graphs into data.

I'm doing this now for a systematic review, but have found in the first two papers that the data extracted from the graphs (using both WebPlotDigitizer and imageJ) do not match the (few) numbers which are discussed in the text. If I create a graph in Excel and then measure it, the data are very close. But what I measure from the points in published graphs is way off, and not in any systematic way I can determine. Perhaps I am just unlucky :(

Adobe Acrobat's measuring tool is very easy for extracting numbers (Mean, SEM, SD) from graphs.

Just go through Rosie Morland's answer and the attachment.

I have used Adobe Acrobat's measuring tool only for graph mining to one of my recent Systematic Reviews (see the Link).

http://onlinelibrary.wiley.com/doi/10.1111/jphp.12398/full

There is software "dcsDigitiser Graph Digitizer" that I have developed worthy of trying. It needs only one click to automatically digitize a graph. YouTube shows how it is done.

https://www.youtube.com/watch?v=-GW5ETRWcBs

An example (attachment) how the graph values are mined using Measuring Tool in Adobe Reader.

The attached document shows the method of extracting the data (numerical) from graph, using Measuring Tool in Adobe Reader, when the y-axis is not originating from zero (0).

Webplotdigitizer is the best and free. It works online as a well as offline

https://youtu.be/PW1nYO8ayMc

This is so helpful! Thanks all.

There is one software freely available online whose name is "Dexter". You can use it for this purpose. In case you wanna perform this in Origin then it is very much easy and available in the versions above 8.5.1. Check out this link to quickly learn this method - https://www.youtube.com/watch?v=42tl-Rxmj2A

… 

Yes! You can use Plot Digitizer software.

Raul,

Use the software entitled "GetData Graph Digitizer"