October 23, 2012

Trying to grab that text? Use OCR!

Copy & paste - a docs best friend! Every now and then you will encounter situations where you can't use the copy & paste technique. Indispencible in the hospital to quickly write consultation texts, letters etc. - nothing is more frustrating as when the medical record system (in an hilarious attempt to stop data theft) cripples the copy/paste function.

Other scenarious where you might need to quickly get some text in electronic format is that when you get a medical letter from a foreign patient. It has happened to me a several times and once in the middle of the night where the patient had acute dyspnea.

Here is what you need to do to do online OCR, "Optical Character Recognition"

1. Convert the text to image

If the text is on your computer you can use any screen grab app to just capture a screen image while having the text open. A favorite of mine is Lightscreen, a portable app thus allowing you to install on USB drive or within your local folder without administration rights.

Text on paper needs to be scanned of course - easily done in modern offices. Make sure the new image is in high quality, PNG is a lossless (no pixels lost while compressing) format highly recommended.

2. Remove patient identifying information

As we are going to use an online application we need to delete any information that might identify the patient. Not that I distrust the website used, just a good habit "just in case".
Open www.pixlr.com - a free, online Photoshop mimic where you can easily manipulate your new image. Just remember to save it back to PNG format agin.

3. Convert to text

So here's the magic ingredient in the cake; www.onlineocr.net is an online tool which takes an image and converts all text it can find to electronic format. Why this one stands out from the other lots out there is that you can even select which language to use! Now that wasn't so difficult!

And now you can do what you want with your text. If you're translating a medical letter just take it to Google translate.