Home Forums Discussions General Discussion OCR and Imported Items; Pros and Cons

There have been a couple of mentions here of using OCR to create searchable files for AC imports in recent weeks. For example, Bert said "it is possible, as we have done, to convert all of our files (> 6GB) to OCR."

I am interested in the thoughts of Bert and others about the value of this, as well as any potential issues that would be created. In theory, it would be great to be able to search imported items for specific terms, and create reports from these (as can be done elsewhere in AC). On the other hand, I am not sure that such searches would yield much useful information, and I am not sure reliable reports could really be created.
On the negative side, errors would invariably introduced as OCR is never perfect. Also, one runs the risk of creating documents that can be altered which is a problem for a medical record.

While we are at it...could someone explain the definition of an "OCR-pdf" and what that means, practically?

I consider OCR PDF or OCR Document I think of a desktop software program and allows you to convert documents into searchable or editable text if your PDF application or other document software cannot do this.

It is also used in Scanners or fax machines (more often scanners) where documents which are scanned to PDFs or other documents are automatically made searchable or editable by the use of the OCR.

OCR also allows PDF to be marked up, e.g. using a highlighter that one would otherwise not be able to do.

There are two types of PDF (probably more, but hey I'm only a doc).
There are, in effect, searchable and non searchable.
The former include the actual text embedded in the file, and are often PDFs created by programs like Word ( and some print to PDF utilities) and of course Adobe. They are usually smaller in size.
The latter are just images ( in TIFF format ). It is these that need to be OCR'd, they come from fax machines and scanners, etc. most of your items will be these.
there are (having seen web sites for) utilities that turn the image only into searchable. They OCR the image and embedded the text, but remain PDF files.
PDF files are not tamper proof (witness POTUS birth certificate), so the security of medical records is really moot.

Be careful of some OCR software as they replace the source file with the recognized text. The good thing about Acrobat is that it adds a text layer to the original document. Most new software does that nowadays. So even if something is really off in the text layer (copy+paste), you can see the original document.

I'm not sure if Acrobat Standard has it, but you can batch recognize text in Acrobat Pro. (I don't think it does.) I.e. go back and OCR all of the II. You can even set up a batch sequence that will monitor a folder and will recognize text within the PDF documents arriving inside. Would be good for a fax board or PaperPort incoming faxes. Acrobat's OCR will even auto-straighten the original document. I do wish it were cheaper. You could probably just get one license for the server.

Does UpDox have OCR on their faxes? Do faxes come in a folder?

No, they don't. Not many programs do. Xerox and Toshiba MFC do.

Everyone should be aware that while Adobe Acobat 9.0 is about $350, you get two licenses. For printing to PDFs, as mentioned on here a million times and should be, is fine as far as working with a PDF, you simply cannot be Adobe. And, their support is very good.

One thing about Adobe is they are like Word. You use 1% and not the 99%. But, when you need the next 1% it is not. Adobe has a free (I think) site in the cloud, which you can upload to. And others can share it. It has the best security. I will many times make a password such as oriole, then send the password under separate cover to JBS, then password protect it. Works pretty well. There is extremely good encryption and public key to private key, but there you need others to know how to interact. This is where we are behind and this is where I think the government instead of doing CCHIT, MU, etc. should have been working on portals and exchange of documents. Just because the Pentagon can't do it, we should be able to.

Not sure what this is referring to.

Btw, if you're looking for Acrobat Pro. I'd recommend you try ebay before using Amazon or something. It'll cost you close to $400 on most retail sites. I grabbed Acrobat Pro X for $180 on ebay. That's 60% off the MSRP.

To be clear - are you saying, Bert, that the faxes that UpDox manages are not OCR-searchable?

Melanie

You'd have to ask an UpDox user, but my bet is that they are not OCR searchable.

They are not Updox searchable

We have two OCR threads running in parallel. I would like to pursue this one, though I will quote from another.

To answer the above questions about Updox, Sandeep (and Melanie), the faxes (or emails) come into a folder, and everything in the folder is "sucked-up" into Updox. I don't believe there is any OCR in the process (as Marty says). Once the fax is in AC, I don't think it is any different from any other II item. So you can use Adobe Reader to view it, and with a paid Adobe product you can search it, but nothing more.

I appreciate the descriptions of OCR-pdf. I am still unclear about exactly what benefits they provide. Sandeep, you say that Acrobat Pro "can batch recognize text in Acrobat Pro...I.e. go back and OCR all of the II." That speaks to file conversion, but how about searching?
I assume that you can search that individual file but am I correct that you cannot search across files in that chart, or across all patient's charts? If that is the case, what is the utility of the OCR conversion? Can you guys give me some examples of how you would use an AC that included all OCR-pdf's, as opposed to regular ones?
I know you can copy and paste from them; that seems of limited value.

Thanks, that's what I thought.

So, a revision to my plan...

What if I used the Brother MFS for incoming faxes like labs, then used software to convert to OCR (like ABBYY or OmniPage)..could the documents then be routed through Updox so that they can be directly sent to the pts. chart (and also send me a message in my inbox that I have something to review?)

I would have pharmacy faxes (since I don't need them to be searchable) get faxed directly into Updox and from there to the pts. chart for review.

Outside labs need to be searchable so that I can generate a report, for instance, of all results of a certain test (not for MY benefit, mind you - for the PQRS we have to do). So these I would have faxed to the MFS, get converted to OCR, and then routed via Updox to the respective chart.

Would this be doable, does anybody know?

Melanie

John,
Ahhhhh.... there may be the rub. I'm curious what Sandeep says. If it's only the text in an individual chart that is searchable, then there is no real benefit to OCR conversion. I was looking to find a way that would search across all pt. charts - so that if I wanted to find all files with the word "mammogram" in them, I could pull them up.
Every year, our IPA gives bonuses based upon how well we have our pts. meeting their metrics. Being able to search across all charts would be incredibly valuable and make the process so much easier...

I think that if you run a batch sequence, you can scan across multiple files! I just found this from the Adobe Acrobat site:

Run a Batch Sequence
Now, all you need to do is to run the batch sequence.

Place all the files you wish to process in a single folder on your hard drive.
In Acrobat Professional 7, choose Advanced?>Batch Processing
? or ?
In Acrobat Professional 8, choose Advanced?>Document Processing?>Batch Processing
Select the sequence to run
Click OK
Select the folder to process
Click the Select button.
Select the Output Folder


Melanie

Not that anyone seems to care about file sizes except for me, I believe that OCR PDFs are larger than regular PDFs which would expand the II folder more quickly. File size remains an issue for those of us who want to do off site backup using a (slowwwww) DSL connection.

When I originally looked at UpDox, I was told that they disassemble then reassemble files going through their service. I believe they are all image files, and so I wonder if a file would retain OCR if converted and then sent through.

Yes, the individual file is searchable. Like I said it would be impossible to a scanner to know the positions and definitions of every field of a medical chart in existence. You would be able to find text across PDF documents using Finder, but it wouldn't be in a database format.

I don't know how large your PDFs are, but some of ours can end up being 80+ pages. Having to scroll through and find a test result or active diagnoses can take a while scrolling through 80 pages. Or maybe I'm looking for a reference to something specific. That's where it's useful.

I'm sorry but that's not true. Acrobat uses an algorithm which is obviously secret, but it actually reduces the file size. For instance I just had a PDF document that 5.5 MB, I OCR'd it and now it's 3MB. So that's not really valid. Think of it this way a full page picture can be a few MB. A full page of text is 1KB. I don't know the exact workings of its algorithm. But you can even optimize for web publishing which really shrinks it down to almost nothing.

As to your backups, you should just not include II in your offsite backup ENC file. Let that be backed up by a third party application like Jungle Disk or CrashPlan. Those do incremental backups so you're not wasting time uploading the same stuff over and over again. II are not part of the database. (Even the SQL database can be incremental, but AC refuses to release the SQL SA password.) For now just make the II incremental. I could be uploading a few MB a day, but since they won't tell me the password, I have to upload my full database everyday. I do keep II incremental because like you my connection is slow and I had to adapt.

Kind of useful when populating demographic fields or other AC related fields.

Also highlighting text.

You can do this using the search function. Just don't be expecting a completed database. It will tell you of all the PDFs containing the word mammogram. Hit CTRL+SHIFT+F in Acrobat to give it a try assuming you have OCRd docs.

It's changed a bit. Now it's called Action Sequences in Acrobat X instead of batch sequence. I'm just using the term from the old days. This is how you would automate it. Keep the batch sequence running on the incoming fax folder. Everytime it runs, move completed OCR files to the real fax folder. Import as desired. Invisible to the user. To them it just looks like a fax has arrived but little do they know it's already been OCR'd Only available in Pro. Standard has the option to OCR all the documents in a folder, but can't be automated and finished files can't be moved.

If you want to be able to generate reports. You'd be better off using Tracked Data Fields in AC. E.g. recording HbA1c level. You could filter for all people above 6.5%. That's something that requires a database. Some users just use Excel/Access to do this, but why not just AC if you can.

But if you're just trying to find a word or something across a large number of PDFs, a search will do it. OCR is really meant to be used inside the document though. I doubt search HbA1c would return anything useful.

UpDox has no way of knowing which fax belongs to which person. Either way you're going to have to lookup the person whose chart you want to import the fax into. This will be the same for the Brother MFC or a faxboard.

Thank you Sandeep, for that clarification.

Melanie

You're welcome. Good luck.

Sandeep,

What happens when you move files to the batch convert folder and there is unreadable text in a document. Does it throw out a bunch of error messages that you then have to manually manipulate? What about formatting, say a lab report or path report with pictures?

Thx

By default, it will suppress all error messages. You can go and look for OCR suspects after the 1st OCR. But it's pretty good. I also have OmniPage for OCR and it doesn't even come close. That one throws out error messages nonstop. Acrobat will add a text layer to the existing PDF. If there is something like an image, you can just copy and paste it. It doesn't look like it retains formatting. Just a plain text layer underneath.

You can even have it down sample the original to reduce the file size of the output file.

Just to be clear by adding a text layer, I mean that the source PDF isn't altered. The text gets added underneath the image. So what you see doesn't change. It also auto straightens the pages.

Sandeep,

Do you have any experience with using barcodes? If I understand correctly, with a barcode, an image of a document is represented by numbers ( as opposed to characters with OCR).

Perhaps for this reason, barcoding is 100% accurate while OCR is not, as there can be errors made in recognizing some characters.

Melanie

Yea, bar codes don't give false positives. It's either all or nothing. I'm a bit confused as to what you're expecting acrobat to do. Before we go any further, what exactly are hoping will happen when you scan a bar coded document?