The difference between scanned vs digital lab reports is whether the file already holds the text. A digital lab report carries a text layer, so software can copy each test name, value and unit as written. A scanned report is a picture of a page, so OCR has to read it first, which adds errors and more review.
How to tell a scanned lab report from a digital one
A scanned lab report is an image of paper, while a digital lab report is a file a lab system produced with real text inside it. The quickest test is to try to select one result value with the cursor. If the value highlights as text and pastes correctly into a note, the file has a text layer.
The text layer is the hidden or visible layer of characters that a PDF viewer can select, search and copy. A report printed straight from a laboratory information system to PDF usually has one. A report run through a scanner usually does not, unless someone applied OCR afterwards and saved the recognized text back into the file.
Three more checks settle most cases:
- Search for a test name. Use the viewer's find box to look for a word you can see, such as "Potassium". No hits on a visible word means the page is an image.
- Zoom in hard. Digital text stays sharp at high zoom. A scan turns blocky or grey at the edges of each letter, and you may see specks, a tilted margin or a dark border.
- Paste and compare. Copy a whole result row and paste it into a plain text editor. A digital report pastes the row in reading order. An OCR'd scan may paste it with wrong characters, merged columns or missing decimal points.
The paste check matters because a text layer is not proof of a clean text layer. A scan that went through OCR has selectable text, but that text is the OCR engine's guess. Treat it as a scan with a draft transcript, not as a digital original.
Scanned vs digital lab report extraction: a side-by-side table
Scanned vs digital lab report extraction differs mainly in the first step. A digital report lets extraction start from exact characters. A scanned report needs OCR before any field can be found, and every OCR guess becomes something a person may have to check. The table compares the two on the points that drive effort.
The table below uses a fictional two-page chemistry report as the example. File sizes are example values for that report and change with scanner settings.
| Point of comparison | Digital lab report | Scanned lab report |
|---|---|---|
| Text layer | Present, made by the lab system | Absent, or added later by OCR |
| OCR needed | No | Yes, before any field can be read |
| Typical errors | Columns read in the wrong order, wrapped test names split across rows | Misread digits, lost decimal points, a "less than" sign read as a letter, units dropped |
| Review load | Spot checks on layout problems | Checks on values, units and flags, page by page |
| File size (example) | 90 KB for the two pages | 1.2 MB for the same two pages at 300 dpi |
| Page quality issues | None from the scanner | Skew, specks, faint print, dark borders |
The open-source PDF library pdfplumber states the same split in its own notes: the pdfplumber project README says it "works best on machine-generated, rather than scanned, PDFs." Tools that read text layers directly cannot see anything on an image-only page.
Digital reports still need checks. A layout parser can pair a value with the wrong row when a test name wraps onto two lines, or when a comment sits between two results. Those are layout errors, and they repeat the same way on every report from the same lab, which makes them easier to catch once you know the pattern.
Scanned reports add character errors on top of layout errors. The OCR errors that change a value, such as a dropped decimal or a misread less-than sign, are covered in the guide to OCR misreads on decimals, units and less-than signs.
How do you digitize old lab reports kept on paper?
To digitize old lab reports kept on paper, scan each report at a resolution high enough for small print, run OCR to add a text layer, and index every file by patient and date before anyone extracts a value. Old reports need extra care because the paper has often faded or been faxed before filing.
Scan settings for old lab reports
Resolution is the setting that matters most. The Tesseract OCR guide to improving quality says the engine "works best on images which have a DPI of at least 300 dpi." Lab reports often print reference ranges and footnotes in very small type, so scan at 300 dpi or more rather than at a lower office default.
The same Tesseract guide lists the page problems that hurt recognition. It warns that line reading "reduces significantly if a page is too skewed," that some noise cannot be removed during processing, and that scanned pages "often have dark borders around them" which can be picked up as extra characters. Remove staples, flatten folds, and feed pages straight so the scanner adds as few of these problems as possible. The full set of scanner choices is in the guide on how to scan paper medical records for OCR.
Indexing old lab reports
An index is the set of fields that lets someone find a file later. For a lab report, the useful index fields are the patient's name, date of birth, record number, the collection date and the performing laboratory. Index first and extract later, so that every extracted value can be traced back to a named file. The post on medical document indexing explains how to pick those fields and apply them during the scan.
A worked example batch
Here is a fictional batch, with every value made up for the example. A clinic finds a folder of 40 paper chemistry reports for one patient, Ana Example, record number EX-20417, covering five years. Staff remove staples, scan the folder at 300 dpi, and run OCR. They then split the stack into one file per report and name each file by record number and collection date, for example EX-20417_2021-03-08_chem.pdf. Only after the 40 files are named does anyone start pulling values.
Keep the paper until the scans are checked. Count the pages scanned against the pages in the folder, and open each file to confirm nothing is upside down or cut off. A missing back page is far easier to fix while the folder is still on the desk.
The broader steps for chart scanning projects, such as choosing which charts go first, belong to the medical OCR basics category.
When a digital lab report is still a picture: printed-to-PDF and faxed copies
A digital lab report is still a picture when it reached you as an image inside a PDF wrapper. The two common cases are a report someone printed to an image-based PDF and a report that travelled by fax. Both look like modern files in the inbox, but neither has a usable text layer.
A printed-to-PDF copy depends on how it was printed. Printing from a lab portal to a PDF driver often keeps the text. Printing a screenshot, or saving a web page as an image, does not. A file that came from a patient's phone camera is always an image, even when it arrives as a PDF.
A faxed report loses its text on the way. The sending fax machine turns the page into an image, and the receiving system stores that image. Small print, such as reference ranges and footnote markers, is the hardest part to read back. What to do with a fax you cannot rescan is covered in the guide to the best scan resolution for medical document OCR. Run the select-and-search checks from the first section on every faxed report, and route it through the same OCR and review steps as a paper scan.
A third case is a mixed file. A referral packet can hold a digital cover sheet, then two scanned pages of results, then a digital signature page. Check each page rather than the first page only. Lab results that arrive with referrals by fax are covered in more depth in the medical faxes and referrals category.
Which lab report source should you request from the lab?
The best lab report source to request from an outside lab is an electronic result feed, and failing that, a PDF produced directly by the lab's own system and sent electronically. A paper copy or a fax should be the last resort, because each one turns a text file into an image that someone must read back into text.
An electronic result feed sends each result as structured data, so nobody reads a page at all. The trade-offs of setting one up are covered in the comparison of HL7 ORU messages and PDF lab reports. For a small outside lab that sends only a few reports a month, a direct PDF is often the realistic request.
When you ask a lab for PDFs, ask for three things:
- Reports generated by the lab system, not scans of printed copies.
- One report per file, so a single PDF does not mix patients or dates.
- Delivery through a secure electronic channel that the lab and your office have agreed on, rather than fax.
The lab's own obligations help here. The CLIA test report rule sets the items every report must show, listed in the guide to lab report header fields and the CLIA report elements, and those items appear on a scan and a digital file alike. The digital file simply lets you copy them without retyping.
Old reports are another matter. The CLIA retention standard at 42 CFR 493.1105 requires the laboratory to keep or be able to retrieve a copy of each original report for at least two years after reporting. That rule binds the lab, not your office, but it means a lab may be able to resend a recent report as a digital file instead of you scanning a faded paper copy.
Once you know what format your reports arrive in, you can size the data entry work behind them. The comparison of outsourcing lab result data entry or automating it helps you weigh who should do that work. Schema Care is built for the case where results still come as pages: it reads test names, results, units, reference ranges and abnormal flags from lab reports and sends any field it is unsure of to a person, as the list of supported documents shows.