# How to Scan Paper Lab Notebooks for ELN Migration By the end of this guide you will be able to take a shelf of bound paper lab notebooks and turn them into a organized, searchable digital archive that is ready to load into an electronic [lab notebook](/knowledge/molecular-biology/lab-notebook) (ELN). You will know what resolution and file format to use, how to handle taped-in gels and folded foldouts, how to run optical character recognition (OCR, software that converts pictures of text into selectable text), how to name files so nothing gets lost, and how to verify that the scan set is complete before the paper ever leaves the building. Here is what to have on hand before you start: - A flatbed or sheet-fed scanner, or a document-capture phone app as a fallback for odd-sized pages - A workstation with enough free disk space (a single notebook scanned at 300 DPI often runs several hundred megabytes, so budget at least 5 GB per notebook) - Free software such as NAPS2 or SimpleScan for capture, ScanTailor Advanced for cleanup, and OCRmyPDF with the Tesseract engine for OCR - A spreadsheet or text file for the inventory and the manifest - Sticky flags (not tape) for marking pages that need special handling - Clean, dry hands and a pencil, never a pen, anywhere near the notebooks ## Why Scanning Matters Before an ELN Migration An ELN is only as useful as the historical record behind it. When a lab moves from paper to an electronic lab notebook, the old notebooks hold the experimental context that makes new data interpretable: the lot number of a reagent used in 2021, the hand-drawn calibration curve, the note that a run was repeated because of a failed water bath. Scanning paper lab notebooks for ELN migration preserves that context in a form the ELN can link to, search, and back up. There is also a records-integrity reason to do this carefully. In regulated environments, the FDA expects data to satisfy the ALCOA principles, meaning records must be attributable, legible, contemporaneous, original, and accurate. A rushed, low-resolution scan that clips page margins or drops a witness signature undermines all five of those properties at once. A disciplined scan project protects them. Scanning is also a hedge against physical risk, since paper notebooks are vulnerable to water, fire, and simple loss in a way that a backed-up digital archive is not. ## What You Need Installed The workflow below assumes a computer with a handful of free, widely used tools. Exact menu locations differ between versions, so treat the commands as representative rather than word-for-word required. **On Windows or macOS:** - NAPS2 (Not Another PDF Scanner 2) for scanner control, or your scanner manufacturer's software - ScanTailor Advanced for deskewing, cropping, and contrast cleanup - OCRmyPDF (installed with Python and Tesseract) for searchable PDFs **On Linux:** - SANE tools such as `scanimage` for capture - ScanTailor Advanced, available through most package managers - OCRmyPDF and Tesseract from your distribution's repositories You will also want a spreadsheet program for the manifest and a checksum utility such as `sha256sum` for verification. ## Step 1: Inventory and Prepare the Notebooks Scanning goes wrong most often in the preparation stage, not the scanning stage. Before any machine is turned on, work through the notebooks on the shelf. 1. Assign every notebook an identifier. A simple scheme such as `NB-PI-LASTNAME-YEAR` works, for example `NB-JG-OKONKWO-2019`. Write the identifier on a slip of acid-free paper tucked inside the front cover, not on the notebook itself. 2. Record each notebook in an inventory spreadsheet with columns for identifier, owner, date range, page count, condition, and any attachments. A realistic row looks like this: 3. Flag pages with glued or taped-in items, foldouts, carbon copies, and damaged corners with removable sticky flags placed on the edge of the page, never on the written surface. 4. Remove loose items such as gel photos, sticky notes, and printouts, and scan them as separately labeled attachments. Keep them in labeled archival folders matched to their notebook identifiers. 5. Photograph the notebook cover, spine, and any signature or witness page before scanning the interior. Those images become the first pages of the digital record. An inventory entry should look something like this: ```csv notebook_id,owner,year_range,pages,condition,attachments,notes NB-JG-OKONKWO-2019,J. Okonkwo,2019-01 to 2019-08,142,good,3,"foldout gel on p. 58, loose SDS-PAGE photos in pocket" NB-JG-PRICE-2020,K. Price,2020-02 to 2020-11,178,fair,1,"water stain pp. 30-32, text still legible" ``` The `attachments` count is your completeness check later. If the inventory says a notebook has three loose items, the digital folder must contain three attachment files plus the main scan. ## Step 2: Choose Resolution, Color Mode, and File Format ![Four-step card showing resolution, color mode, lossless capture and PDF delivery settings for scanning lab notebooks.](/article-images/illustrations/scanning-paper-lab-notebooks-for-eln-scan-settings.webp) The two settings that determine whether a scan is useful long-term are resolution and color handling. **Resolution.** 300 DPI (dots per inch) is the working standard for text documents and is the setting recommended by major archives and the scanner industry for general document capture. Use 400 to 600 DPI when pages contain fine detail, such as small chromatogram peaks, thin pen strokes in pencil, or photographic figures you may want to reuse in a figure. Scanning at 72 or 150 DPI is the single most common way to make a scan project worthless, because you cannot re-capture detail that was never recorded. **Color mode.** Full color at 24 bits per pixel preserves everything, including blue or green ink, pencil graphite, and the faint color of a hand-drawn mark. Black-and-white bitonal scanning is smaller but destroys anything written in light ink or pencil. For lab notebooks, scan in color. Grayscale is a reasonable compromise for dense text-only pages, but color is the safe default. **File format.** Capture each page as a lossless TIFF or PNG. Lossless means no image data is discarded in compression. Save the final deliverable as a PDF, ideally PDF/A, the ISO-standardized PDF profile designed for long-term archiving. JPEG is a lossy format and should be reserved for photographs, never for text pages. | Setting | Recommended value | Use this when | Avoid this when | |--|--|--|--| | Resolution | 300 DPI | Standard text pages, tables, signatures | Pages have fine detail, then use 400 to 600 DPI | | Color mode | 24-bit color | Any page with color ink, pencil, stains, or photos | You only need bitonal text and storage is critical | | Page format | TIFF or PNG | All capture and archiving steps | You plan to compress aggressively to JPEG | | Deliverable | PDF or PDF/A | The final file loaded into the ELN | The ELN cannot ingest large PDFs, then split volumes | | OCR language | Match the notebook language | Any text-heavy notebook | The text is purely numeric or graphical | ## Step 3: Scan the Notebooks This is the core walkthrough for scanning paper lab notebooks for ELN migration, step by step. The example uses a Linux workstation with SANE and OCRmyPDF, but the same settings exist in NAPS2 on Windows under similar names. 1. Confirm the scanner is detected and note the device name: ```bash scanimage -L ``` Expected output looks like: ```text device 'genesys:libusb:001:004' is a Canon Canon LiDE 220 flatbed scanner ``` 2. Set the scan mode, resolution, and page size, and capture a test page: ```bash scanimage -d genesys:libusb:001:004 \ --mode Color --resolution 300 \ -l 0 -t 0 -x 215 -y 297 \ -o test_page.png ``` Here `-l`, `-t`, `-x`, and `-y` set the scan area in millimeters. The values above cover a full A4 page. For US letter pages, use `-x 216 -y 279`. Open `test_page.png` and check that the full page edge to edge is captured, text is sharp at 100 percent zoom, and handwriting is readable. 3. Capture each page as a numbered file. For a notebook scanned two facing pages at a time on a flatbed, one pass per opening is typical: ```bash scanimage -d genesys:libusb:001:004 \ --mode Color --resolution 300 \ -l 0 -t 0 -x 215 -y 297 \ -o NB-JG-OKONKWO-2019_p%04d.png ``` The `%04d` pattern produces zero-padded filenames such as `NB-JG-OKONKWO-2019_p0001.png`. Zero padding matters, because without it many file browsers and ELN importers sort page 10 before page 2. 4. If a page has a foldout, scan it in overlapping horizontal strips and keep the strip filenames sequential, for example `p0058a`, `p0058b`, `p0058c`. Photograph-based attachments get their own files, such as `NB-JG-OKONKWO-2019_att01_gelphoto.png`. A full notebook of 142 pages at 300 DPI color typically takes one to three hours on a flatbed, depending on page-turning and repositioning. Batch work by notebook, not by the whole shelf, so each notebook stays a coherent unit. **Phone capture fallback.** For oversized foldouts or notebooks too fragile for the scanner glass, a document-capture app that applies perspective correction is acceptable for reference copies. Photograph each page straight on, under even indirect light, with the phone parallel to the page, and record in the manifest that the page was phone-captured. Phone capture is a fallback, not the primary route for the archive of record. ## Step 4: Clean Up and Deskew the Images Raw scans are almost never square. A few degrees of rotation makes OCR fail on entire lines, so deskewing is not cosmetic, it is functional. In ScanTailor Advanced, load the batch of PNGs and apply, in order: deskew, page split if you scanned two-page openings, content selection to trim black scanner borders, and modest contrast adjustment. Do not aggressively binarize (convert to pure black and white), because pencil and light inks disappear. If you prefer a command line, ImageMagick handles rotation and a gentle contrast stretch in one step: ```bash magick NB-JG-OKONKWO-2019_p0001.png \ -deskew 40% -level 5%,95% \ NB-JG-OKONKWO-2019_p0001_clean.png ``` The `-deskew 40%` argument lets ImageMagick search for skew up to a wide angular range, and `-level 5%,95%` stretches tonal range without clipping faint strokes. Check one cleaned page against the original before batch-processing the notebook. If pencil text has vanished, back the level adjustment off. ## Step 5: Run OCR and Build Searchable PDFs ![Decision path for OCR results, covering a page that already has a text layer, handwriting and skew.](/article-images/illustrations/scanning-paper-lab-notebooks-for-eln-ocr-check.webp) OCRmyPDF takes a PDF or image set, runs the Tesseract OCR engine over it, and writes a PDF where the page image is preserved and an invisible text layer sits underneath it. That text layer is what makes Ctrl+F work inside a ten-year-old notebook. First assemble the pages into a single PDF per notebook, then OCR it: ```bash magick NB-JG-OKONKWO-2019_p*_clean.pdf NB-JG-OKONKWO-2019_raw.pdf ocrmypdf -l eng -r -j 4 \ NB-JG-OKONKWO-2019_raw.pdf \ NB-JG-OKONKWO-2019_searchable.pdf ``` Here `-l eng` selects the English language model, `-r` rotates pages automatically where the engine detects upside-down or sideways text, and `-j 4` runs four OCR workers in parallel. Expected console output looks like: ```text Output file is a PDF/A-2b because input was not PDF/A and textconversion needed it Postprocessing... OCR: 142 pages ``` Then spot-check the result. Open the searchable PDF, select a block of handwritten-adjacent printed text such as a reagent lot number, and confirm it copies out as real characters. Handwriting itself will not OCR reliably with general-purpose engines, and that is expected. The goal is that printed text, table headers, dates, and lot numbers become searchable, while handwriting stays as the image. A common OCRmyPDF error is: ```text PriorOcrFoundError: page already has text! ``` This means the source PDF already contained a text layer, often from the scanner software. Rebuild the PDF from the raw images, or rerun with the appropriate redo flag for the version you have installed, and check the OCRmyPDF documentation for your release. ## Step 6: Name Files and Build a Manifest File naming is the part of the migration people skip, then regret. Establish one convention for the whole project and never deviate. A workable pattern is: ```text NB-{notebook_id}_p{page}.png single page images NB-{notebook_id}_searchable.pdf final deliverable NB-{notebook_id}_att{nn}_{label}.png attachments NB-{notebook_id}_cover.jpg cover record ``` Alongside the files, keep a manifest that maps digital files back to physical page numbers, because ELN importers and auditors both ask for it: ```csv notebook_id,file,physical_pages,content_type,operator,scan_date NB-JG-OKONKWO-2019,NB-JG-OKONKWO-2019_searchable.pdf,cover + 1-142,notebook,A. Rivera,2026-11-05 NB-JG-OKONKWO-2019,NB-JG-OKONKWO-2019_att01_gelphoto.png,"loose, filed after p. 58",attachment,A. Rivera,2026-11-05 ``` Store the manifest as both a CSV and a PDF printout inside the notebook folder. When the ELN import runs months later, this file is the map. ## Step 7: Verify the Result Before Retiring the Paper Verification is where a scan project earns its credibility. Work through three checks per notebook. **Completeness.** Count pages in the PDF against the physical count in the inventory. A quick command-line count: ```bash pdfinfo NB-JG-OKONKWO-2019_searchable.pdf | grep Pages ``` Expected output: ```text Pages: 143 ``` That is 142 notebook pages plus one cover image, matching the inventory. If the numbers disagree, rescan the gap before the notebook leaves your hands. **Legibility.** Open a random sample of at least ten percent of pages at 100 percent zoom and check that signatures, dates, and small annotations are readable. Pay special attention to pages you flagged during preparation. **Integrity.** Generate checksums so you can prove later that files have not changed: ```bash sha256sum NB-JG-OKONKWO-2019_searchable.pdf > NB-JG-OKONKWO-2019.sha256 ``` Expected output: ```text c3f1a9...e77b NB-JG-OKONKWO-2019_searchable.pdf ``` Store the checksum file with the manifest. Anyone who needs to confirm the archive is unchanged can rerun `sha256sum -c` and compare. Keep the paper notebooks until the digital archive has been verified, imported into the ELN, and backed up in at least two locations. Many institutions have their own retention schedules for research records, so confirm yours with your records office or quality unit before discarding anything. ## Step 8: Import Into the ELN ELN platforms differ in how they accept bulk files, but the pattern is similar: create a notebook or project structure, upload the searchable PDF as an entry attachment or a linked document, and attach the metadata from the manifest. Before you commit to hours of uploads, evaluate the platform against what this project needs. A useful electronic lab notebook comparison checklist for a migration includes: - Bulk upload of multi-hundred-page PDFs without manual splitting - Full-text search across attached documents, not just across typed entries - Version history and an audit trail on the stored record - Export in an open format so the archive outlives the vendor - Permissions that mirror your lab's signing and witness structure As you structure the ELN workspace, many teams adopt a digital page layout inspired by a conventional lab notebook template, with dated entries, an experiment title, an objective, and a results block. Keeping the digital structure parallel to the paper structure makes the archive intuitive for people who remember the paper originals. ## Common Mistakes and Limitations Most failed scan projects fail the same few ways. These are the mistakes to design out from day one. **Scanning too low a resolution.** 150 DPI looks fine on screen and falls apart the moment anyone zooms in to read a margin note. Commit to 300 DPI minimum for the archive of record. **Compressing to JPEG early.** JPEG artifacts blur the exact thin strokes, pencil marks, and small numbers that make lab notebooks valuable. Capture lossless, and only produce compressed derivatives if the ELN demands them. **Skipping the inventory.** Without a pre-scan page count and attachment count, you cannot prove the scan is complete, and a missing page in a notebook is indistinguishable from a page that was never written on. **Not scanning covers and signature pages.** The cover links the scan to the physical object, and witness or approval signatures carry the record's authority. Treat both as first-class pages, not decorations. **Trusting OCR on handwriting.** General-purpose OCR engines handle printed text and often struggle badly with cursive handwriting. Plan for search over printed text only, and locate handwritten content by page number and date. **Letting filenames drift.** One notebook named `notebook_final2.pdf` inside a set of `NB-` files will be lost by the next person. Enforce the naming convention mechanically, with a script if needed. **Retiring the paper too early.** Discarding or boxing off the originals before verification and a successful ELN import leaves no way to fix gaps. Keep the paper through at least one full review cycle. **Ignoring fragile and attached media.** Forcing a glued-in gel photo through a sheet-fed scanner destroys both the photo and the scanner. Handle attachments deliberately. **One limitation to plan around.** Individual records vary widely in condition, media type, and institutional retention requirements, so a lab planning a large migration should confirm handling rules with its own records office, quality unit, or data-management office before discarding or altering any original. ## Frequently Asked Questions ### What resolution should I use when scanning paper lab notebooks for ELN migration? Use 300 DPI as your standard for text pages, and 400 to 600 DPI for pages with fine detail such as small plots, pencil notes, or photographic figures. Lower resolutions cannot be fixed after the fact, because the detail was never captured. ### Should I scan lab notebooks in color or black and white? Scan in full color by default. Color preserves blue and green inks, pencil marks, stains, and photographic figures that black-and-white scanning destroys, and storage costs are rarely a good reason to lose that information. ### How do I handle glued-in items like photos, gels, and printouts? Scan them as separate attachment files, label each with the notebook identifier and the page it came from, and record them in the manifest. Do not try to peel items off the page, and never force thick or delicate media through a sheet-fed scanner. ### Do I need OCR, or are page images enough? Images alone preserve the record, but OCR makes the archive searchable, which is the main practical benefit of migration. Run OCR to get a searchable text layer under the page images, and expect strong results on printed text but limited results on handwriting. ### What file format should scanned notebooks be saved in? Capture pages as lossless TIFF or PNG, then deliver a PDF, ideally PDF/A, the archiving profile of the PDF standard. Avoid JPEG for text pages, because its compression discards detail permanently. ### Are scanned notebooks acceptable as data integrity records? A high-fidelity scan that preserves the original page, including signatures and dates, supports the ALCOA data integrity principles the FDA applies to records. The scan's value depends on capture quality, verification, and controlled storage, which is why the checklist and checksum steps above matter. ### How do I keep notebook pages in the right order after scanning? Use zero-padded filenames such as `p0001` through `p0142`, scan in strict page order, and keep a manifest that maps each file to its physical page number. Zero padding prevents page 10 from sorting before page 2. ### Can I scan notebooks with my phone instead of a flatbed scanner? Yes, as a fallback. Use a document-capture app with perspective correction, shoot straight on under even light, and note in the manifest that the page was phone-captured. For the permanent archive, a flatbed or sheet-fed scanner gives more consistent quality.