# Lab Notebook Image Annotation: How to Do It Well By the end of this guide you will be able to annotate images in a [lab notebook](/knowledge/molecular-biology/lab-notebook) so that another scientist, months or years later, can tell exactly what the image shows, under what conditions it was acquired, and why you captured it. You will set up a file naming scheme, preserve raw data, add scale bars and labels with FIJI (ImageJ), write captions that carry real information, and embed everything in a notebook entry that stays searchable. Here is what to have on hand before you start: - The raw image files from your instrument (microscope, gel imager, plate reader camera, or phone) - A copy of FIJI, the free ImageJ distribution, installed from the official download page - A command-line metadata reader such as ExifTool if you work with camera or microscope files - Your lab notebook, whether paper with printed inserts or an electronic lab notebook (ELN) ## Why Annotate Images in a Lab Notebook Image annotation means attaching identifying information to an image so it can be interpreted without the person who made it standing nearby. The information lives in two places: marks on the image itself (scale bars, arrows, region outlines, false color) and text in the notebook entry (caption, acquisition conditions, sample identity, verdict). The need grows quickly with data volume. Developers of the Electron Microscopy Electronic Notebook (EMEN), a web-based system for transmission electron microscopy (TEM) data, described the problem directly. As high-resolution biological TEM grew in popularity, the volume of data and the number of parallel projects rose dramatically, and effective [data management](/blog/guides/data-management-basics-principles-processes-and-best-practices) became essential to process large data sets and extract maximum information from available data [1]. Their solution organized the database so multiple collaborators at different locations could access and interpret the data without assistance [1]. That standard, an image a distant collaborator can interpret unaided, is a good target for any lab notebook, paper or electronic. Annotation is also about trust. In regulated and collaborative settings, a figure that cannot be traced to a raw file, a date, and an instrument setting is a figure nobody can defend. Good annotation is cheap at the bench and expensive to reconstruct afterward. ## Paper Notebooks, Electronic Notebooks, and Where Images Fit An electronic lab notebook is software that replaces or supplements the bound paper notebook. It typically stores entries with timestamps, supports attachments, and offers search. Choosing between paper and electronic formats shapes how you annotate images, so it helps to compare them on the features that matter for image-heavy work. | Feature | Paper notebook | Generic electronic notebook | Structured, database-backed notebook | |--|--|--|--| | Image embedding | Taped prints or pasted thumbnails | Drag-and-drop attachments | Images stored as records with fields | | Search across images | Page flipping only | Free-text search of captions | Structured search on sample, instrument, condition | | Multi-user access | One physical copy | Depends on platform | Designed for distributed teams | | Extensibility | None | Limited to platform fields | Schema can grow with new data types | | Data export | Manual scanning | Usually supported | Often direct, for analysis pipelines | The third column describes systems like EMEN, which was built as a portable, object-oriented, web-based tool for TEM data archival and project management [1]. Two design choices in EMEN are instructive for anyone annotating images today. First, the database was logically organized and annotated so that collaborators in different places could interpret the data without help [1]. Second, it was designed to be used much as a paper lab notebook would be, while keeping a structure suitable for data mining and direct interaction with data-processing software [1]. The object-oriented database structure was dynamic, so information not defined in the original specification could be added later [1]. That flexibility matters for images, because the metadata you need in year one (microscope, magnification) often differs from what you need in year three (detector, acquisition software version, image analysis parameters). Practical takeaway: whether you use paper or an ELN, structure your image annotations the way EMEN structured its database. Separate the raw image, the annotations, and the interpretation, and keep all three linked to one entry. ## What to Record Before You Annotate Annotation fails when the underlying metadata was never written down. Before touching the image, confirm that your notebook entry contains the minimum set below. 1. **Date and operator.** The date the image was acquired, not the date you wrote it up, if they differ. 2. **Sample identity.** Sample ID, batch, passage number, plate position, or animal number, following your lab's convention. 3. **Instrument and settings.** Instrument name, objective or lens, magnification, exposure, channel, voltage, or camera setting. For microscope work, the numerical aperture of the objective and the calibration (micrometers per pixel) are essential. 4. **Treatment and condition.** What was done to the sample, with concentrations and times where applicable. 5. **Why the image exists.** One sentence: what question this image was captured to answer. If you use a phone camera for documentation shots, note that phone files carry their own metadata. Capture it before it is stripped by messaging apps, which commonly remove EXIF (exchangeable image file format) data when images are shared. ## Lab Notebook Image Annotation Step by Step This walkthrough uses a worked example: a gel image and a set of cell microscopy frames that need to go into a notebook entry. Commands are shown for a Unix-style terminal and for FIJI. Expected outputs follow each command. ### Step 1. Preserve the raw file and record its fingerprint Never annotate the only copy of a raw image. Copy the raw file into a project folder and compute a checksum, a short hash that uniquely identifies the file contents. ```bash cp IMG_4471.CR2 ./raw/20261103_projectX_gel_A1_raw.CR2 md5sum raw/20261103_projectX_gel_A1_raw.CR2 ``` Expected output: ```text 9f3a1c4e7b2d8a6f0e5c1b9d3a7f2e6c raw/20261103_projectX_gel_A1_raw.CR2 ``` Write the hash, or at least the last eight characters, in your notebook entry. If anyone later questions whether the annotated figure matches the raw data, the hash settles it. ### Step 2. Name files so they sort and read correctly Use a fixed pattern: date, project, sample, condition, instrument, serial number. The example above follows it. Dates in ISO format (YYYYMMDD) sort chronologically in every file browser. Avoid spaces, which complicate command-line work, and use underscores or hyphens. ### Step 3. Read the embedded metadata Camera and microscope files often store acquisition settings inside the file. Read them before you transcribe anything into the notebook. ```bash exiftool -Model -DateTimeOriginal -ExposureTime -ISO raw/20261103_projectX_gel_A1_raw.CR2 ``` Expected output: ```text Model : Canon EOS R7 Date/Time Original : 2026:11:03 14:22:07 Exposure Time : 1/60 ISO : 800 ``` Transcribe the values that matter into your entry. This step prevents the most common [transcription error](/knowledge/molecular-biology/transcription-error), copying the wrong exposure from a previous session. ### Step 4. Set the physical scale in FIJI ![Worked example dividing 100 micrometers by 412 pixels to give 0.243 micrometers per pixel for the FIJI Set Scale command.](/article-images/illustrations/lab-notebook-image-annotation-pixel-calibration.webp) Open your image in FIJI. If you have an image of a stage micrometer taken with the same objective and camera, set the spatial calibration so pixels convert to real units. From the menus, choose Analyze, then Set Scale, and enter the known distance. Programmatically, an ImageJ macro line looks like this: ```text run("Set Scale...", "distance=412 known=100 unit=micron global") ``` This tells FIJI that 412 pixels span 100 micrometers in this optical setup, so the calibration is 0.243 micrometers per pixel. The `global` flag applies the calibration to subsequent images from the same session. To verify, use Analyze, then Measure on a feature of known size. The Results window should show a Length close to the [expected value](/blog/data-analysis/expected-value-definition-formula-examples). If it does not, the calibration is wrong and every scale bar you add will be wrong too. ### Step 5. Add the scale bar and annotation marks With calibration set, add a scale bar: ```text run("Scale Bar...", "width=50 x=10 y=10 font=14 color=White background=None location=[Lower Right] bold overlay") ``` Expected result: a white bar labeled "50 µm" appears in the lower right corner. The `overlay` keyword keeps the bar as a non-destructive overlay in the TIFF file, meaning you can remove or adjust it later. Write down which objective and camera the calibration came from, because a scale bar is only valid for images acquired under the same optical conditions. Add arrows, text, and region outlines with FIJI's overlay tools, or in a macro: ```text run("Colors...", "foreground=white background=black selection=red") setFont("SansSerif", 14) drawString("A1 untreated", 20, 30) makeRectangle(520, 210, 120, 90) run("Draw", "overlay") ``` Expected result: the label "A1 untreated" appears at the top left, and a red rectangle outlines the region of interest at coordinates x 520, y 210, width 120, height 90 pixels. Record what the rectangle marks in the caption, for example "nucleus selected for intensity measurement." ### Step 6. Save two files, not one ![Diagram of the raw file copied and hashed, an annotated TIFF master made from the copy, and a flat PNG export embedded in the notebook entry.](/article-images/illustrations/lab-notebook-image-annotation-file-roles.webp) Save the annotated image as a TIFF with the overlay intact, under a new name that marks it as annotated: ```text raw/20261103_projectX_gel_A1_annotated.tif ``` Then export a flat copy for embedding, such as a PNG at a reasonable size for the notebook. Keep the raw file untouched. Three files, three roles: raw (untouched truth), annotated (master annotated version), and export (convenience copy for the notebook page or ELN attachment). ### Step 7. Write the caption A caption should answer what, when, where, how, and so what in two or three sentences. Structure it the same way every time so entries stay comparable across a project. ### Step 8. Embed, link, and index Paste or attach the export into your notebook entry next to the caption. Then add the searchable text: sample ID, condition, and image type, as plain words in the entry or as tags. If your ELN supports custom fields or a growing schema, follow the EMEN approach and add fields as your project needs them rather than forcing all images into one fixed form [1]. The goal is that searching the sample ID returns the entry, the entry links to the annotated image, and the annotated image links back to the raw file. ## Annotation Types and When to Use Them | Annotation | Purpose | Typical tool | Watch out for | |--|--|--|--| | Scale bar | Shows physical size | FIJI Scale Bar command | Bar valid only for the calibrated objective and camera | | Arrow or marker | Directs the eye to a feature | FIJI overlay tools, drawing tools | Arrows without a matching caption note | | Region of interest outline | Marks the area measured or discussed | FIJI `makeRectangle` or freehand ROI | Outline with no statement of what was measured | | Text label | Names wells, lanes, conditions | FIJI `drawString`, annotation layers | Labels that collide with data or sit outside a crop | | False color | Makes dim signal visible | FIJI lookup tables (LUTs) | Changing LUT without noting the original channel | | Crop or inset | Focuses on the decisive region | FIJI crop, duplicate and paste | Cropping that hides a failed or informative area | Two rules govern all of them. Annotate a copy, never the raw file. And make every mark on the image correspond to a sentence in the caption, so nothing on the figure is unexplained. ## A Lab Notebook Template for Image Entries A consistent template turns annotation from a chore into a checklist. Copy this block into each entry that includes an image, and fill every line. ```text ## Image entry - Entry date: YYYY-MM-DD - Operator: name - Project: project code - Sample ID: - Instrument and settings: instrument, objective/lens, exposure, channel - Calibration: micrometers per pixel, from (micrometer slide date/file) - Raw file: path and md5 (first 8 characters) - Annotated file: path - Export file: path (attached to this entry) - Annotations present: [scale bar / arrow / ROI / label / LUT] - Caption: one to three sentences (what, condition, question, observation) - Conclusion or next step: ``` The caption line deserves the most care. Compare two captions for the same image. "Gel image" tells a reader nothing. "Agarose gel, sample A1 through A6, PCR products of the target locus, 1 percent agarose, 100 volt run for 40 minutes, ladder in lane 1. Bands at the expected size in A2 and A5 only." That second caption lets a reader reuse the image years later. The EMEN project emphasized exactly this property, a database organized and annotated so that other people can interpret the data without assistance [1]. ## Lab Notebook Image Annotation Tips **Calibrate once per optical setup, verify every session.** Store calibration files next to the images they apply to. A scale bar computed from the wrong objective is a persistent, persuasive error. **Keep annotation layers separate where the format allows.** FIJI overlays and ELN annotation layers preserve the underlying pixels. Flattened marks are permanent. **Use false color with discipline.** A lookup table (LUT) is a mapping from pixel values to display colors. Fire, Ice, and Grayscale are common choices. Note the LUT name in the caption and, when quantification matters, keep the original grayscale data available. **Write the negative result into the caption.** If the image shows what you expected to see and did not, say so. An annotated null observation is data. **Annotate at acquisition time or within a day.** Memory of instrument settings degrades fast. The EMEN developers built their system around capturing information during the workflow rather than reconstructing it later [1], and the same principle applies at the bench. **Batch the repetitive parts.** If you annotate fifty frames from one acquisition, write a short FIJI macro that applies the calibration and scale bar to a folder of images, then inspect the batch. Consistency across a series is itself information. **Keep one canonical identifier per image.** The file name from Step 2 should appear in the caption, the ELN attachment name, and the raw file path. One string, used everywhere, ties the chain together. ## Common Mistakes and Limitations **Annotating the raw file.** Drawing on the only copy destroys evidence. Always annotate a duplicate and keep the hash of the original. **A scale bar with no recorded calibration.** A bar drawn after guessing the magnification looks authoritative and can be wrong by a factor of ten. Record the micrometers-per-pixel value and its source in the entry. **Annotations that drift from the data.** If a crop, rotation, or contrast stretch is applied after annotation, marks may no longer point where they should. Finish geometric changes first, then annotate. **Captions that describe procedure but not observation.** "Image of cells after treatment" is a procedure note. Add what is visible and what it means for the question at hand. **Missing negative space.** Cropping to the beautiful cell and discarding the empty field hides useful information about transfection efficiency or contamination. Note in the caption when a crop excludes relevant context. **No link back to the raw file.** An annotated PNG embedded in an ELN with no pointer to the raw data cannot be audited. Always include the path or identifier of the raw file and its checksum. **Over-trusting embedded metadata.** Messaging apps, some converters, and screenshot tools strip or rewrite EXIF data. Read metadata early and transcribe the values that matter into the notebook itself. **Inconsistent naming across a team.** A per-sample scheme that each person invents defeats search. Agree on one pattern in writing and enforce it in the template. **A limitation to keep in mind.** Every lab has its own conventions, instrument documentation, and, in regulated settings, formal procedures for data records. The workflow here is a framework. Adapt the field list and file naming to your lab's standard operating procedures, and when a record has regulatory weight, follow your institution's data management requirements rather than this general guide. ## How to Check Your Work Run through this short audit before you consider an entry done. 1. Can a reader identify the sample, date, and condition from the entry alone, without asking you? 2. Does the annotated file name appear in the entry, and does the entry link to the raw file? 3. Is every mark on the image (bar, arrow, rectangle, label) explained in the caption? 4. Is the raw file hash recorded, and does recomputing it give the same value? 5. Does a search for the sample ID find this entry in your ELN or, for paper, in your index? If all five answers are yes, the entry meets the standard EMEN was designed around, remote collaborators interpreting data without assistance [1]. ## Next Steps Once the basic workflow is habit, three upgrades are worth pursuing. First, write a FIJI macro for your most common acquisition type so calibration, scale bar, and labels apply consistently across batches. Second, if you use an ELN, explore whether it supports structured fields for image metadata, following the model of a dynamic, extensible schema that grows with the project rather than a fixed form [1]. Third, agree with your lab on the template above, adapt it, and store the agreed version where new members will find it. Annotation quality is a team property, not an individual one. ## Frequently Asked Questions ### What is lab notebook image annotation? It is the practice of adding identifying marks and text to images recorded in a lab notebook, including scale bars, labels, region outlines, captions, and links to raw data, so the images remain interpretable without the original operator. ### Do I need an electronic lab notebook to annotate images well? No. A paper notebook with printed images, structured captions, and file references can be excellent. An ELN adds search, timestamps, and easy embedding, which matters most for large data volumes and distributed teams. ### What metadata should accompany every image? Date, operator, sample identity, instrument and settings, treatment condition, and one sentence on why the image was captured. For microscopy, include the calibration in micrometers per pixel. ### Should I annotate the raw image file or a copy? Always annotate a copy. Keep the raw file untouched, compute and record its checksum, and save the annotated version under a separate name. ### How do I add a scale bar to a microscopy image? In FIJI, set the spatial calibration with a stage micrometer image taken under the same objective, then use the Scale Bar command. The bar is only valid for images acquired with the same optical setup. ### What file format should I save annotated images in? Save the annotated master as TIFF with annotation kept as a non-destructive overlay where possible, then export a PNG or JPEG copy for embedding in the notebook. Never overwrite the raw file. ### How can I make annotated images searchable later? Use a consistent file naming pattern, repeat the sample ID and condition in the caption, and embed the image in an ELN entry that contains those same terms as searchable text or tags. ### How detailed should captions be? Detailed enough that a reader can interpret the image without asking you. Cover what the image shows, the acquisition conditions, and the observation, usually in two or three sentences. ## Sources 1. [Object oriented database and electronic notebook for transmission electron microscopy.](https://pubmed.ncbi.nlm.nih.gov/14750990/)