ImageJ Western Blot Analysis: The 15-Step Process (and a Faster Way)

ImageJ (and its distribution Fiji) is the gold standard for biological image analysis. It's free, open source, cited in tens of thousands of papers, and maintained by the NIH. It's also, for western blot densitometry specifically, a bit of a pain.

The built-in gel analysis workflow requires 15+ manual steps, produces no normalized values, and leaves you to copy numbers into Excel for the actual quantification. If you've ever closed an ImageJ window and lost your ROIs, you know the feeling.

Here's the full workflow, step by step, so you know exactly what you're dealing with.

The ImageJ Gel Analysis Workflow

Step 1: Open Your Image

File > Open. If it's a 16-bit TIFF, you'll see a very dark image. You can adjust brightness/contrast for display (Image > Adjust > Brightness/Contrast), but do not click "Apply" — that permanently alters pixel values.

Step 2: Convert to 8-bit (Maybe)

ImageJ's gel analyzer works on 8-bit images. If your image is 16-bit, you need to convert (Image > Type > 8-bit). This reduces your dynamic range from 65,536 levels to 256. Yes, this is a problem. The workaround is to set your display range carefully before converting so the relevant signal falls within the 8-bit range.

Step 3: Draw the First Lane

Select the Rectangle tool. Draw a box around your first lane, making sure it covers all bands you want to quantify. Then go to Analyze > Gels > Select First Lane (or press Ctrl+1).

Step 4: Select Subsequent Lanes

Move the rectangle to lane 2. Analyze > Gels > Select Next Lane (Ctrl+2). Repeat for every lane. ImageJ does not auto-detect lanes — you place each one manually. If your lanes aren't perfectly straight (they rarely are), you just do your best.

Step 5: Plot Lanes

Analyze > Gels > Plot Lanes (Ctrl+3). ImageJ generates a profile plot for each lane — intensity vs. vertical position. Each lane appears as a separate peak profile.

Step 6: Draw Baselines

For each peak in the profile plot, you need to draw a baseline using the Straight Line tool. This is ImageJ's version of background subtraction: the area above the baseline and under the peak curve is your signal. The area below the baseline is background.

This is the most subjective step. Different people draw different baselines, and there's no undo. If you mess up, you start over from Step 5.

Step 7: Measure Peak Areas

Click inside each closed peak with the Wand tool. ImageJ measures the enclosed area and adds it to the Results table. You need to click each peak individually — one for target band, one for loading control, per lane.

Step 8: Copy Results to Excel

The Results table shows raw area values. Copy these into Excel or Google Sheets. ImageJ does not do normalization for you.

Step 9: Calculate Background-Subtracted Intensities

Depending on your baseline drawing, the areas may or may not need further background subtraction. Many protocols suggest measuring a "blank" region and subtracting it from each band.

Step 10: Calculate Ratios

In Excel, divide each target band intensity by its corresponding loading control intensity. This gives you the normalized ratio for each lane.

Step 11: Calculate Fold Change

Divide each normalized ratio by the control sample's ratio to get fold change relative to control.

Step 12: Repeat for Replicates

Open your next blot image and do Steps 1–11 again. Repeat for each biological replicate.

Step 13: Calculate Statistics

In Excel, calculate mean, standard deviation, and SEM across your biological replicates. Run your statistical tests.

Step 14: Make a Figure

Build a bar chart in Excel, GraphPad Prism, or whatever you use. Add error bars, significance brackets, axis labels.

Step 15: Write Methods

Write a methods paragraph describing your quantification approach: software (ImageJ, version), background subtraction method, normalization strategy, number of replicates, statistical tests. Reviewers will ask.

Where This Gets Painful

To be fair to ImageJ — it's an incredible tool. It does everything from cell counting to particle tracking to 3D reconstruction. But for western blot densitometry specifically, the workflow has real friction points:

• The 8-bit conversion problem. ImageJ's gel analyzer requires 8-bit images, which means you lose the dynamic range you carefully preserved at the imager. Workarounds exist, but they add steps and are easy to get wrong.
• Manual baseline drawing is subjective. Two people analyzing the same blot will draw different baselines and get different numbers. There's no record of where the baseline was drawn, making it hard to reproduce.
• No normalization. ImageJ gives you raw areas. The normalization math happens in Excel, which means transcription errors and no audit trail connecting your image to your final numbers.
• No QC checks. ImageJ won't warn you about saturated bands, uneven loading, or other common problems. You have to catch those yourself.
• No reusable ROIs across sessions. Close the window and your ROIs are gone (unless you remember to save them via the ROI Manager, which adds more steps).

A Simpler Approach

VoilaBlot was built specifically for this use case. Instead of 15 steps across two programs, it's three steps in your browser:

1. Upload your blot. TIFF, PNG, or JPEG. 16-bit images stay 16-bit — no forced conversion to 8-bit.
2. Define lanes and bands. Draw or auto-detect lane ROIs, mark your target and loading control bands.
3. Get results. Background-subtracted, normalized densitometry with publication-ready figures, QC checks, and an auto-generated methods paragraph.

All processing happens in your browser — your images never touch a server. And unlike ImageJ's gel analyzer, VoilaBlot works directly on 16-bit data, runs automated QC checks (saturation, loading consistency, background uniformity), and gives you normalized fold-change values without touching a spreadsheet.

ImageJ is still the right tool for many image analysis tasks. But if your specific need is western blot densitometry, a purpose-built tool saves you real time and reduces the chance of quantification errors.

References

• Gassmann M et al. (2009) Electrophoresis 30(11):1845-1855.
• Janes KA. (2015) Sci Signal 8(371):rs2.
• Butler TAJ et al. (2019) Biomed Res Int 2019:5214821.
• Schneider CA, Rasband WS, Eliceiri KW. (2012) NIH Image to ImageJ: 25 years of image analysis. Nat Methods 9(7):671-675.