Table 5 Overview of the recommendations for each step of an image-based particle measurement. The time estimates per sample are exemplarily given for around 10,000 microplastics that are used in effect studies. In general, the higher the degree of automation, the faster the method

1: Sample preparation

Few hours^a

Subsampling

Take ten subsamples of equal weight from the homogenized bulk

Minimum number of particles

300 for environmental microplastics and 10,000 for microplastics that are used in effect studies

Contamination

Clean instruments and sample holders. Use procedural blanks and avoid plastics. Work with a laminar flow box and cover samples

De-agglomeration

Sprinkle a small amount of microplastics through a sieve with a mesh size slightly larger than the largest particle

Measuring points

With effect studies, measure after production, in vivo, and post mortem. For environmental microplastics, measure the filter

2: Image acquisition

1 h – 1 day^a

Calibration

Calibrate the light microscope by measuring a given distance ten times at different locations on a certified graticule

Magnification

The magnification should be set so that the smallest particle is represented by at least five to ten pixels in a montage

Illumination

Always use Köhler illumination. Choose the type of illumination based on the microplastics in the sample

Depth of focus

Use Z-stacks to maximize the depth of focus

Image storage

16-bit grey scale images stored as uncompressed.tiff files with all relevant metadata

Number of field of views

Multiply the minimum number of particles by 5 to 50 particles per field of view

3. Digital image processing

Few seconds (computer vision) – 30 min [128]

Smoothing

Cancel noise by filters. Use them very carefully. Correct uneven illumination a digital image of the empty background

Contrast

Use auto contrast functions in the whole digital image

Segmentation

Choose an adequate computer algorithm for global automatic thresholding by qualitative and quantitative comparisons

Separation of touching particles

Avoid the usage of computer algorithms. Better de-agglomerate the subsamples

4. Measurement

Few seconds (computer vision) [128] – few minutes^a

Measurement frames

The choice generally depends on the automation of the stage. Neighboring measurement frames can be processed the easiest

Particle size measurement

Choose a size metric based on the research objective. As a compromise, report the maximum Feret’s diameter and open data

Particle shape measurement

Calculate roundness, solidity, and elongation and provide open data

Computer algorithms

Measure the perimeter with the Freeman algorithm. Measuring the projection area is done by counting pixels

5. Quality control and quality assurance

1 – 2 days^a

Outlier detection in subsamples

Apply Grubb’s test/Dixon’s Q to test whether the extreme values of the subsamples are outliers

Background contamination

Use procedural blanks to estimate the number of non-plastic particles by spectroscopy

Excluding of small microplastics

Remove all microplastics < 3 µm from the data analysis

Validation of particle shape

Use a Sphericity-Elongation diagram to detect microplastics with a combination outside the theoretical range

6. Data reporting

Few hours^a

Test report

Write a test report according to ISO 13322–1:2014

Particle size distribution

Draw a normalized histogram with corrected counts. Choose the type of visualization (linear/logarithmic) based on the width of the distribution. Alternatively, report a cumulative distribution

Distributions of shape descriptors

Apply the same recommendations as for particle size distributions. Use a size interval of 0.1

Summary statistics

Report the median and the (geometric) standard deviation