ASTM E3060-23

This international standard was developed in accordance with internationally recognized principles on standardization established in the Decision on Principles for the
Development of International Standards, Guides and Recommendations issued by the World Trade Organization Technical Barriers to Trade (TBT) Committee.
Designation: E3060 − 23
Standard Guide for
Subvisible Particle Measurement in Biopharmaceutical
Manufacturing Using Dynamic (Flow) Imaging Microscopy
This standard is issued under the fixed designation E3060; the number immediately following the designation indicates the year of
original adoption or, in the case of revision, the year of last revision. A number in parentheses indicates the year of last reapproval. A
superscript epsilon (´) indicates an editorial change since the last revision or reapproval.
1. Scope 1.3 This guide will describe best practices and consider-
ations in applying dynamic imaging to identification of poten-
1.1 Biotherapeutic drugs and vaccines are susceptible to
tial sources and causes of particles during biomanufacturing.
inherent protein aggregate formation which may change over
These results can be used to monitor these particles and where
the product shelf life. Intrinsic particles, including excipients,
possible, to adjust the manufacturing process to avoid their
silicone oil, and other particles from the process, container/
formation. This guide will also address the fundamental
closures, equipment or delivery devices, and extrinsic particles
principles of dynamic imaging analysis including image analy-
which originate from sources outside of the contained process,
sis methods, sample preparation, instrument calibration and
may also be present. Monitoring and identifying the source of
verification and data reporting.
the subvisible particles throughout the product life cycle (from
initial characterization and formulation through finished prod- 1.4 The values stated in SI units are to be regarded as
uct expiry) can optimize product development, process design, standard. No other units of measurement are included in this
improve process control, improve the manufacturing process, standard.
and ensure lot-to-lot consistency.
1.5 This standard does not purport to address all of the
safety concerns, if any, associated with its use. It is the
1.2 Understanding the nature of particles and their source is
responsibility of the user of this standard to establish appro-
a key to the ability to take actions to adjust the manufacturing
priate safety, health, and environmental practices and deter-
process to ensure final product quality. Dynamic imaging
mine the applicability of regulatory limitations prior to use.
microscopy (also known as flow imaging or flow microscopy)
1.6 This international standard was developed in accor-
is a useful technique for particle analysis and characterization
dance with internationally recognized principles on standard-
(proteinaceous and other types) during product development,
ization established in the Decision on Principles for the
in-process and commercial release with a sensitive detection
Development of International Standards, Guides and Recom-
and characterization of subvisible particles at ≥2 μm and
mendations issued by the World Trade Organization Technical
≤100 μm (although smaller and larger particles may also be
Barriers to Trade (TBT) Committee.
reported if data are available). In this technique brightfield
illumination is used to capture images either directly in a
2. Referenced Documents
process stream, or as a continuous sample stream passes
2.1 ASTM Standards:
through a flow cell positioned in the field of view of an imaging
system. An algorithm performs a particle detection routine. E2589 Terminology Relating to Nonsieving Methods of
Powder Characterization
This process is a key step during dynamic imaging. The digital
particle images in the sample are processed by image morphol- 2.2 ISO Standards:
ISO 3951-1 Sampling Procedures for Inspection by Vari-
ogy analysis software that quantifies the particles in size, count,
image intensity, and morphological parameters. Dynamic im- ables
ISO 8871 Elastomeric Parts for Parenterals and for Devices
aging particle analyzers can produce direct determinations of
the particle count per unit volume (that is, particle for Pharmaceutical Use
ISO 9276-6 Representation of Results of Particle Size
concentration), as a function of particle size by dividing the
particle count by the volume of imaged fluid (see Appendix Analysis Part 6: Descriptive and Quantitative Representa-
tion of Particle Shape and Morphology
X1).
1 2
This guide is under the jurisdiction of ASTM Committee E55 on Manufacture For referenced ASTM standards, visit the ASTM website, www.astm.org, or
of Pharmaceutical and Biopharmaceutical Products and is the direct responsibility of contact ASTM Customer Service at service@astm.org. For Annual Book of ASTM
Subcommittee E55.03 on General Pharmaceutical Standards. Standards volume information, refer to the standard’s Document Summary page on
Current edition approved Oct. 1, 2023. Published October 2023. Originally the ASTM website.
approved in 2016. Last previous edition approved in 2016 as E3060 – 16. DOI: Available from American National Standards Institute (ANSI), 25 W. 43rd St.,
10.1520/E3060-23. 4th Floor, New York, NY 10036, http://www.ansi.org.
Copyright © ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959. United States
E3060 − 23
2.3 Other Standards: captured still frame projected images of particles in motion
ANSI/ASQ Z1.9-2003 Sampling Procedures and Tables for (also referred to as flow imaging, flow microscopy, direct
Inspection by Variables for Percent Nonconforming imaging).
ASME BPE-2022 Bioprocessing Equipment
3.2.8 equivalent diameter, n—the diameter of a sphere or
USP <787> Subvisible Particulate Matter in Therapeutic
circle with the same particle volume or area measured by a
Protein Injections
particle sizing instrument.
USP <788> Particulate Matter in Injections
3.2.8.1 Discussion—For dynamic imaging, the equivalent
USP <1663> Assessment of Extractables Associated with
diameter is calculated from the projected area of a measured
Pharmaceutical Packaging/Delivery Systems
particle.
USP <1664> Assessment of Drug Product Leachables Asso-
3.2.8.2 Discussion—Depending on the choice of software
ciated with Pharmaceutical Packaging Delivery Systems
and software settings, the projected area may have any holes in
USP <1787> Measurement of Subvisible Particulate Matter
the image filled or left unfilled.
in Therapeutic Protein Injections
3.2.9 extrinsic particle, n—an unexpected particle intro-
USP <1788> Methods for Determination of Subvisible Par-
5 duced from sources that are foreign or external to the manu-
ticulate Matter
facturing process.
USP <1788.3> Flow Imaging Method for the Determination
3.2.10 Feret diameter, F, n—apparent diameter of an object
of Subvisible Particulate Matter
determined from the distance between two parallel tangents on
opposite sides of a binary object.
3. Terminology
3.2.10.1 Discussion—There are an infinite number of Feret
3.1 Definitions:
diameters; the maximum and the minimum Feret find most use
3.1.1 For definitions of terms used in this standard, refer to
within imaging.
Terminology E2589.
3.2.11 field of view, n—the two dimensional, lateral extent
3.2 Definitions of Terms Specific to This Standard:
of the imaged area.
3.2.1 aspect ratio, n—ratio of particle width to particle
length.
3.2.12 frequency distribution, n—a representation, as a
3.2.1.1 Discussion—The definition of particle width and table, graph, or mathematical function, that gives the frequency
length depends on the image analysis software being used, and
or count of values within a set of specified intervals.
reported aspect ratio may not be interchangeable between
3.2.13 inherent particle, n—a particle made entirely of
different software packages.
components of the formulated drug product or its manufactur-
3.2.2 binary image, n—a transformation of a camera image
ing intermediate, arising from the product itself.
into pixels identified as particles and pixels identified as
3.2.14 intrinsic particle, n—a particle composed of materi-
background.
als that the product or intermediate contacts or is mixed with
3.2.3 brightfield illumination, n—a method of providing during the manufacturing process or during storage in primary
light into a measurement space whereby the illuminated objects
packaging components.
are located between the light source and the viewing receiver.
3.2.15 particle, n—mobile, self-contained, and undissolved
3.2.4 circularity, n—degree to which a particle (or its objects.
projection area) is similar to a circle, mathematically expressed
3.2.15.1 Discussion—In the context of particle counting
as 4πA/P , where A is particle area and P is particle perimeter.
instruments, the term particle may be used to designate any self
contained object that is optically distinguishable from the
3.2.5 cumulative particle size distribution, n—a
background image, including a liquid droplet or gas-phase
representation, as a table, graph, or mathematical function, that
bubble.
gives the total fraction or concentration of particles greater than
or less than a set of specified size values.
3.2.16 particle size distribution (PSD), n—a frequency or
3.2.5.1 Discussion—Cumulative particle size distributions volume distribution of the concentration of particles versus
may be expressed as either mass, volume, area, number, or
particle size.
concentration values.
3.2.16.1 Discussion—Dynamic imaging particle analyzers
of use to the biopharmaceutical industry report the PSD as the
3.2.6 depth of field, n—the distance along the optical axis
concentration of particles per unit volume within specified size
between the nearest and farthest objects that are in acceptably
ranges, where the size is most commonly the equivalent
sharp focus in an image.
diameter but may be another morphological size attribute. See
3.2.7 dynamic imaging, n—particle size and shape analysis
Appendix X1.
using computer image analysis techniques on instantaneously
3.2.17 subvisible particle, n—a particle with a measured
equivalent diameter within the approximate range 1 μm to
100 μm.
Available from American Society of Mechanical Engineers (ASME), ASME
International Headquarters, Two Park Ave., New York, NY 10016-5990, http://
NOTE 1—When it is necessary to specify an exact size range, the range
www.asme.org.
should be defined explicitly rather than by such adjectives as subvisible.
Available from U.S. Pharmacopeial Convention (USP), 12601 Twinbrook
Pkwy., Rockville, MD 20852-1790, http://www.usp.org. 3.2.17.1 Discussion—The 100 μm upper limit is based on
E3060 − 23
the historical definition of subvisible particle as used in the environmental fibers, hair, airborne particles, etc.). Extrinsic
field of drug inspection. Particles of 20 μm or smaller of particle types should be a rare occurrence and eliminated.
sufficient optical contrast are readily visible under bright
illumination, especially when present in numerous quantity. 6. Sources of Particles
3.2.18 threshold, n—the minimum quantitative change in
6.1 Subvisible particles may be generated by a number of
intensity (of either positive or negative sign) from the back-
sources during the manufacturing process. In analyzing the risk
ground pixel value for a pixel to be identified as a possible
of particle generation introduced by various process steps, it is
particle.
useful to understand the sensitivity of the drug product or
substance to a variety of stresses known to promote particle
3.2.19 volume distribution, n—a frequency distribution that
formation.
gives the distribution of particle volume as a function of
particle size.
6.2 Sources of Inherent Particles:
6.2.1 Stresses which may cause inherent particle changes
4. Significance and Use
may include:
4.1 This guide will encompass considerations for manufac-
6.2.1.1 Interaction with interfaces or other particles.
turers regarding sources and potential causes of subvisible
(1) Increased interfacial transport resulting from agitation,
particles in biomanufacturing operations and the use of dy-
stirring, etc.
namic imaging particle analyzers as a suggested common
(2) Interfacial adsorption: both liquid/vapor and liquid/
method to monitor them. The guide will address the following
solid
components of particle analysis using dynamic imaging mi-
(3) Nucleation on other particles
croscopy: fundamental principles, operation, image analysis
(4) Trace metals and other molecules promoting oxidation
methods, sample handling, instrument calibration, and data
and aggregation
reporting.
6.2.1.2 Chemical environment.
(1) Formulation, which may promote or hinder particle
5. Types of Particles
generation
5.1 USP <1787> defines three subcategories of particles
(2) Excipients
related to their source or nature. When combined with appro-
(3) Impurities
priate strategies for characterizing particle types, this catego-
6.2.1.3 Physical environment.
rization scheme provides a framework for assessing the root
(1) Vibration
cause and acceptable concentrations of different types of
(2) Mechanical shock
particles.
(3) Cavitation
5.1.1 Inherent particles are related to the product formula-
(4) Temperature and humidity
tion (for example, chemical and physical properties and con-
(5) Environment—contamination
centration of the Active Pharmaceutical Ingredient (API)
(6) Intense light exposure
proteins, excipients, API solid suspensions, emulsions, adju-
6.2.2 The count and characteristics of the particles formed
vant aluminum salts added to vaccines). Packaging of the
as a result of these stresses will vary in general with the
product and external stresses (including temperature, mechani-
duration of the stress and subsequent storage time and condi-
cal shock or movement, light exposure, and interaction with
tions.
liquid/solid and liquid/air interfaces) can all have substantial
6.3 Sources of Intrinsic Particles:
impact on the concentration and characteristics of protein
6.3.1 Intrinsic particles may be formed when materials in
aggregates. Protein aggregates may change over time, in both
concentration and characteristics, and some levels of protein contact with drug substance or product are stressed, such as the
degradation or related aggregation, or both, may be expected. shedding of particles by pumps used in fill and finish opera-
tions. In other cases, the stresses may be minimal, but the
Inherent particles must be well characterized and monitored
over the product shelf-life. materials are not verified to be sufficiently particle free; an
example would be the shedding of particles from a filter. As
5.1.2 Intrinsic particles include product contact materials
from the manufacturing process or primary packaging compo- with inherent particles, the creation of particles depends both
on the duration of particular stresses and the time of storage.
nents (that is, silicone oil, glass, stainless steel, rubber closure,
polymer tubing, semi-solid silicone lubricant, process related
6.4 Combinations of particular stresses may arise in differ-
fibers, etc.). This category also includes stability-indicating
ent process steps during manufacturing operations, including:
particles found predominantly during development or stability
6.4.1 Formulation,
studies (formulation degradation, container closure-related,
6.4.2 Sterilization,
glass delamination, stopper degradation, etc.). The presence of
6.4.3 Storage: conditions, time of storage, and choice of
intrinsic particle types must be minimized, and if they are
container,
stability-indicating, they should be eliminated whenever pos-
6.4.4 Transport,
sible.
6.4.5 pH adjustments,
5.1.3 Extrinsic particles comprise any particles not sourced
6.4.6 Viral inactivation steps,
from the manufacturing process or product contact materials
including particles of a biological source (that is, external 6.4.7 Ultrafiltration/diafiltration,
E3060 − 23
6.4.8 Container or closure siliconization, which may pro- these data are given in Section 14). Changes in quantities or
mote aggregation of proteins, distribution of subvisible particles should be investigated to
6.4.9 Freeze-thaw, identify root cause. Manufacturers may consider particle con-
6.4.10 Mixing, and tributions from other process steps and studies, including:
6.4.11 Fill/finish.
7.2.1 Scale-up,
7.2.2 Freeze/thaw studies,
6.5 Components in the manufacturing process may contrib-
7.2.3 Development stability studies,
ute particles directly (for example, polymer particles shed by a
7.2.4 Container/closure studies, and
single use system component or other flexible system
7.2.5 Transport/storage studies.
components), or may contribute to increased particle load
indirectly (for example, protein adsorption and subsequent
7.3 Monitoring should also be considered during key manu-
desorption as a particle from a hydrophobic polymer surface).
facturing operations, in particular:
The use of components and filters requires the development of
7.3.1 Sterilization,
compatibility profiles with the product and solutions to assure
7.3.2 Filling,
leachable substances are not a concern as discussed in USP
7.3.3 Container/closure supplies and use,
<1663> and USP <1664>. The therapeutically active drug
7.3.4 Marketed product stability studies,
substance (small or large molecule) would have to be shown
7.3.5 Manufacturing site changes, and
not to bind to the filter system as evidenced by loss of potency
7.3.6 Manufacturing device process changes.
or any indications of API degradation. Process steps may either
increase or decrease particle concentrations, or a combination
7.4 Once the baselines are available, significant deviations
thereof. For example, filtration will remove inherent particles
from the baseline should be noted and particles should be
but may introduce intrinsic particles shed from the filtration
characterized if possible. This characterization may help iden-
media or even promote further growth in inherent particles by
tify root cause. Studies should be undertaken to address the
nucleating interfacial growth of protein aggregates. ISO 8871
sources or adjust the process, or both, to minimize their
is a guide to the compatibility of rubber or elastomeric
formation. In addition, the contribution of particles from the
components for most aspects of stopper performance testing. In
external environment during the manufacturing process, par-
addition, many protein solutions or drug formulation impurities
ticularly during filling operations, should be evaluated, under-
can interact with medical grade silicone used to lubricate the
stood and minimized.
container, closure or plunger, and result in increased protein
7.5 As part of the baseline characterization, it is desirable to
aggregate formation over time in the absence of surfactant.
identify the dominant subpopulations of particle types. One
Also, residual tungsten from the manufacturing of syringe
useful approach is to generate samples with particles of known
barrels with staked cannulas has been implicated in protein
composition and known mechanism of generation. From these
aggregation and particle formation. Pumps are another com-
samples, images representing different categories of particle
mon source of particles and should be inspected frequently for
types can be used to generate parameters for filtering of the
indicators of wear or particulate generation. Piston pumps can
images to categorize them, based on assessment of risk. Care
generate stainless steel particles, peristaltic pumps can cause
should be taken in using the PSD of the sample particles to
spallation or abrasion of the inner tubing wall and generate
generate size-based filter parameters, since the size of particles
polymeric particles, and diaphragm pumps can generate rubber
in test samples may differ significantly from intentionally
diaphragm particles over time. Close attention to pump main-
created particles. Image distinction may be straightforward for
tenance is recommended.
some common uniform particle types such as silicone oil,
whereas distinguishing rare particles such as extrinsic fibers
7. Baseline Monitoring During the Manufacturing
from fibrous protein particles is difficult. Image analysis is a
Process
rapid means of identifying particle types, but care in interpre-
7.1 Biopharmaceutical manufacturers should establish base-
tation of images is necessary, especially for irregularly shaped
lines for particle levels at key steps in the manufacturing
particles. Shape information (for example, aspect ratio,
process to evaluate the effects of component changes, process
circularity, etc.) and image intensity analysis measurements
changes and stability on the product. Baseline data should be in
(for example, average intensity, intensity differences, etc.) may
place to assess and understand how these changes impact the
also be included. Accurate morphological analysis may not be
particle formation during and after the manufacturing process.
possible for particles below 5 μm, depending on the instrument
Particle baselines may be developed during:
used. Because dynamic imaging does not provide direct
7.1.1 Formulation development
chemical information, the specificity of image analysis, espe-
7.1.2 Clinical lot manufacturing
cially (but not only) for small particles, cannot equal the
7.1.3 Routine manufacturing
specificity of microspectroscopy techniques. While use of
7.2 Testing should be conducted at time of release and at the Fourier Transform Infrared spectrometry (FTIR) or Raman
conclusion of shelf life in order to assess the formation and microspectroscopy and Scanning Electron Microscopy-Energy
change in distribution of subvisible particles over time. Particle Dispersive Spectroscopy (SEM-EDS) methods can identify
data should be collected according to size in the following particle types with greater confidence than dynamic image
categories: 2 μm to 5 μm, 5 μm to 10 μm, 10 μm to 25 μm, analysis, these methods have greatly reduced throughput and
25 μm to 50 μm, and 50 μm to 100 μm (Options for reporting have limitations on minimum particle size or composition.
E3060 − 23
SEM-EDS gives basic elemental composition of both organic 8.1.1 Dynamic image analysis is a particle analysis tech-
and inorganic particles as small as 100 nm, but the method is nique using light microscopy to examine microscopic particles
not appropriate for fragile and highly hydrated protein in a moving fluid. Basic instruments are identical to a standard
particles, or similar particles. FTIR and Raman are generally light microscope, with the difference being that in a Dynamic
limited to particle sizes greater than ≈10 μm, with greatly Image Particle Analyzer the sample fluid is imaged
reduced throughput and less chemical specificity near the low dynamically, while in motion, as opposed to the sample being
end of the size range. Positive identification of particles below imaged statically as it is with a stationary sample in light
≈10 μm will depend on the analysis instrument and method microscopy. The primary benefit to dynamic image particle
capabilities and in some cases may not be possible. When analysis is that since the fluid is being imaged dynamically,
investigating deviations from process control, dynamic image larger numbers of particles can be imaged, stored and measured
analysis and investigation by spectroscopic or other chemically in a short period of time. The larger number of particles
specific methods may be warranted. analyzed yields much higher levels of statistical confidence
versus static microscopy. An additional advantage is that
7.6 Dynamic image analysis provides a highly sensitive
background subtraction to correct for image intensity varia-
method for measuring the particle size and counting the
tions other than particles is very effective, enabling detection of
number of particles. Typical limits of detection for dynamic
particles with low optical contrast.
image analysis correspond to very low volume fractions of
particles. For example, 200 particles per milliliter at a diameter
8.2 Basic Hardware Configuration:
-8
of 5 μm is equivalent to a volume fraction of only 10 . As a
8.2.1 Two distinct configuration types for flow imaging
result, for many common particle types, detection of particles
systems are designated here: (1) stand-alone instruments using
is possible at concentrations far below levels that would impact
a sample obtained from a batch and (2) in-line configurations
product quality.
whereby a probe containing the system components is inserted
7.7 From the perspective of risk analysis, particles may be into a process vessel or pipe. While this document will
categorized as: concentrate on the stand-alone type of system, since it is the
7.7.1 Particles that may be present in the final drug product most common (largely because samples are usually drawn
and represent a potentially significant risk to safety or efficacy from the final drug product in its packaged form), the basic
(for example, aggregated protein, foreign material), techniques are very similar for the in-line type of technology
7.7.2 Particles with low intrinsic risk (for example, silicone with the exception that no “sample handling” is involved.
oil in products intended for IV administration), and Dynamic Image Particle Analyzers (see Fig. 1) consist of three
7.7.3 Particles of unknown composition. basic components: fluidics, optics and electronics:
8.2.2 The optics are essentially microscope components,
8. Apparatus
while the electronics consist of the image sensor (camera) and
8.1 Principles of Measurement: supporting electronics required to obtain and process the digital
FIG. 1 Components of a Dynamic Imaging Particle Analyzer
E3060 − 23
images of the particles. The fluidic system consists of sample on the size of particles that may be imaged; particles of sizes
introduction fittings, tubing, a flow cell and a pump. In some close to the flow cell depth or greater may not be properly
systems, samples are introduced into the flow cell by a robotic counted or may cause blockages.
fluid sampler. The pump can be either peristaltic or syringe
8.4 Imaged Volume and Particle Size Limits:
type, and may be controlled by the system computer. The
8.4.1 The volume of space within the interior of the flow
fluidics flow is generally as follows: the pump (typically
cell where particles are imaged is termed the imaged volume.
located downstream of the flow cell), pulls sample fluid from
The imaged volume is defined laterally by the camera view
the sample introduction fittings through the flow cell and out
width and height, or by the flow cell width if this is smaller
into waste (the sample can be recirculated back to introduction
than the camera view width. Similarly, along the optical axis,
if desired, but generally it only passes through once so that
the depth of the imaged volume is the smaller of the flow cell
every particle is only imaged once).
depth or the optical depth of field. The exact dependence on
8.2.3 For in-line systems, hardware design must be compat-
depth of field is influenced by the threshold values selected
ible with any cleaning or sterilization requirements of the
which define a particle from its background and the size and
process.
degree of optical contrast of the particles. The manufacturer
8.3 Flow Cell/Fluidics: must properly calculate the volume of sample being imaged in
8.3.1 In stand-alone instruments, the flow cell is a critical
order to get valid number concentration values.
system component as it must restrict the position of the
8.4.2 There are several instrument variants where the im-
particles to lie in an approximate plane perpendicular to the
aged volume is defined by the depth of field of the imaging
microscope’s optical axis in order to keep them in reasonable
objective, and not by the physical depth of a flow cell. In some
focus (within the depth of field). The flow cell itself is typically
cases, the sample passes through a flow cell or flow passage,
a transparent fluid channel (typically glass) with two flat
but the depth of field is substantially smaller than the flow cell
surfaces through which the sample is actually illuminated and
depth. In this case, the imaged volume in which particles are
imaged. See Fig. 2.
detected, and consequently the reported number concentration,
8.3.2 The flow cell is typically designated by the depth as
may depend on particle type, illumination, and threshold
shown in Fig. 2. Different types and configurations of flow
settings. Some systems use a different arrangement to hydro-
cells may be available. In some of these, the width of the flow
dynamically focus the particles relative to the optics, usually
cell may be greater than the actual camera field of view or referred to as a sheath flow system. A sheath flow system uses
illuminated area, while in other configurations the flow width
a wide tube of glass through which a tunnel of fluid (sheath
may be restricted to stay within or match the camera field of fluid) is created for the sample to pass through in the center by
view or illuminated area. In either case, the manufacturer must
varying the velocity and density of the two fluids so that they
properly calculate the volume of sample being imaged in order do not mix. The net result is that the sample particles pass
to get valid concentration values. Because the optical depth of
through the imaging zone in a very narrow stream (typically in
field (the distance along the optical axis between the nearest single file), and thus remain in sharp focus. These systems
and farthest objects that are in acceptably sharp focus in an
often have very small measured imaged volume, and concen-
image) decreases with increasing magnification, it is important
tration determination may have increased uncertainty. Effects
to match flow cell depth to optical magnification in the system.
of the sheath flow interaction with the product sample are
While most systems have only a single magnification/flow cell
unknown.
size available, some systems do offer different magnifications;
8.4.3 Since dynamic imaging systems use light microscopy,
in these systems it is critical that the manufacturer’s recom-
the minimum particle size which can be counted is set to
mended flow cell size is matched to the magnification. The
approximately 1 μm, or larger for morphological determina-
choice of flow cell depth also sets an approximate upper limit
tions. This is due to diffraction effects and camera pixel size,
which create hard limits on the minimum size. Specialized
instruments can resolve smaller particles using special optics,
but with decreased sample throughput. The maximum size
particle that can be measured is typically restricted by flow cell
depth: particles larger than the flow cell depth may cause
clogging of the flow cell, which is to be avoided. In the case of
biopharmaceuticals, where the particles (particularly protein
aggregates) may be pliable, some particles larger than the flow
cell depth may be seen.
8.5 Illumination:
8.5.1 As particles pass through the flow cell, they are
illuminated most commonly from behind (“back-lit”, although
some systems may use “front lighting”) by a light source,
typically a modulated source. The light source is “strobed”
(typically at a synchronous interval) in order to capture a
blur-free image of the particles as they flow through the cell.
FIG. 2 Flow Cell Most commonly, the transmitted or reflected light is collected
E3060 − 23
and focused without alteration by filters or polarizers, which is particle will obstruct or refract, or both, the light from the
termed brightfield illumination. Some types of particles, such illumination source. As a result, most gray-scale methods
as glass shards, may be challenging to identify with brightfield involve setting a threshold darker than the background; if the
illumination. Alternate illumination types (for example, dark- intensity value for that pixel is equal to or less than the
field) may provide improved sensitivity in principle, but these threshold value (darker) when compared to the background,
alternate illumination schemes may be technically challenging then the pixel is considered to be “particle”. The background
to implement in commercial dynamic imaging particle analyz- value for each pixel in the camera array is the value obtained
ers. when no particles are present, typically derived by averaging
8.5.2 For in-line systems, all of the above descriptions the pixel value over several frames as the fluid is moving
apply, with the exception that the imaging system is now through the flow cell. In general, each manufacturer will either
integral to a pipe or vessel and views the sample therein. The pre-set the threshold method/values or supply recommended
cell, as defined above, still exists for the in-line installation. settings to use. Some systems allow thresholding that selects
The illumination configurations available should be the same pixels that are either darker or lighter than the specified
for the in-line instrument as for the lab instrument in order to threshold. The system must be set up and calibrated per the
allow for the comparison of images and data from both manufacturer’s recommendations. It is, however, important to
analyses. understand what thresholding method and values are used
because they can affect particle measurements dramatically.
8.6 Data Acquisition and Analysis Software:
The lower size limit for particle detection depends both on the
8.6.1 The acquisition and analysis software is an integral
optical resolution of the system and on the pixel size of the
part of commercial dynamic imaging systems. Acquisition
camera; this limit is approximately 1 μm for typical dynamic
software controls the sample fluidics, camera settings, and
imaging particle analyzers.
image capture. Software settings are determined during the
9.2.2 In Step 3, once every pixel in the original image has
Method Development process, discussed in Section 9. Analysis
been classified as “particle” or “not particle” by the threshold-
software performs the steps discussed in Section 10 and often
ing process, contiguous groups of pixels classified as “particle”
generates reports automatically. Details of both types of
are grouped together to form objects. In this case each
software vary substantially between instruments.
stand-alone object is an actual particle in the original image,
9. Image Processing consisting of one or more touching pixels classified in the
binary image as “particle” pixels. This process, referred to as
9.1 In all imaging systems, the stages of image processing in
“particle detection”, can be simple (for example, pixels that are
general can be defined as follows:
touching each other in the binary image), or complex (propri-
9.1.1 Step 1—Acquisition of raw gray-scale (or color)
etary algorithms).
image (full camera field of view).
9.2.3 The thresholding and particle detection processes also
9.1.2 Step 2—Reduction of gray-scale or color image to a
become important when looking at how concentrated a sample
binary image where each pixel can only have one of two
can be when being imaged. As the number concentration
values: particle or not particle.
increases, there is an increased likelihood of particles being too
9.1.3 Step 3—Grouping of contiguous particle pixels to
crowded together to be separated and counted individually. In
form “isolated objects” (individual particles).
a sample with a high particle concentration, two particles in
9.1.4 Step 4—Once each particle is isolated in the binary
close proximity to each other in the camera image may be
image, measurements are made and stored for each particle.
interpreted as a single particle. While there are some algo-
Measurements may be related to particle morphology or image
rithms available for “declumping” or “de-aggregating”
intensity, as well as particle size.
particles, their use is typically limited to particles of known
9.2 Thresholding and Pixel Grouping (“particle detec-
shape, and generally the analyses are too computationally
tion”):
expensive to work in a real time environment. Separation of
9.2.1 Step 2 above, where the original gray-scale image is
particles by the use of image analysis software should not be
converted into a binary image, is one of the most critical steps
used as particles may be genuinely agglomerated. Commercial
in the process, and can sometimes produce different results
dynamic imaging systems in common use for biopharmaceu-
depending on the method used. The most common method
tical applications conform to this recommendation. Refer to
used is gray-scale thresholding: in this process, a threshold is
10.4.1 for further discussion.
chosen. Pixels with an intensity below this threshold are
9.3 Image Analysis:
classified as “particle” while those above the threshold are
classified as “not particle”. Threshold settings may be fixed by 9.3.1 Once the particle detection is complete and binary
the manufacturer or adjustable. It should be noted that for a particle images have been produced, the software will then
mixture of either particle types or sizes that there is no unique make measurements of each particle to be stored and associ-
threshold that will distinguish the edge of a particle in an ated with each corresponding particle. The actual measure-
identical manner. Note also that the measured particle size ments made on each particle may vary by system, but typically
depends on the illumination conditions, and as a result the include equivalent diameter, Feret diameter, aspect ratio
illuminating light level needs careful control. Since most (width/length) and other morphological measurements like
dynamic imaging systems are back-lit, particles will typically circularity, area, etc. For many dynamic imaging particle
have a darker value than the background due to the fact that the analyzers, equivalent diameter is calculated from the projected
E3060 − 23
area of a particle (with any holes filled first) as if that area fibrous protein particles), and particles or droplets at sizes at or
formed a circle, and in equation form is: below the optical resolution of the instrument. Digital filters
are most useful in distinguishing irregular solids (predomi-
Projected Equivalent Particle Diameter 5 = area ⁄ 0.785
~ !
nantly aggregated protein in most cases) from liquid droplets or
gas bubbles at diameters >5 μm.
9.3.2 Some systems provide the option to fill or not fill holes
9.5.2 Particle classification should not be inferred directly
in the binary image, or to use alternative definitions of the
from measurements of test samples with heterogeneous particle
equivalent diameter, such as averaging the Feret diameter over
populations. Instead, samples containing known particle types
multiple angular orientations. The images obtained by dynamic
at the size of interest should be produced and then measured.
imaging capture a 2-D projected area of particles. Shear flow
For example, silicone oil droplets may be created by sonication
inside the flow cell or field of view of an in-line system will
of silicone oil in a detergent solution, or protein particles may
often orient plate-like or fiber-like particles, with the degree of
be created from a variety of physical or chemical stresses.
orientation larger for large particles of symmetric shape. For
From measurements of these known particles, image parameter
oriented particles, equivalent circular diameter is not equiva-
values characteristic of this particle type may be obtained.
lent to the diameter of a sphere of the same volume as the
Combinations of several different image parameters may be
detected particle.
used for image identification in subsequent measurements by
9.3.3 Most systems also make and record intensity-based
application of an appropriate software filter on the measured
measurements such as average intensity and intensity variation
particles. Most filters are value filters where a range of values
for each particle as well. It should be noted that these
for specific measurements are defined and only particles
intensity-based measurements are made from the original
meeting those requirements are counted in that subpopulation.
gray-scale (or color) image as now defined by the binary
Development and verification of digital filters is still under
particle outline. It should be further noted that such measure-
development, and application of digital filters to multiple
ments are 2-D ratios when the particles are actually 3-D.
particle types in routine measurements remains a challenge.
9.3.4 Once the images and measurements are gathered, the
Advanced filters based on machine learning may also be
resultant data are stored, typically in spreadsheet or database
applied to particle classification.
form. Most systems will also save at least some (if not all) of
9.5.3 Although dynamic imaging analysis is a useful tool to
the original particle images so that they can be viewed/
monitor the concentration of particles in a process, with high
reviewed after analysis, providing visual proof of the measure-
ments. Data may be post processed to produce detailed sensitivity and reasonably fast throughput, investigations of the
root cause of particle formation will require additional methods
statistics about the population(s) contained in the sample.
to confirm particle identification. Dynamic imaging alone is
Typical outputs here might be particle size distributions (PSDs)
insufficient to confidently identify particle type.
based on number or converted to another parameter such as
volume or particle shape distributions.
9.6 Reference Library:
9.6.1 Filters for creating subpopulations and particle classi-
9.4 Use of Filters to Create Subpopulations:
9.4.1 Since dynamic imaging systems produce multiple fication may be stored simply as a set of conditions (value or
statistical), but they can also be stored as a reference library of
measurements for each particle, the output data can be used to
identify subpopulations of the original sample. Distinguishing images. This may be useful especially for classification, as the
library can be used as a visual aid for the operator in either
subpopulations has varying success based upon the uniqueness
of image attributes collected for a particular subpopulation. verifying particle types classified by the filters or as a reference
set of images for operator training. A trained operator can often
Digital filters can be defined using any particle measurement or
group of measurements applied to the spreadsheet data. Simple distinguish unusual or heterogeneous particles with more
success than a digital filter, although manual inspection of
size filters are frequently used to quantify subpopulations by
size, such as particles ≥10 μm and particles ≥25 μm. images is tedious. A library may be used simply as a set of
reference images, but it can also be used as the basis from
9.5 Image Characterization:
which values or statistics are generated for filter creation.
9.5.1 Digital filters can be used simply to bin data by
parameter values, as is the case in size filters, and can also be
10. Sampling and Sample Handling
made specific enough to create subpopulations that classify
particles by nominal particle type. The success of particle 10.1 Overview:
classification depends on the degree of parameter differences 10.1.1 Sample handling will depend on the nature of the
(for example, circularity, image intensity) between the images particles to be measured, the particle size range, and the
of different particle types. In practice, the best filters will particle concentration. General procedures and recommenda-
combine both morphological and image intensity criteria, and tions for sample selection, preparation and handling of drug
will be verified for the particular instrument and optical set up products have been covered in the United States Pharmacopeia,
in use. At the present state of the art, filters are often not specifically USP <787>, USP <788>, and the informational
effective at classifying heterogeneous particles (for example, chapters USP <1788> and <1788.3>. Specific guidance on
protein aggregates associated with silicone oil droplets), par- dynamic imaging systems is given in USP <1788.3>, including
ticles or droplets of unusual morphology (for example, dou- method development and qualification, instrument validation,
blets of silicone oil), particles of different types but similar and system suitability tests. Although USP chapters are not
morphology (
...