High-Throughput Segmentation Analysis for PLM Images

Transform qualitative Polarized Light Microscopy into fast, repeatable measurements of crystalline domains and morphology, without using up scarce early material.

Solid-state characterization

Conventional PLM Has Real Limits

  • Qualitative & SubjectiveStandard workflows rely almost entirely on visual, expert interpretation rather than quantitative metrics.
  • Slow, Manual MeasurementMeasuring particles one by one in image software is slow, so only a small sample of each image set ever gets sized.
  • Unused Data PotentialValuable image sets captured during pre-formulation are rarely analyzed quantitatively.
  • Iterative Material LossWithout reliable, non-destructive quantitative data, teams risk inefficient characterization cycles and depleted early lots.

Getting More From Limited Early Material

In early discovery pharmaceutics, new APIs and intermediates such as spray-dried dispersions (SDDs) and granulations come in small lots, and most of that material is already spoken for by release and in vivo testing. That leaves room for only a few non‑destructive tests.

Polarized Light Microscopy (PLM) is one of them. It needs very little sample and uses none of it up, which makes it a natural first look at crystalline domains and particle morphology. The images just rarely get measured. segHT PLM measures them at essentially no material cost, before the lot goes on to its intended use.

See what your PLM images hold

With segHT PLM

The same material goes to the same uses, and every portion now carries crystalline domain and morphology data.

Without segHT PLM

The same material goes to the same uses, and the PLM images go unmeasured.

New APIsSDDsIntermediatesIn vivotestingMaterialcharacterizationFormulationdevelopmentsegHT PLMNo material used

Limited early material

Polarized light microscopy

Introducing segHT PLM

segHT PLM is a software module for the automated detection and analysis of crystalline and amorphous domains in PLM images, for APIs and intermediates such as SDDs and granulations. Every image becomes quantitative crystalline domain and morphology data.

Core capabilityHow it works
Dual Deep Learning ModelsOne model separates crystalline from amorphous domains. The other outlines every particle.
Automated RunRuns each analysis step in sequence, automatically, as soon as images are imported.
Crystalline Domains and MorphologyReports crystalline area % and crystalline count %, plus particle and domain size distributions, from standard PLM images.
Integrated Image LongevityStores images and quantitative records together in a structured database.

Impact on your workflow

Every run delivers crystalline domain and morphology metrics, so qualitative visual estimates give way to repeatable, quantitative data. Because images and records are stored together, your characterization data stays available for future studies.

Request a Live Demo

A raw grayscale polarized light image of drug particles, before any analysis The same image with crystalline domains segmented in red by Model 1 The same image with every particle segmented in blue by Model 2 Both results overlaid: particles in blue, crystalline domains in red
Raw image · before analysis

Particle size analysis

One-Click Analysis Workflow

segHT PLM turns image processing into a hands‑off procedure designed for busy discovery labs.

  1. 01

    Upload

    Import raw PLM images directly into segHT PLM.

  2. 02

    Automate

    Just press play. The deep learning models run each analysis step automatically.

  3. 03

    Quantify

    In about a minute, get quantitative particle sizing, segmentation overlays and a structured analytical report.

  4. 04

    Compare

    Line up runs across lots in the stored database to spot shifts in crystalline area or particle size before they cost you material.

31.4%

Crystalline area

Standard deviation ± 5.7

74.4%

Crystalline count

Standard deviation ± 8.9

MeasureD10D50D90
Particles16.6 ± 2.426.6 ± 1.333.4 ± 0.3
Crystalline domains13.1 ± 0.722.3 ± 3.527.5 ± 2.4
Size percentiles in µm, volume weighted

Particle size distribution

010203040505101520253035Equivalent spherical diameter (µm)Volume (%)01020304050102030Equivalent spherical diameter (µm)Volume (%)

Crystalline domain size distribution

051015202530051015202530Equivalent spherical diameter (µm)Volume (%)0510152025300102030Equivalent spherical diameter (µm)Volume (%)

The line shows volume % at each size. The shaded band is the spread the segHT PLM report draws around it.

Representative data from an example run on dipyridamole

Show the data as a table
Particles
Diameter (µm)Volume (%)Spread (%)
5.30.20 to 0.4
6.80.40 to 0.7
8.30.80.3 to 1.2
9.80.20 to 0.4
11.31.91 to 3.6
12.81.20.6 to 1.8
14.44.62 to 7.5
15.96.21.5 to 11
17.400 to 0
18.95.43 to 7.8
20.45.83 to 9
21.918.89 to 29
23.46.90 to 15
24.92.50 to 5
26.40.20 to 0.4
27.910.81.5 to 20.5
29.500 to 0
31.000 to 0
32.58.83.5 to 15
34.029.410 to 49
Crystalline domains
Diameter (µm)Volume (%)Spread (%)
0.83.20.5 to 6
2.300 to 0
3.81.30.8 to 1.8
5.31.80.5 to 4
6.80.70.3 to 1.5
8.32.51.5 to 3
9.91.20.6 to 2.5
11.48.42 to 17
12.94.92.5 to 6
14.422.314 to 31
15.92.21 to 3.5
17.400 to 0
18.910.26.5 to 14
20.41.70.8 to 3.5
21.900 to 0
23.412.42.3 to 22.5
25.013.82.8 to 25.5
26.500 to 0
28.000 to 0
29.517.86 to 30

Pharmaceutical R&D

Built for Solid-State & Pre-Formulation Teams

segHT PLM is designed for the demands of pharmaceutical R&D, for teams across big pharma, mid-sized biopharma and specialty chemical applications.

  • Pre-Formulation Scientists & DirectorsAccelerate solid-state characterization and decisions before in vivo testing.
  • Solid State Analytical ChemistsStandardize quantitative particle sizing and reduce variation between visual reviewers.
  • Particle Engineers & Process ChemistsGet rapid, non-destructive insight into particle size distributions.
  • Materials ChemistsBuild a reliable historical database of early discovery lots.

Talk to a solid-state imaging specialist

Dissolution modeling

From PLM Image to Release Prediction

The particle size distribution segHT PLM measures goes straight into dissoLab, digiM’s dissolution modeling software. Add material properties such as solubility and density, and dissoLab predicts how each API will release, so two samples that look alike under the microscope can be compared on predicted release while material is still scarce.

Measured in segHT PLM

Particle size distribution

020406080100010203040506070Equivalent spherical diameter (µm)Cumulative volume (%)0204060801000204060Equivalent spherical diameter (µm)Cumulative volume (%)

API 1API 2

Predicted in dissoLab

Simulated drug release

02040608010000.511.522.5Time (minutes)Drug released (%)020406080100012Time (minutes)Drug released (%)

API 1API 2

NameDrug productBench typeDrug density (g/mL)Solubility (mg/mL)Bulk diffusivity (cm²/s)
API 1ParticleD1.5120.450.0000075
API 2ParticleD1.5120.450.0000075

Example: two APIs measured in segHT PLM and run in dissoLab with the same material inputs, so the gap in predicted release comes from particle size alone.

Explore dissoLab

Pharmaceutical imaging

The digiM Advantage

digiM brings deep specialization in pharmaceutical imaging, with years of domain expertise in materials characterization and neural network development. segHT PLM applies that expertise to crystalline and amorphous material alike, from new APIs to intermediates such as SDDs and granulations, and turns each PLM image into quantitative crystalline domain and morphology data.

Ready to Quantify Your PLM Workflows?

Schedule a live demonstration with digiM’s solid-state imaging specialists and see what segHT PLM can do with your pre-formulation images.

Request a Live Demo