Standard extracted fields

ChromaParse extracts the following fields from each chromatography PDF (availability depends on report layout):

Sample info

• Sample name / sample ID / batch ID / operator / injection time / injection volume / detection wavelength

Method header

• Method name / column / mobile phase / flow rate / column temp / detector / run time

Peak table

• / retention time (RT) / peak name / area / height / area % / height % / theoretical plates / tailing factor / resolution / concentration / units

System suitability

• RSD / tailing / plates / resolution / acceptance criteria comparison result

Quantitation summary

• Assay content / known + unknown impurities / total impurity / sum / acceptance result

Precision

We maintain a test set for every certified template:

• Numeric fields (RT, area, height, etc.): 100% exact — extracted from the PDF text layer, no OCR involved
• Text fields (sample names, method names, etc.): >99.5% — rare character-segmentation ambiguities are flagged for review
• Scanned PDFs (OCR pipeline): numeric >99.8%, text >98%

Fields below precision targets are explicitly tagged review_needed and never pass silently — that’s a hard requirement for GMP compliance.

Multi-page and continuation tables

Chromatography reports often span multiple pages for a single sample (continuation peak tables, SST on its own page, etc.). ChromaParse automatically:

1. Detects “continuation” markers (different vendors use different conventions; all are catalogued)
2. Merges cross-page peaks into one logical table
3. Preserves original page numbers as audit-trail metadata

What we don’t extract

• Chromatogram image itself — we currently extract data only; image-based chromatogram extraction may come later but isn’t high-ROI
• Handwritten annotations — flagged as handwritten_annotation on scanned PDFs but content not parsed
• Stamps, signatures, QR codes — location detected, content not parsed

Custom field mappings

Per-customer configuration available:

• Field renaming: Area % → Relative peak area (matches LIMS schema)
• Unit conversion: area mAU·s → AU·min
• Precision truncation: RT 4 decimals → 3
• Default values: if a field is missing from the PDF, emit N/A or null

See Enterprise / Delivery & operations.