pysuv 0.1.0

Python library for calculating Standardized Uptake Value (SUV) from PET DICOM images

A Python library for calculating Standardized Uptake Value (SUV) from PET DICOM images.

Table of Contents

• Why pySUV?
• Features
• Installation
• Quick Start
• Documentation
• API Reference
• Validation
• References
• Disclaimer
• License

Why pySUV?

SUV calculation is essential for PET image analysis, whether for traditional clinical reading (PERCIST criteria, Deauville score) or machine learning pipelines (radiomic features, preprocessing). Yet no simple, validated Python library exists.

Alternative	Strength	Limitation
3D Slicer	Complete, multi-vendor	Heavy dependency (~500 MB), GUI-oriented, hard to script
Z-Rad	Excellent, validated (arXiv:2410.13348)	Broad scope (full radiomics + GUI), many dependencies
ACRCode	Vendor-neutral, lightweight	Incomplete implementation (edge cases)
GitHub snippets	Simple	Unvalidated, BQML-only

pySUV provides a focused, validated solution: one function call, multi-vendor support, < 1% error vs QIBA reference. Beyond accurate calculations, it validates DICOM metadata against clinical guidelines—catching data entry errors and protocol deviations before they affect your results.

Features

• Validated: Cross-validated against QIBA DRO (< 1% error)
• Multi-vendor: Siemens, GE, Philips (BQML and CNTS units)
• Four normalizations: SUVbw, SUVlbm, SUVbsa, SUVibw
• Simple API: Compute SUV arrays or get conversion factors
• Clinical validation: Optional module for protocol compliance checks

Installation

pip install pysuv

Quick Start

import pysuv

# Calculate SUV from a PET DICOM file
suv = pysuv.compute_suv("pet.dcm")
print(f"SUVmax: {suv.max():.2f}")

# Use different normalizations
suv_lbm = pysuv.compute_suv("pet.dcm", normalization="lbm")
suv_bsa = pysuv.compute_suv("pet.dcm", normalization="bsa")

Documentation

• SUV Conversion Guide: Supported vendors, formats, normalizations, and edge cases
• Validation Guide: Clinical protocol validation for FDG PET

API Reference

compute_suv(source, normalization='bw')

Compute SUV array from a PET DICOM image.

• source: Path to DICOM file or pydicom Dataset
• normalization: 'bw' (default), 'lbm', 'bsa', or 'ibw'

Returns a numpy array of SUV values (same shape as input image).

get_suv_factor(source, normalization='bw')

Get the SUV conversion factor without computing the full array. Useful for applying to ROIs or storing for reproducibility.

import pydicom
import pysuv

factor = pysuv.get_suv_factor("pet.dcm")

# Apply to pixel data (works for all vendors)
ds = pydicom.dcmread("pet.dcm")
scaled = ds.pixel_array * factor.rescale_slope + factor.rescale_intercept
suv = scaled * factor.suv_factor

get_suv_parameters(source)

Inspect the parameters used for SUV calculation.

info = pysuv.get_suv_parameters("pet.dcm")
print(info)
# SUV Parameters:
#   Manufacturer: SIEMENS
#   Injected dose: 370.0 MBq
#   Patient weight: 70.0 kg
#   Half-life: 109.77 min
#   Elapsed time: 60.0 min

Normalizations

Type	Description	Required fields
bw	Body weight (default)	weight
lbm	Lean body mass	weight, height, sex
bsa	Body surface area	weight, height
ibw	Ideal body weight	weight, height, sex

Validation

pySUV is validated against:

• QIBA DRO: Digital Reference Object with known SUV values
• Reference implementations: Z-Rad, ACRCode, 3D Slicer

References

• Z-Rad: Pfaehler et al. "Technical Note: Vendor-Specific Approach for Standardized Uptake Value Calculation", arXiv:2410.13348 (2024)
• QIBA SUV Profile: RSNA QIBA Wiki
• QIBA DRO: University of Washington PET/CT DRO

Disclaimer

For research use only. This software is not intended for clinical decision-making or diagnostic purposes. It has not been cleared or approved by any regulatory agency.

The authors provide this software "as is" without warranty of any kind. In no event shall the authors be liable for any claim, damages, or other liability arising from the use of this software.

Users are responsible for validating results against their institution's reference standards before use in any research or clinical context.

License

MIT