---
id: highdicom
version: "0.28.1"
license: MIT
license_treatment: permissive
maintenance: active
---

# highdicom — High-level DICOM abstractions.

License: permissive · Maintenance: active · Downloads: 83.6K/mo

## What it is and what it does

Highdicom is a pure Python library that sits on top of pydicom to simplify working with DICOM medical images in computational contexts. It handles three main workflows: reading DICOM files of various modalities (radiology, pathology, etc.) and preparing their frames for analysis, including spatial arrangement and pixel transforms that pydicom does not handle; writing derived DICOM objects such as annotations, segmentations, parametric maps, structured reports, and secondary captures; and reading and filtering those derived DICOM files to extract stored information.

The package targets machine learning, computer vision, and similar computational analyses where standard DICOM tooling leaves gaps. It depends on pydicom for core DICOM parsing, numpy and pillow for image handling, pyjpegls for JPEG-LS codec support, typing-extensions for type hints, and packaging for version management. It requires Python 3.10 or later and installs as a pure wheel with no compiled dependencies.

Use it for:

- Prepare medical imaging datasets for machine learning by reading DICOM frames and applying standardized pixel transforms for computational analysis.
- Store machine learning model outputs (segmentations, heatmaps, predictions) back into DICOM format for clinical workflow integration and archival.
- Create structured DICOM reports containing numerical results and annotations from computational analyses or human review.
- Convert single-frame DICOM images to multi-frame format or apply presentation state transformations for standardized display.
- Extract and filter information from existing derived DICOM objects (segmentations, parametric maps) for downstream processing.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Highdicom provides a high-level Python API for reading, writing, and manipulating DICOM medical image files, with particular focus on machine learning and computational analysis workflows.

Yes. Highdicom is actively maintained, permissively licensed (MIT), has low install friction, and fills a genuine gap in DICOM tooling for computational workflows. It is well-supported by the imaging research community (NCI Imaging Data Commons, QIICR, Radiomics) and has no known security vulnerabilities. Install it if you work with DICOM files in machine learning or computational analysis contexts.

## Install

pip install highdicom
uv add highdicom
poetry add highdicom

## Installing highdicom

Before you install: Low friction installation with pure Python distribution. Active maintenance with recent releases; last commit 2026-08-07. Supports current Python versions (3.10–3.14).

License in practice: MIT license permits commercial and private use with minimal restrictions, making it suitable for research and production deployments.

Quickstart:

pip install highdicom

import highdicom
from pydicom import dcmread

# Read a DICOM file and access its image data
dataset = dcmread('image.dcm')
dicom_image = highdicom.from_dicom(dataset)
frames = dicom_image.pixel_array

Requires Python 3.10 or later. JPEG-LS compressed images require pyjpegls, which may need system libraries depending on your platform.

Verify before relying:

- Whether pyjpegls is required for all use cases or only for JPEG-LS compressed images
- Performance characteristics when processing large multi-frame DICOM datasets
- Completeness of support for all DICOM modalities and derived object types

## Package facts

- License: MIT (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 83.6K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

DICOM image processing, medical image annotation, DICOM segmentation, radiology pathology images, DICOM derived objects, structured report DICOM, parametric map images, medical-imaging, dicom-processing, ml-data-prep

[View on SkillFed](https://skillfed.io/packages/highdicom) · [View on PyPI](https://pypi.org/project/highdicom/)