Skip to main content

A Python package for data handling and evaluation in deep learning-based medical image analysis.

Project description

pymia

Documentation status

pymia is an open-source Python (py) package for deep learning-based medical image analysis (mia). The package addresses two main parts of deep learning pipelines: data handling and evaluation. The package itself is independent of the deep learning framework used but can easily be integrated into TensorFlow and PyTorch pipelines. Therefore, pymia is highly flexible, allows for fast prototyping, and reduces the burden of implementing data handling and evaluation.

Main Features

The main features of pymia are data handling (pymia.data package) and evaluation (pymia.evaluation package). The data package is used to extract data (images, labels, demography, etc.) from a dataset in the desired format (2-D, 3-D; full- or patch-wise) for feeding to a neural network. The output of the neural network is then assembled back to the original format before extraction, if necessary. The evaluation package provides both evaluation routines as well as metrics to assess predictions against references. Evaluation can be used both for stand-alone result calculation and reporting, and for monitoring of the training progress. Further, pymia provides some basic image filtering and manipulation functionality (pymia.filtering package). We recommend following our examples.

The following figure depicts pymia in the deep learning environment. The data package allows to create a dataset from raw data. Extraction of the data from this dataset is possible in nearly every desired format (2-D, 3-D; full- or patch-wise) for feeding to a neural network. The prediction of the neural network can, if necessary, be assembled back to the format before extraction. The evaluation package allows to evaluate predictions against references using a vast amount of metrics. It can be used stand-alone (solid) or for performance monitoring during training (dashed).

The pymia package in the deep learning environment

Getting Started

If you are new to pymia, here are a few guides to get you up to speed right away:

Citation

If you use pymia for your research, please acknowledge it accordingly by citing our paper:

Jungo, A., Scheidegger, O., Reyes, M., & Balsiger, F. (2021). pymia: A Python package for data handling and evaluation in deep learning-based medical image analysis. Computer Methods and Programs in Biomedicine, 198, 105796.

BibTeX entry:

@article{Jungo2021a,
author = {Jungo, Alain and Scheidegger, Olivier and Reyes, Mauricio and Balsiger, Fabian},
doi = {10.1016/j.cmpb.2020.105796},
issn = {01692607},
journal = {Computer Methods and Programs in Biomedicine},
pages = {105796},
title = {{pymia: A Python package for data handling and evaluation in deep learning-based medical image analysis}},
volume = {198},
year = {2021},
}

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

pymia-0.3.4.tar.gz (59.7 kB view details)

Uploaded Source

Built Distribution

pymia-0.3.4-py3-none-any.whl (73.5 kB view details)

Uploaded Python 3

File details

Details for the file pymia-0.3.4.tar.gz.

File metadata

  • Download URL: pymia-0.3.4.tar.gz
  • Upload date:
  • Size: 59.7 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/5.1.1 CPython/3.9.16

File hashes

Hashes for pymia-0.3.4.tar.gz
Algorithm Hash digest
SHA256 d795c8fcf49011a054d18376103e2660bab1be8a92029c923d55e8281de32b10
MD5 2760cd5200632ce8c9ee64410c7931c6
BLAKE2b-256 59b0237c3cb1b9991ef2ee0c89380e19006373469ad006c65ffae3db090382b3

See more details on using hashes here.

File details

Details for the file pymia-0.3.4-py3-none-any.whl.

File metadata

  • Download URL: pymia-0.3.4-py3-none-any.whl
  • Upload date:
  • Size: 73.5 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/5.1.1 CPython/3.9.16

File hashes

Hashes for pymia-0.3.4-py3-none-any.whl
Algorithm Hash digest
SHA256 106aa03ff75c24bf419fb7fd5f634cb70bed034025ec1cc48570cf309a9aeeb5
MD5 46332270f009c2a19bd5c03fc3e2db1a
BLAKE2b-256 43eb41526650378c07498572018c92cf384b00e0136aff4cfe4ea4d28d1693bc

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page