AutoProSeq - automatic classification of sequence types for prostate mpMRI

XGBoost-based automatic classification of sequence types for prostate mpMRI into T2W, DWI, ADC, DCE and others (any other sequence type) using metadata alone. Works as a CLI, allows direct application to DICOM datasets or CSV/tsv/Excel/parquet datasets, features integration for Orthanc and DICOM-web as RESTful APIs. This was applied to all data on ProCAncer-I to derive initial results for sequence types. Refines predictions using rule-based heuristics for T2W, DWI and ADC.

Topics

Collections

Details

License:
MIT
Maturity:
Emerging
Cost:
Free of charge (with restrictions)
Tool Type:
api, command-line tool
Operating Systems:
Linux
Programming Languages:
Python
Added:
12/2/2025
Last Updated:
12/3/2025

Operations

Data Inputs & Outputs

Image annotation

Publications

Documentation

Other
https://github.com/CCIG-Champalimaud/metadata-classification
Repository containing the code, README and additional information on how to run the tool. If used as an API (when integrated with Orthanc or DICOM-web), the documentation is available upon launching the API at <api-url>/docs.

Links

Repository
https://github.com/CCIG-Champalimaud/metadata-classification
(Repository containing code, README and additional usage information for this tool.)