Data

Tools

Indexes

Applications

Experiments

Open Source Science

f18-psma-pet-ct-ai

Detection of local prostate cancer recurrence from PET/CT scans using deep learning

https://github.com/biomeds/f18-psma-pet-ct-ai

Science Score: 57.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ✓
    CITATION.cff file
    Found CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ✓
    .zenodo.json file
    Found .zenodo.json file
  • ✓
    DOI references
    Found 5 DOI reference(s) in README
  • ○
    Academic publication links
  • ○
    Academic email domains
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (7.6%) to scientific vocabulary
Last synced: 6 months ago · JSON representation ·

Repository

Detection of local prostate cancer recurrence from PET/CT scans using deep learning

Basic Info
  • Host: GitHub
  • Owner: BioMeDS
  • License: mit
  • Language: Jupyter Notebook
  • Default Branch: main
  • Size: 556 KB
Statistics
  • Stars: 1
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Created about 1 year ago · Last pushed 9 months ago
Metadata Files
Readme License Citation

README.md

Detection of local prostate cancer recurrence from PET/CT scans using deep learning

This repository contains the accompanying code for the article:

Korb M, Efetürk H, Jedamzik T, Hartrampf PE, Kosmala A, Serfling SE, Dirk R, Michalski K, Buck AK, Werner RA, et al. Detection of Local Prostate Cancer Recurrence from PET/CT Scans Using Deep Learning. Cancers. 2025; 17(9):1575. https://doi.org/10.3390/cancers17091575

[!IMPORTANT] Large outputs (e.g. model weights) are deposited on Zenodo (10.5281/zenodo.15174580). Training data can not be shared publically. To understand the structure of the data, these files are included as symbolic links that point outside of the repository.

Pre-processing

Nifti conversion

Exported and pseudonomized PET and CT dicom images for each examination were converted to nifti format using dcm2niix (Chris Rorden's dcm2niiX version v1.0.20220720 (JP2:OpenJPEG) (JP-LS:CharLS) GCC5.5.0 x86-64 (64-bit Linux)). Those niftis are saved in data/nifti (train and validation set) and data/nifti_ts2024 (test set).

Prostate segmentation

The prostate and urinary bladder were segmented with TotalSegmentator (version 2.1.0) in all ct images.

bash for i in data/nifti*/*_ct.nii.gz do TotalSegmentator -i $i -o analysis/totalsegmentator2/$(basename $i _ct.nii.gz) -rs prostate urinary_bladder done

Cropping around prostate (or urinary bladder)

Cropping a 20x20x20 cm³ cube around the centroid of the prostate (if detected) or urinary bladder (otherwise) with the script code/preprocessing/crop_by_prostate_or_ub.py. The cropped files are saved in data/cropped_nifti.

Conversion from BQML to SUV

Determine factors for SUV conversion

The converted nifti PET files have values in the BQML unit. In order to convert them to SUV, individual conversion factors have to be determined. This was done by applying code/preprocessing/bqml_to_suv.py to the dicom files (containing the relevant header information) to create the factors in data/suv_factors.tsv.

Convert cropped PET niftis

bash while read pid suv do fslmaths data/cropped_nifti/${pid}_pet.nii.gz -mul $suv data/cropped_nifti_suv/${pid}_pet.nii.gz done <<(tail -n +2 data/suv_factors.tsv)

Get ranges for scaling

bash python code/preprocessing/nii_range.py data/cropped_nifti/*_ct.nii.gz >analysis/cropped_ct_range.tsv python code/preprocessing/nii_range.py data/cropped_nifti_suv/*_pet.nii.gz >analysis/cropped_pet_suv_range.tsv

Owner

  • Name: BioMedical Data Science
  • Login: BioMeDS
  • Kind: organization

Group at the Center for Computational and Theoretical Biology, University of Würzbrug

Citation (CITATION.cff) 

cff-version: 1.2.0
message: If you use this data/code, please cite both the software repository and the paper (see preferred-citation).
authors:
  - family-names: Korb
    given-names: "Marko"
  - family-names: Efetürk
    given-names: "Hülya"
  - family-names: Jedamzik
    given-names: "Tim"
  - family-names: Hartrampf
    given-names: "Philipp E"
  - family-names: Kosmala
    given-names: "Aleksander"
  - family-names: Serfling
    given-names: "Sebastian E"
  - family-names: Dirk
    given-names: "Robin"
  - family-names: Michalski
    given-names: "Kerstin"
  - family-names: Buck
    given-names: "Andreas K"
  - family-names: Werner
    given-names: "Rudolf A"
  - family-names: Schlötelburg
    given-names: "Wiebke"
    orcid: 0000-0001-9685-0947
    affiliation: "Department of Nuclear Medicine, University Hospital Würzburg, Germany"
    email: schloetelb_w@ukw.de
  - family-names: Ankenbrand
    given-names: "Markus J"
    orcid: 0000-0002-6620-807X
    affiliation: "Center for Computational and Theoretical Biology, University of Würzburg, Germany"
    email: markus.ankenbrand@uni-wuerzburg.de
title: "Detection of local prostate cancer recurrence from PET/CT scans using deep learning"
version: 0.3.0
doi: 10.5281/zenodo.15188969
date-released: 2025-04-10
repository-code: https://github.com/BioMeDS/f18-psma-pet-ct-ai
keywords:
  - PET/CT
  - Cancer
  - Prostate
  - Deep Learning
license: MIT
url: https://github.com/BioMeDS/f18-psma-pet-ct-ai/new/main
preferred-citation:
  type: article
  scope: Cite this paper if you use this code in your research
  authors:
    - family-names: Korb
      given-names: "Marko"
    - family-names: Efetürk
      given-names: "Hülya"
    - family-names: Jedamzik
      given-names: "Tim"
    - family-names: Hartrampf
      given-names: "Philipp E"
    - family-names: Kosmala
      given-names: "Aleksander"
    - family-names: Serfling
      given-names: "Sebastian E"
    - family-names: Dirk
      given-names: "Robin"
    - family-names: Michalski
      given-names: "Kerstin"
    - family-names: Buck
      given-names: "Andreas K"
    - family-names: Werner
      given-names: "Rudolf A"
    - family-names: Schlötelburg
      given-names: "Wiebke"
    - family-names: Ankenbrand
      given-names: "Markus J"
  title: "Detection of Local Prostate Cancer Recurrence from PET/CT Scans Using Deep Learning"
  year: 2025
  journal: Cancers
  volume: 17
  issue: 9
  doi: 10.3390/cancers17091575
  url: https://doi.org/10.3390/cancers17091575

GitHub Events

Total
  • Release event: 3
  • Watch event: 1
  • Push event: 20
  • Public event: 1
  • Create event: 3
Last Year
  • Release event: 3
  • Watch event: 1
  • Push event: 20
  • Public event: 1
  • Create event: 3