Prostate Fused-MRI-Pathology
Fused Radiology-Pathology Prostate Dataset (PROSTATE-FUSED-MRI-PATHOLOGY)
Pre-prostatectomy 3 T multiparametric prostate MRI of 28 men with prostate cancer, paired with H&E pseudo-whole-mount histology of the excised glands, pathologist cancer annotations and registration files that map tumor extent onto the MRI.
Overview
The Fused Radiology-Pathology Prostate Dataset pairs in vivo prostate MRI with digitized histology of the same glands after radical prostatectomy. It was contributed by Anant Madabhushi (Case Western Reserve University) and Michael D. Feldman (Hospital of the University of Pennsylvania) and published on The Cancer Imaging Archive in 2016. Version 2 (April 2023) added a table that links MR slices to pathology slides; the images did not change. The collection supports work on MRI-histology registration, tumor mapping on MRI and imaging markers of tumor grade.
Composition
All 28 subjects are men with prostate cancer, each with one MR study (324 DICOM series, 32,508 images). Every subject has T2-weighted and multiphase dynamic contrast-enhanced series, 25 have separate pre-contrast T1-weighted series and 12 have diffusion-weighted series with ADC maps. Annotated pseudo-whole-mount H&E pathology images exist for 16 subjects (114 images), and registration files in MATLAB and MHA format for 15. Spreadsheets list the slice correspondences between MRI and pathology.
Acquisition
All series come from a 3 T Siemens Verio, according to the DICOM headers. The paper recommended by the authors describes an endorectal coil and a T1-weighted VIBE sequence for DCE in its 23-patient analysis. Each excised gland was cut into sections, each section into quarters, and each quarter slide was scanned at 20x on an Aperio scanner. The four quarter images were then stitched into one pseudo-whole-mount section.
Annotations
An expert pathologist outlined cancer on the stitched sections. Software matched each section to a T2-weighted MR slice, and a pathologist and a radiologist checked the matches. Deformable registration then carried the cancer outlines from histology onto the MRI, giving a map of tumor extent on the MR images.
Known limitations
- Pathology images and registration files cover only about half of the subjects (16 and 15 of 28).
- The protocol varies between subjects: diffusion imaging is missing for more than half, and series naming differs.
- Age and Gleason grade are not given for the 28 subjects on the collection page.
- All MR data come from a single scanner model.
Cohort
Aggregate numbers from the sources below. Bars are relative to the 28 subjects.
Sex
- Male 28 100%
Contrast combinations
How many subjects have exactly each set of contrasts.
| T1w | T1w_ce | T2w | dwi | dce | Subjects with exactly this set |
|---|---|---|---|---|---|
13 | |||||
7 | |||||
5 | |||||
3 |
Modality
subjects
- MRI 28 100%
- Pathology slides 16 57%
Contrast / sequence
subjects, values can overlap
- T2-weighted 28 100%
- DCE perfusion 28 100%
- T1-weighted 25 89%
- Diffusion-weighted 12 43%
- T1-weighted, contrast enhanced 7 25%
Condition
subjects, values can overlap
- Prostate cancer 28 100%
Scanner vendor
subjects
- Siemens Healthineers 28 100%
Field strength
subjects
- 3 T 28 100%
License and access
Our reading of the license, not legal advice. Before you use the data, read the original license and confirm that your use is allowed. We take no responsibility for how you use a dataset. Full disclaimer
Download without an account
MRI as DICOM via the TCIA Data Retriever, whole-slide pathology images and annotations via IBM Aspera, and the registration MATLAB files and correspondence tables as direct downloads.
Creative Commons Attribution 3.0 Unported
Older version of CC BY. Use, share and adapt the data for any purpose, including commercial use, as long as you credit the creators.
Every item on the TCIA collection page is listed under CC BY 3.0. TCIA also asks users to follow its Data Usage Policy and to cite the dataset DOI.
What you can do
- Yes
- Yes
- Yes
- Yes
What you can share
- Yes
- Yes
- Yes
What you must do
- Yes
- Share alike No
- No
- No
- Manuscript review No
- Release code No
- Return results No
- Delete after use No
Limits
- No
- Location limits No
Citation
Madabhushi A, Feldman M. (2016). Fused Radiology-Pathology Prostate Dataset (Prostate Fused-MRI-Pathology). The Cancer Imaging Archive. https://doi.org/10.7937/K9/TCIA.2016.TLPMR1AM
All numbers
Every number on this page, as stored in stats.csv, with its source.
| Measure | Breakdown | Value | Source |
|---|---|---|---|
| Subjects | total | 28 | tcia-prostate-fused-mri-pathology Data Access table |
| Studies | total MR studies | 28 | tcia-prostate-fused-mri-pathology Data Access table |
| Scans | total DICOM MR series, including localizers and derived ADC, FA and subtraction series | 324 | tcia-prostate-fused-mri-pathology Data Access table |
| Images | total DICOM MR images (slices) | 32,508 | tcia-prostate-fused-mri-pathology Data Access table |
| Subjects | modality=MR | 28 | tcia-prostate-fused-mri-pathology Data Access table |
| Subjects | modality=SM subjects with annotated whole-slide pathology images | 16 | tcia-prostate-fused-mri-pathology Data Access table |
| Images | modality=SM annotated whole-slide pathology images (TIFF and XML) | 114 | tcia-prostate-fused-mri-pathology Data Access table |
| Subjects | condition=prostate_cancer | 28 | tcia-prostate-fused-mri-pathology Summary |
| Subjects | field_strength=3 | 28 | tcia-prostate-fused-mri-pathology Summary |
| Subjects | vendor=siemens all 324 series from a Siemens Verio | 28 | tcia-nbia-series Manufacturer, ManufacturerModelName |
| Subjects | sex=male | 28 | tcia-nbia-series getPatientStudy, PatientSex |
| Subjects | contrast=T2w series descriptions starting with T2 (TSE and HASTE) | 28 | tcia-nbia-series PatientID, SeriesDescription |
| Subjects | contrast=dce multiphase fl3d_vibe series with at least 200 images; for one subject only the subtracted dynamic series is present | 28 | tcia-nbia-series PatientID, SeriesDescription, ImageCount |
| Subjects | contrast=T1w non-dynamic pre-contrast T1 series (T1 axial, T1 VIBE in/opposed phase, T1 FLASH) | 25 | tcia-nbia-series PatientID, SeriesDescription |
| Subjects | contrast=dwi ep2d_diff or MDDW DTI series | 12 | tcia-nbia-series PatientID, SeriesDescription |
| Subjects | contrast=T1w_ce non-dynamic post-contrast T1 series | 7 | tcia-nbia-series PatientID, SeriesDescription |
| Subjects | contrast_set=dce+T1w+T2w | 13 | tcia-nbia-series PatientID, SeriesDescription, ImageCount |
| Subjects | contrast_set=dce+dwi+T1w+T1w_ce+T2w | 7 | tcia-nbia-series PatientID, SeriesDescription, ImageCount |
| Subjects | contrast_set=dce+dwi+T1w+T2w | 5 | tcia-nbia-series PatientID, SeriesDescription, ImageCount |
| Subjects | contrast_set=dce+T2w | 3 | tcia-nbia-series PatientID, SeriesDescription, ImageCount |
Sources
The keys used in the table above.
- tcia-prostate-fused-mri-pathology TCIA PROSTATE-FUSED-MRI-PATHOLOGY collection page (version 2, updated 2023/04/10) website
- tcia-nbia-series TCIA public NBIA API, series and patient metadata of the collection (CC BY 3.0, no account needed), getSeries and getPatientStudy computed
- singanamalli2016 Singanamalli et al. 2016, Journal of Magnetic Resonance Imaging (PMC4691230) paper