README.md

<p align="center">
  <img src="https://github.com/MrGiovanni/R-Super/raw/main/documents/r_super_pdac-8.gif"
       alt="R-Super PDAC demo animation"
       width="400">
</p>

# Merlin Plus

This repository provides Merlin Plus, per-voxel annotations for organs and other anatomical structures in the 25,494 CT scans of the Merlin dataset (Stanford). The per-voxel annotations were created by AI models trained on more than 14,000 CT scans at Johns Hopkins University. 
Merlin Plus is part of a collaboration between the Merlin Project at Stanford and the R-Super Project at Johns Hopkins University. R-Super is a novel AI training methodology that uses radiology reports and organ annotations to significantly improve tumor-segmentation AI. Merlin Plus makes R-Super easily reproducible for the medical AI community.

>[!NOTE]
>See the [Report Supervision (R-Super) GitHub](https://github.com/MrGiovanni/R-Super) to discover how you can use Merlin Plus to improve **tumor segmentation**!

- Merlin Plus includes per-voxel annotations for organs, blood vessels, organ parts (liver and pancreas sub-segments), and ducts:

<details>
<summary style="margin-left: 25px;">Class List</summary>
<div style="margin-left: 25px;">
  
```
adrenal gland left
adrenal gland right
aorta
bladder
cbd stent
celiac artery
celiac trunk
colon
common bile duct
duodenum
esophagus
femur left
femur right
gall bladder
hepatic vessels
intestine
kidney left
kidney right
liver
liver segment 1
liver segment 2
liver segment 3
liver segment 4
liver segment 5
liver segment 6
liver segment 7
liver segment 8
lung left
lung right
pancreas body
pancreas head
pancreas tail
pancreas
pancreatic duct
portal vein and splenic vein
postcava
prostate
rectum
renal vein left
renal vein right
spleen
stomach
superior mesenteric artery
superior mesenteric vein
```

</details>

- Merlin Plus presents annotations for the CT scans in the original Merlin dataset. To download these CT scans, please visit this website: https://stanfordaimi.azurewebsites.net/datasets/60b9c7ff-877b-48ce-96c3-0194c8205c40

### Merlin Dataset
Merlin Abdominal CT Dataset is an abdominal CT dataset consisting of 25,494 scans from 18,317 patients. Each scan is paired with its corresponding radiology report. The dataset includes abdominal and pelvis CT exams conducted between 2012 and 2018 at the Stanford Hospital Emergency Department, selected using CPT codes (72192, 72193, 72194, 74150, 74160, 74170, 74176, 74177, and 74178) through the STARR tool. For each exam, the DICOM series with the largest slice count was converted into NIfTI format, compressing the scans and removing patient-identifiable metadata.


# Papers

<b>Learning Segmentation from Radiology Reports</b> <br/>
[Pedro R. A. S. Bassi](https://scholar.google.com/citations?user=NftgL6gAAAAJ&hl=en), [Wenxuan Li](https://scholar.google.com/citations?hl=en&user=tpNZM2YAAAAJ), [Jieneng Chen](https://scholar.google.com/citations?user=yLYj88sAAAAJ&hl=zh-CN), Zheren Zhu, Tianyu Lin, [Sergio Decherchi](https://scholar.google.com/citations?user=T09qQ1IAAAAJ&hl=it), [Andrea Cavalli](https://scholar.google.com/citations?user=4xTOvaMAAAAJ&hl=en), [Kang Wang](https://radiology.ucsf.edu/people/kang-wang), [Yang Yang](https://scholar.google.com/citations?hl=en&user=6XsJUBIAAAAJ), [Alan Yuille](https://www.cs.jhu.edu/~ayuille/), [Zongwei Zhou](https://www.zongweiz.com/)* <br/>
*Johns Hopkins University* <br/>
MICCAI 2025 <br/>
<b>Best Paper Award Runner-up (top 2 in 1,027 papers)</b>  <br/>
<a href='https://link.springer.com/chapter/10.1007/978-3-032-04971-1_29'><img src='https://img.shields.io/badge/Paper-PDF-purple'></a><a href='https://link.springer.com/chapter/10.1007/978-3-032-04971-1_29'><img src='https://img.shields.io/badge/Springer-Link-orange'></a>

<p align="center">
  <img src="https://github.com/MrGiovanni/R-Super/raw/main/documents/miccai_2025_best_paper_award.png"
       alt="Prize"
       width="400">
</p>

<b>Merlin: A Vision Language Foundation Model for 3D Computed Tomography</b> <br/>
Louis Blankemeier, Joseph Paul Cohen, Ashwin Kumar, Dave Van Veen, Syed Jamal Safdar Gardezi, Magdalini Paschali, Zhihong Chen, Jean-Benoit Delbrouck, Eduardo Reis, Cesar Truyts, Christian Bluethgen, Malte Engmann Kjeldskov Jensen, Sophie Ostmeier, Maya Varma, Jeya Maria Jose Valanarasu, Zhongnan Fang, Zepeng Huo, Zaid Nabulsi, Diego Ardila, Wei-Hung Weng, Edson Amaro Junior, Neera Ahuja, Jason Fries, Nigam H. Shah, Andrew Johnston, Robert D. Boutin, Andrew Wentland, Curtis P. Langlotz, Jason Hom, Sergios Gatidis, Akshay S. Chaudhari  <br/>
*Stanford University* <br/>
<a href='https://arxiv.org/pdf/2406.06512v1'><img src='https://img.shields.io/badge/Paper-PDF-purple'></a>

# Download Merlin Plus

You can manually download and unzip the Merlin Plus annotations, or you can save the script below as download.sh and execute it with 'bash download.sh'.

<details>
<summary style="margin-left: 25px;">Download Code</summary>
<div style="margin-left: 25px;">
  
```bash
#!/bin/bash
set -e

cd data/

echo "Downloading README.md..."
wget -O README.md "https://huggingface.co/datasets/AbdomenAtlas/MerlinPlus/resolve/main/README.md?download=true"

mkdir -p MerlinMasks

# Download and extract the 1,000-case shards (00000001 → 00025000)
for i in {1..25}; do
    start=$(printf "%08d" $(( (i - 1) * 1000 + 1 )))
    end=$(printf "%08d" $(( i * 1000 )))
    file="MerlinMasks_BDMAP_${start}_${end}.tar.gz"
    url="https://huggingface.co/datasets/AbdomenAtlas/MerlinPlus/resolve/main/${file}?download=true"

    echo "[${i}/25] Downloading $file..."
    wget --show-progress -O "$file" "$url"

    echo "Extracting $file..."
    tar -xzf "$file" -C MerlinMasks
    rm "$file"
done

# Download and extract the final partial shard (00025001 → 00025456)
file="MerlinMasks_BDMAP_00025001_00025456.tar.gz"
url="https://huggingface.co/datasets/AbdomenAtlas/MerlinPlus/resolve/main/${file}?download=true"
echo "Downloading final shard..."
wget --show-progress -O "$file" "$url"
echo "Extracting $file..."
tar -xzf "$file" -C MerlinMasks
rm "$file"

cd ..
```

</details>

# Citations
If you use this data, please cite the papers below (R-Super and Merlin Projects):

```
@inproceedings{bassi2025learning,
  title={Learning segmentation from radiology reports},
  author={Bassi, Pedro RAS and Li, Wenxuan and Chen, Jieneng and Zhu, Zheren and Lin, Tianyu and Decherchi, Sergio and Cavalli, Andrea and Wang, Kang and Yang, Yang and Yuille, Alan L and others},
  booktitle={International Conference on Medical Image Computing and Computer-Assisted Intervention},
  pages={305--315},
  year={2025},
  organization={Springer}
}

@article{blankemeier2024merlin,
  title={Merlin: A vision language foundation model for 3d computed tomography},
  author={Blankemeier, Louis and Cohen, Joseph Paul and Kumar, Ashwin and Van Veen, Dave and Gardezi, Syed Jamal Safdar and Paschali, Magdalini and Chen, Zhihong and Delbrouck, Jean-Benoit and Reis, Eduardo and Truyts, Cesar and others},
  journal={Research Square},
  pages={rs--3},
  year={2024}
}

```