Mammo-CLIP: A Vision Language Foundation Model to Enhance Data Efficiency and Robustness in Mammography
Shantanu Ghosh1
, Clare B. Poynton2
, Shyam Visweswaran3,
Kayhan Batmanghelich1
1BU ECE, 2 BUMC, 3 Pitt DBMI
⚠️ WARNING: Look for /restricted/projectnb/batmanlab/shawn24/PhD and replace it with your own path. E.g, .src/codebase/breastclip/data/datasets/imagetext.py, change the json path
⚠️ WARNING: There is a plethora of pre-processing settings available for RSNA and VinDr Mammo datasets. We recommend using the pre-processing discussed in the following sections. We are not responsible for any discrepancies in the results due to different pre-processing settings. If you use the VinDr png dataset uploaded in kaggle, it is fully pre-processed. Else you can use the pre-processing scripts provided in the following sections.
import nltk
nltk.download('punkt_tab')After going through the instruction, it is recommended to visit the following queries logged in the issues:
- Issue-2 for any further clarification on pretraining.
- Issue-10 for getting the embeddings.
- Issue-6 for further clarification on the downstream tasks and corresponding datasets.
- Issue-13 for setting up the baselines.
- Issue-9 for problems related to BioClinincalBert from Hugging Face.
If we hear more queries, we may add a separate FAQs in the future.
- Environment Setup
- Data Download
- Pre-processing Images
- Data Preparation for Pretraining
- Data Preparation for Downstream Evaluation Tasks
- Final Dataset Directory Structures
- Mammo-CLIP checkpoints
- Pretraining Mammo-CLIP
- Creating classifiers and detectors for downstream evaluations
- Evaluation
- Tutorial Notebooks
- Additional Scripts
- Mammo-FActOR
- Citation
- License and Copyright
- Contact
- Acknowledgements
- Contributing
Use environment.yml to setup the environment.
git clone [email protected]:batmanlab/Mammo-CLIP.git
cd Mammo-CLIP
conda env create --name Mammo-CLIP -f environment.yml
conda activate Mammo-CLIPMammo-CLIP is implemented with following specification:
- Python version: 3.8.18
- PyTorch version: 2.2.2
- CUDA version: 11.8
Download the original versions VinDr and RSNA from the links for downstream evaluations:
For the PNG images converted from the original Dicom images, as mentioned in the preprocessing steps in the paper, refer to the following links:
To preprocess the dicom images directly, follow the instructions in the next section. If you downloaded the PNG images, skip the preprocessing steps.
python ./src/preprocessing/preprocess_image_to_png_kaggle.py \
--phase="test" \
--base_folder="/restricted/projectnb/batmanlab/shawn24/PhD/RSNA_Breast_Imaging/Dataset/RSNA_Cancer_Detection"python ./src/preprocessing/preprocess_image_to_png_vindr.py \
--phase="test" \
--base_folder="/restricted/projectnb/batmanlab/shawn24/PhD/RSNA_Breast_Imaging/Dataset/External/Vindr/vindr-mammo-a-large-scale-benchmark-dataset-for-computer-aided-detection-and-diagnosis-in-full-field-digital-mammography-1.0.0"-
Our image-text dataset is an in-house dataset from UPMC. You can have your own image+text dataset where images are 2D mammograms and texts are radiology reports. If you have access to such dataset, follow the setup here. Extract the
IMPRESSIONandFINDINGSsections from the report and create a csv. The sample csv: upmc_dicom_consolidated_final_folds_BIRADS_num_1_report.csv -
Note the
FINDINGSandIMPRESSIONcolumns are used to generate the text for the image. TheHISTORY,FINDINGSandIMPRESSIONcolumns contains templated text due to privacy. -
Next run the following command to augment the text with
upmc_dicom_consolidated_final_folds_BIRADS_num_1_report.csvfile:
# input: upmc_dicom_consolidated_final_folds_BIRADS_num_1_report.csv
# output: clip_pretrain_100.csv
python ./src/codebase/augment_text.py \
--dataset-path="/restricted/projectnb/batmanlab/shawn24/PhD/Mammo-CLIP/src/codebase/data_csv" \
--csv-path="upmc_dicom_consolidated_final_folds_BIRADS_num_1_report.csv" \
--dataset="upmc" The augment_text.py script will do the preprocessing for the FINDINGS and IMPRESSION columns by converting them to
lower case and removing punctuations. Look at the _split_report_into_segment_concat function in the script for more
details.
- The csv file of the final image-text dataset should have the following format:
| index | patient_id | laterality | image | view | CC | MLO | text | text_augment |
|---|---|---|---|---|---|---|---|---|
| 0 | patient_id | laterality ('R' or 'L') | List of all image_paths for patient_id-laterality combo | List of views for patient_id-laterality combo (only 'CC' and 'MLO' are used) | List of image paths for CC view for patient_id-laterality combo | List of image paths for MLO view for patient_id-laterality combo | List of [findings, impression] | List of [augmented findings, augmented impression] |
- The final sample csv file as the output of
step3is here: clip_pretrain_100.csv.clip_pretrain_100.csvis used for pretraining the image-text variant of Mammo-CLIP.
We use VinDr dataset as image-label dataset though it can be expanded to any such datasets. Make sure that every patient should have atleast one CC and MLO image per laterality. So if you are planning to use it in the pre-training setup, use the following notebook to preprocess the VinDr dataset:
./src/codebase/notebooks/preprocess-clip/VinDr.ipynbWhen you download the VinDr dataset, you will get these two csv
files: breast-level_annotations.csv
and finding_annotations.csv
. We preprocess the finding_annotations.csv file to
get vindr_detection_v1_folds.csv
. VinDr.ipynb
notebook requires vindr_detection_v1_folds.csv file as input and
generate clip_vindr_final.csv
file.
Both clip_pretrain_100.csv and clip_vindr_final.csv files are used for pretraining the image-text and image-label variant of Mammo-CLIP.
The csv file of the final image-label (VinDr) dataset should have the following format:
| index | patient_id | laterality | image | view | CC | MLO | CC_FINDING | MLO_FINDING |
|---|---|---|---|---|---|---|---|---|
| 0 | patient_id | laterality ('R' or 'L') | List of all image_paths for patient_id-laterality combo | List of views for patient_id-laterality combo (only 'CC' and 'MLO' are used) | List of image paths for CC view for patient_id-laterality combo, e.g, [CC_img ..] | List of image paths for MLO view for patient_id-laterality combo, e.g, [MLO_img .. ] | Findings per image per laterality for CC view (see below for the format) | Findings per image per laterality for MLO view (see below for the format) |
Explanation for CC_FINDING and MLO_FINDING Columns: In the above table, for the row, CC_FINDING can be expanded as:
[
[+ve findings for CC_img if laterality of CC_img is R],
[+ve findings for CC_img if laterality of CC_img is L],
[-ve findings for CC_img if laterality of CC_img is R],
[-ve findings for CC_img if laterality of CC_img is L],
]
As VinDr contains a single image per patient-laterality combo, we did n
Similarly, in the above table, for the row, MLO_FINDING can be expanded as:
[
[+ve findings for MLO_img if laterality of MLO_img is R],
[+ve findings for MLO_img if laterality of MLO_img is L],
[-ve findings for MLO_img if laterality of MLO_img is R],
[-ve findings for MLO_img if laterality of MLO_img is L],
]
Use the following csv files as metadata for the downstream tasks (classification, detection, zero-shot):
| Dataset | CSV |
|---|---|
| VinDr | vindr_detection_v1_folds.csv |
| RSNA | train_folds.csv |
For detection/localization tasks, we have included the coordinates of the resized bounding boxes of VinDr in the above csv file. Somebody interested in resizing the bounding boxes by themselves, run the following command with finding_annotations.csv file as input:
python ./src/preprocessing/preprocess_VinDr_detector.py.
├── list_tree_files.sh
├── upmc_dicom_consolidated_final_folds_BIRADS_num_1_report.csv
├── clip_pretrain_100.csv
└── DICOM/images_png_CC_MLO/
├── Patient_100/
│ ├── 1.png
│ ├── 2.png
└── Patient_200/
├── 3.png
├── 4.png
├── 53.png
├── 6.png
└── 7.png
.
├── breast-level_annotations.csv
├── finding_annotations.csv
├── vindr_detection_v1_folds.csv
├── clip_vindr_final.csv
└── images_png/
├── c7811f4575c1229ad4a7606de49ea68f/
│ ├── 9eb4650a2b630e44074c403f6127c5a1.png
│ ├── cc3fdc5d733a671f3000e20838e192d9.png
│ ├── 181fd193d3b785dc9faafdaa8e1695fc.png
│ └── 55eb5ea616abacd225e584ffc8be57da.png
└── a1dd219b28806fc295fac20ceb147870/
├── 887cdcc99ebed66bd062ada6c8210152.png
├── 36f2921a2ac19eba7420c591c4c07ae4.png
├── 12dc17dfd9d30ea7c0c1ccb33a505085.png
└── e22e4f297b4c82279e7b78a98417a6cd.png.
├── train_folds.csv
├── train_images_png/
├── 59549/
│ ├── 1154694388.png
│ ├── 1192817932.png
│ ├── 1979035704.png
│ ├── 2022274082.png
│ ├── 431013616.png
│ ├── 457600713.png
│ ├── 78005871.png
│ └── 856162422.png
└── 28242/
├── 1966298736.png
├── 233201459.png
├── 349787619.png
└── 98615814.png
Following are the pre-training checkpoints of Mammo-CLIP:
| Model architecture | Checkpoints (Google drive) | Checkpoints (Hugging Face) |
|---|---|---|
| Best performance | Efficient-Net B5 | Efficient-Net B5 |
| Lightweight | Efficient-Net B2 | Efficient-Net B2 |
We have also uploaded the downstream checkpoints for classification and localization (both linear probe and finetuning) with the image encoder of Efficient-Net B5 Mammo-CLIP for fold 0 here.
For pretraining Efficient-Net B5 Mammo-CLIP with a single GPU, use the following command:
python ./src/codebase/train.py --config-name pre_train_b5_clip.yamlFor pretraining Efficient-Net B5 Mammo-CLIP with a 4 GPUs using pytorch-ddp, use the following command:
torchrun --nproc_per_node=4 ./src/codebase/train.py --config-name pre_train_b5_clip.yamlAll the yaml files for the config are
found here.
- Use pre_train_b5_clip.yaml for pre-training image-text variant of Efficient-Net B5 Mammo-CLIP
- Use pre_train_b2_clip.yaml for pre-training image-text variant of Efficient-Net B2 Mammo-CLIP
- Use pre_train_b5_w_vindr_clip.yaml for pre-training image-text + image-label variant of Efficient-Net B5 Mammo-CLIP
- For creating classifiers for downstream evaluations using the image encoder of Mammo-CLIP, use the
class
BreastClipClassifierin breast-clip-classifier.py file. - For creating detectors for downstream evaluations using the image encoder of Mammo-CLIP, use the
function
RetinaNet_efficientnetin detector_model.py file.
FOLD=0
CKPT="b2-model-best-epoch-10.tar"
DIR="./Mammo-CLIP/src/codebase/outputs/upmc_clip/b2_detector_period_n"
FULL_CKPT="$DIR/checkpoints/fold_$FOLD/$CKPT"
python ./src/codebase/eval_zero_shot_clip.py \
--config-name zs_clip.yaml hydra.run.dir=$DIR model.clip_check_point=$FULL_CKPTAdjust the CKPT and DIR variables according to your setup.
python ./src/codebase/train_classifier.py \
--data-dir '/restricted/projectnb/batmanlab/shawn24/PhD/RSNA_Breast_Imaging/Dataset' \
--img-dir 'External/Vindr/vindr-mammo-a-large-scale-benchmark-dataset-for-computer-aided-detection-and-diagnosis-in-full-field-digital-mammography-1.0.0/images_png' \
--csv-file 'External/Vindr/vindr-mammo-a-large-scale-benchmark-dataset-for-computer-aided-detection-and-diagnosis-in-full-field-digital-mammography-1.0.0/vindr_detection_v1_folds.csv' \
--clip_chk_pt_path "/restricted/projectnb/batmanlab/shawn24/PhD/Mammo-CLIP/src/codebase/outputs/upmc_clip/b5_detector_period_n/checkpoints/fold_0/b5-model-best-epoch-7.tar" \
--data_frac 1.0 \
--dataset 'ViNDr' \
--arch 'upmc_breast_clip_det_b5_period_n_lp' \
--label "Mass" \
--epochs 30 \
--batch-size 8 \
--num-workers 0 \
--print-freq 10000 \
--log-freq 500 \
--running-interactive 'n' \
--n_folds 1 \
--lr 5.0e-5 \
--weighted-BCE 'y' \
--balanced-dataloader 'n' data-dir: root directory of the datasetimg-dir: directory containing images, absolute path:data-dir/img-dircsv-file: csv file containing image paths and labels, absolute path:data-dir/csv-fileclip_chk_pt_path: path to the checkpoint of the pre-trained Mammo-CLIP modeldataset: dataset name, e.g.,ViNDrorRSNAdata_frac: fraction of the dataset to use for training, e.g.,1.0,0.5etcarch: architecture of the model, e.g.,upmc_breast_clip_det_b5_period_n_lpfor Efficient-Net B5 orupmc_breast_clip_det_b2_period_n_lpfor Efficient-Net B2, pretrained on UPMC dataset. Also,upmc_vindr_breast_clip_det_b5_period_n_lpfor Efficient-Net B5 orupmc_vindr_breast_clip_det_b2_period_n_lpfor Efficient-Net B2, pretrained on UPMC and VinDr datasets.label: target label for classification, e.g.,Mass,Suspicious_Calcificationordensityfor ViNDr dataset;cancerfor RSNA datasetrunning-interactive: running on interactive mode. In this mode,the training will be done using 100 samples for sanity check
python ./src/codebase/train_classifier.py \
--data-dir '/restricted/projectnb/batmanlab/shawn24/PhD/RSNA_Breast_Imaging/Dataset' \
--img-dir 'External/Vindr/vindr-mammo-a-large-scale-benchmark-dataset-for-computer-aided-detection-and-diagnosis-in-full-field-digital-mammography-1.0.0/images_png' \
--csv-file 'External/Vindr/vindr-mammo-a-large-scale-benchmark-dataset-for-computer-aided-detection-and-diagnosis-in-full-field-digital-mammography-1.0.0/vindr_detection_v1_folds.csv' \
--clip_chk_pt_path "/restricted/projectnb/batmanlab/shawn24/PhD/Mammo-CLIP/src/codebase/outputs/upmc_clip/b5_detector_period_n/checkpoints/fold_0/b5-model-best-epoch-7.tar" \
--data_frac 1.0 \
--dataset 'ViNDr' \
--arch 'upmc_breast_clip_det_b5_period_n_ft' \
--label "Mass" \
--epochs 30 \
--batch-size 8 \
--num-workers 0 \
--print-freq 10000 \
--log-freq 500 \
--running-interactive 'n' \
--n_folds 1 \
--lr 5.0e-5 \
--weighted-BCE 'y' \
--balanced-dataloader 'n'data-dir: root directory of the datasetimg-dir: directory containing images, absolute path:data-dir/img-dircsv-file: csv file containing image paths and labels, absolute path:data-dir/csv-fileclip_chk_pt_path: path to the checkpoint of the pre-trained Mammo-CLIP modeldataset: dataset name, e.g.,ViNDrorRSNAdata_frac: fraction of the dataset to use for training, e.g.,1.0,0.5etcarch:arch: architecture of the model, e.g.,upmc_breast_clip_det_b5_period_n_ftfor Efficient-Net B5 orupmc_breast_clip_det_b2_period_n_ftfor Efficient-Net B2, pretrained on UPMC dataset. Also,upmc_vindr_breast_clip_det_b5_period_n_ftfor Efficient-Net B5 orupmc_vindr_breast_clip_det_b2_period_n_ftfor Efficient-Net B2, pretrained on UPMC and VinDr datasets.label: target label for classification, e.g.,Mass,Suspicious_Calcificationordensityfor ViNDr dataset;cancerfor RSNA datasetrunning-interactive: running on interactive mode. In this mode,the training will be done using 100 samples for sanity check
python ./src/codebase/train_detector.py \
--data-dir '/restricted/projectnb/batmanlab/shawn24/PhD/RSNA_Breast_Imaging/Dataset' \
--img-dir 'External/Vindr/vindr-mammo-a-large-scale-benchmark-dataset-for-computer-aided-detection-and-diagnosis-in-full-field-digital-mammography-1.0.0/images_png' \
--csv-file 'External/Vindr/vindr-mammo-a-large-scale-benchmark-dataset-for-computer-aided-detection-and-diagnosis-in-full-field-digital-mammography-1.0.0/vindr_detection_v1_folds.csv' \
--clip_chk_pt_path "/restricted/projectnb/batmanlab/shawn24/PhD/Mammo-CLIP/src/codebase/outputs/upmc_clip/b5_detector_period_n/checkpoints/fold_0/b5-model-best-epoch-7.tar" \
--dataset 'ViNDr' \
--arch 'clip_b5_upmc' \
--epochs 120 \
--batch-size 7 \
--freeze_backbone "y" \
--data_frac 1.0 \
--concepts 'Mass' \
--print-freq 5000 \
--log-freq 300 \
--running-interactive 'n' \
--focal-alpha 0.25 \
--focal-gamma 2.0 \
--score-threshold 0.2data-dir: root directory of the datasetimg-dir: directory containing images, absolute path:data-dir/img-dircsv-file: csv file containing image paths and labels, absolute path:data-dir/csv-fileclip_chk_pt_path: path to the checkpoint of the pre-trained Mammo-CLIP modeldataset: dataset name, e.g.,ViNDrdata_frac: fraction of the dataset to use for training, e.g.,1.0,0.5etcarch: architecture of the model, e.g.,clip_b5_upmcfor Efficient-Net B5 orclip_b2_upmcfor Efficient-Net B2, pretrained on UPMC dataset. Similarly,clip_b5_upmc_vindrfor Efficient-Net B5 orclip_b2_upmc_vindrfor Efficient-Net B2, pretrained on UPMC and VinDr datasets.concepts: target label for classification, e.g.,Mass,Suspicious Calcificationfor ViNDr datasetrunning-interactive: running on interactive mode. In this mode,the training will be done using 100 samples for sanity checkfreeze_backbone: freeze the backbone of the model, for linear probe, set toy
python ./src/codebase/train_detector.py \
--data-dir '/restricted/projectnb/batmanlab/shawn24/PhD/RSNA_Breast_Imaging/Dataset' \
--img-dir 'External/Vindr/vindr-mammo-a-large-scale-benchmark-dataset-for-computer-aided-detection-and-diagnosis-in-full-field-digital-mammography-1.0.0/images_png' \
--csv-file 'External/Vindr/vindr-mammo-a-large-scale-benchmark-dataset-for-computer-aided-detection-and-diagnosis-in-full-field-digital-mammography-1.0.0/vindr_detection_v1_folds.csv' \
--clip_chk_pt_path "/restricted/projectnb/batmanlab/shawn24/PhD/Mammo-CLIP/src/codebase/outputs/upmc_clip/b5_detector_period_n/checkpoints/fold_0/b5-model-best-epoch-7.tar" \
--dataset 'ViNDr' \
--arch 'clip_b5_upmc' \
--epochs 120 \
--batch-size 7 \
--freeze_backbone "n" \
--data_frac 1.0 \
--concepts 'Mass' \
--print-freq 5000 \
--log-freq 300 \
--running-interactive 'n' \
--focal-alpha 0.25 \
--focal-gamma 2.0 \
--score-threshold 0.2data-dir: root directory of the datasetimg-dir: directory containing images, absolute path:data-dir/img-dircsv-file: csv file containing image paths and labels, absolute path:data-dir/csv-fileclip_chk_pt_path: path to the checkpoint of the pre-trained Mammo-CLIP modeldataset: dataset name, e.g.,ViNDrdata_frac: fraction of the dataset to use for training, e.g.,1.0,0.5etcarch: architecture of the model, e.g.,clip_b5_upmcfor Efficient-Net B5 orclip_b2_upmcfor Efficient-Net B2, pretrained on UPMC dataset. Similarly,clip_b5_upmc_vindrfor Efficient-Net B5 orclip_b2_upmc_vindrfor Efficient-Net B2, pretrained on UPMC and VinDr datasets.concepts: target label for classification, e.g.,Mass,Suspicious Calcificationfor ViNDr datasetrunning-interactive: running on interactive mode. In this mode,the training will be done using 100 samples for sanity checkfreeze_backbone: freeze the backbone of the model, for finetune, set ton
- For a quick look at setting up the downstream classifier, follow the notebook: Downstream_classifier_tutorial.ipynb
- For a quick look at downloading the image embeddings from the vision encoder of Mammo-CLIP, follow the notebook: Get_Embedding_Vision_encoder_Mammo_CLIP_tutorial.ipynb
For all the training scripts, we add them in the scripts directory:
| Scripts | Purpose |
|---|---|
| pretrain_mammo_clip_b5.sh | Pretrain Mammo-CLIP b5 with image+text data |
| pretrain_mammo_clip_b5_ddp.sh | Pretrain Mammo-CLIP b5 with image+text data using multiple GPUs |
| pretrain_mammo_clip_b2.sh | Pretrain Mammo-CLIP b2 with image+text data |
| pretrain_mammo_clip_b2_ddp.sh | Pretrain Mammo-CLIP b2 with image+text data using multiple GPUs |
| pretrain_mammo_clip_w_vindr_b5.sh | Pretrain Mammo-CLIP b5 with image+text data and image+label data |
| classifier_fine_tune_b5.sh | Evaluate Mammo-CLIP b5 on fine tuning tasks for classification |
| classifier_fine_tune_b2.sh | Evaluate Mammo-CLIP b2 on fine tuning tasks for classification |
| classifier_linear_probe_b5.sh | Evaluate Mammo-CLIP b5 on linear probing tasks for classification |
| classifier_linear_probe_b2.sh | Evaluate Mammo-CLIP b2 on linear probing tasks for classification |
| detector_fine_tune_b5.sh | Evaluate Mammo-CLIP b5 on fine tuning tasks for detection |
| detector_fine_tune_b2.sh | Evaluate Mammo-CLIP b2 on fine tuning tasks for detection |
| detector_linear_probe_b5.sh | Evaluate Mammo-CLIP b5 on linear probing tasks for detection |
| detector_linear_probe_b2.sh | Evaluate Mammo-CLIP b2 on linear probing tasks for detection |
For training Mammo-FActOR, refer to the following notebook.
@InProceedings{10.1007/978-3-031-72390-2_59,
author="Ghosh, Shantanu
and Poynton, Clare B.
and Visweswaran, Shyam
and Batmanghelich, Kayhan",
editor="Linguraru, Marius George
and Dou, Qi
and Feragen, Aasa
and Giannarou, Stamatia
and Glocker, Ben
and Lekadir, Karim
and Schnabel, Julia A.",
title="Mammo-CLIP: A Vision Language Foundation Model to Enhance Data Efficiency and Robustness in Mammography",
booktitle="Medical Image Computing and Computer Assisted Intervention -- MICCAI 2024",
year="2024",
publisher="Springer Nature Switzerland",
address="Cham",
pages="632--642",
abstract="The lack of large and diverse training data on Computer-Aided Diagnosis (CAD) in breast cancer detection has been one of the concerns that impedes the adoption of the system. Recently, pre-training with large-scale image text datasets via Vision-Language models (VLM) (e.g., CLIP) partially addresses the issue of robustness and data efficiency in computer vision (CV). This paper proposes Mammo-CLIP, the first VLM pre-trained on a substantial amount of screening mammogram-report pairs, addressing the challenges of dataset diversity and size. Our experiments on two public datasets demonstrate strong performance in classifying and localizing various mammographic attributes crucial for breast cancer detection, showcasing data efficiency and robustness similar to CLIP in CV. We also propose Mammo-FActOR, a novel feature attribution method, to provide spatial interpretation of representation with sentence-level granularity within mammography reports. Code is available publicly: https://github.com/batmanlab/Mammo-CLIP.",
isbn="978-3-031-72390-2"
}Licensed under the Creative Commons Attribution 4.0 International
Copyright © Batman Lab, 2024
For any queries, contact Shantanu Ghosh (email: [email protected])
Special thanks to Boston University Masters students Abhishek Varshney & Akshat Gurbuxani for enabling multi-GPU support to Mammo-CLIP.
Did you try Mammo-CLIP on other datasets containing 2D-Mammograms and want to report the results? Feel free to send a pull request.
