Machine-Learning-Based-Radiodosiomics-and-Swin-UNETR-Framework-for-Dose-Prediction

This repository presents a radiodosiomics framework for personalized [¹⁷⁷Lu]Lu-PSMA-617 RLT in patients with metastatic castration-resistant prostate cancer (mCRPC). It includes feature selection and ML models using clinical biomarkers and radiomic and dosiomic (radiodosiomic) features extracted from pretreatment [⁶⁸Ga]Ga-PSMA-11 PET/CT, plus Swin UNETR model with SSL to predict Monte Carlo–based dose rate maps.

🤖 Machine Learning Pipeline 🤖

See radiodosiomics_ML.ipynb for the full pipeline including:

• Feature selection (RFE, Boruta, LASSO, Mutual Information, and Elastic Net)
• Model training and evaluation
• Integration of clinical, radiomic, and dosiomic features

🧠 Deep Learning Pipeline 🧠

🧹 1. Preprocessing

Before pretraining and fine-tuning, data (PET and CT images) should be preprocessed:

python preprocess.py --in_dir=<Input-directory(PET and CT)> --out_dir=<Output-directory>

🏋️ 2. Pre-Training

Pre-Train Swin UNETR encoder on unlabeled data

python main.py --exp=<Experiment Name> --in_channels=2 --data_dir=<Data-Path> --json_list=<Json List Path> \
--lr=6e-6 --lrdecay --batch_size=<Batch Size> --num_steps=<Number of Steps>

3. 🛠️ Fine-Tuning

Fine-Tuning Swin UNETR on labeled data:

python main.py --exp=<Experiment Name> --data_dir=<Data-Path> --json_list=<Json List Path> --in_channels=2 --out_channels=1 \
--pretrained_model_name=<Pretrained Encoder Name> --batch_size=<Batch Size> --max_epochs=<Epochs> --use_ssl_pretrained \
--ssl_pretrained_path=<Pretrained Model Path> --use_checkpoint

📊 4. Evaluation

Evaluating Swin UNETR

python test.py --pretrained_dir=<Pretrained Model Path> --data_dir=<Data-Path> --exp_name=<Experiment Name> \
--json_list=<Json List Path> --pretrained_model_name=<Pretrained Model Name> --save

⚙️ Install Dependencies

Install dependencies using:

pip install -r requirements.txt

📚 Citation

If you find our work useful, please cite the following paper:

Yazdani E, Neizehbaz A, Karamzade‐Ziarati N, Emami F, Vosoughi H, Asadi M, Mahmoudi A, Sadeghi M, Kheradpisheh SR, Geramifar P.
Transforming [177Lu] Lu‐PSMA‐617 treatment planning: Machine learning‐based radiodosiomics and swin UNETR using pretherapy PSMA positron emission tomography/computed tomography (PET/CT).
Medical Physics. 2025 Oct;52(10):e70030. https://doi.org/10.1002/mp.70030

Acknowledgement

Models Implantation and SSL Pipeline are based on MONAI and This repository.