MMDB

PCMMD: A Novel Dataset of Plasma Cells to Support the Diagnosis of Multiple Myeloma is a novel dataset of plasma cells.

This repository contains the database with annotations of plasma cells and non-plasma cells, as well as the diagnoses of 10 patients with multiple myeloma.

Original Manuscript:

LB ANDRADE, Caio et al. PCMMD: A Novel Dataset of Plasma Cells to Support the Diagnosis of Multiple Myeloma. Scientific Data, v. 12, n. 1, p. 161, 2025. URL: (https://www.nature.com/articles/s41597-025-04459-1).

In addition to the database, a training was performed using the deep learning model Yolo version 8 for cell identification, and diagnosis based on the population of plasma cells.

Project Structure

The following folder division was made for dataset organization and to conduct a benchmark with the goal of diagnosing 10 patients. Cross-validation with 5 folds was performed, and the model's variance against the dataset was checked.

• data/:
  • detection/: Data for object detection.
    • patients/: It contains the images, annotations of plasma cells and non-plasma cells, and the diagnoses of 10 patients.
    • train/: It contains the images and annotations of plasma cells and non-plasma cells from various patients.
    • mieloma_fold_{i}.yaml Configuration file for each fold i used for training.
    • split_{1}.txt: File with the path of the images for each fold i.
    • mieloma_final.yaml: Final file used to train with all folds and test with the 10 patients.
  • segmentation/: Data for instance segmentation.
    • plasma cells/: It contains cropped images and masks of plasma cells.
    • non-plasma cells/: It contains cropped images and masks of non-plasma cells.
• src:
  • dignosis.ipynb: Jupyter notebook that performs the diagnosis of the 10 patients.
  • plot.ipynb: Jupyter notebook that visualizes detection and segmentation results for analysis and debugging.
  • weights.pt: Yolo v8 weights from the final training used for diagnosis.

Benchmarking reproduction

Instalation

To setup and install Yolov8: ultralytics

---

Train:

To train model, use folds mieloma_fold_{i}.yaml or final file mieloma_final.yaml in Yolov8.

from ultralytics import YOLO

# Load a model
model = YOLO("yolov8n.pt")  # load a pretrained model (recommended for training)

# Use the model
model.train(data="mieloma_final.yaml.yaml", epochs=300)  # train the model
metrics = model.val()  # evaluate model performance on the validation set

Test

Use MMDB/src/diagnosis.ipynb for test.

📝 Citation

If you find the dataset useful for your research, please consider citing:

@article{lb2025pcmmd,
  title={PCMMD: A Novel Dataset of Plasma Cells to Support the Diagnosis of Multiple Myeloma},
  author={LB Andrade, Caio and Ferreira, Marcos V and M. Alencar, Brenno and SB Filho, Jorge L and A. Guimaraes, Matheus and Porto Cruz Moraes, Iarley and S. Lopes, Tiago J and S. dos Santos, Allan and M. dos Santos, Mariane and CS e Silva, Maria I and others},
  journal={Scientific Data},
  volume={12},
  number={1},
  pages={161},
  year={2025},
  publisher={Nature Publishing Group UK London}
}