Update README.md

README.md

@@ -176,3 +176,23 @@ Instructions on how to build and run sket images are reported below, if you alre
 10) If you want to build the images again, from the project folder type ```docker-compose down --rmi local```, pay attention that this command will remove all the images created (both CPU and GPU). If you want to remove only one image between CPU and GPU see the [docker image documentation](https://docs.docker.com/engine/reference/commandline/image/). Finally repeat steps 5-8. 
 
 Regarding SKET GPU-enabled, the corresponding Dockerfile (you can find the Dockerfile at the following path: sket_server/docker-sket_server-config/sket_gpu) contains the ```nvidia/cuda:11.0-devel```. Users are encouraged to change the NVIDIA/CUDA image within the Dockerfile depending on the NVIDIA drivers installed in their host machine. NVIDIA images can be found [here](https://hub.docker.com/r/nvidia/cuda/tags?page=1&ordering=last_updated).
+
+## Cite
+
+If you use or extend our work, please cite the following:
+
+```
+@article{npj_dig_med-2022,
+  title = "Unleashing the potential of digital pathology data by training computer-aided diagnosis models without human annotations",
+  author = "N. Marini and S. Marchesin and S. Ot\'{a}lora and M. Wodzinski and A. Caputo and M. van Rijthoven and W. Aswolinskiy and J. M. Bokhorst and D. Podareanu and E. Petters and S. Boytcheva and G. Buttafuoco and S. Vatrano and F. Fraggetta and J. der Laak and M. Agosti and F. Ciompi and G. Silvello and H. Muller and M. Atzori",
+  journal = "npj Digital Medicine",
+  year = "2022",
+  url = "http://dx.doi.org/10.1038/s41746-022-00635-4",
+  doi = "10.1038/s41746-022-00635-4",
+  volume = "5",
+  number = "1",
+  pages = "1--18"
+}
+```
+
+We have another submission ongoing that only focuses on SKET and explainability, stay tuned!