By default, the app uses the jupyter/datascience-notebook. The upstream Jupyterhub project maintains many other notebook images.

To use a different notebook image, use the File Manager to place custom configuration under /app/data/customconfig.py. For example, add a line like below:
c.DockerSpawner.image = 'quay.io/jupyter/all-spark-notebook:lab-4.1.5'

It is also possible to use any arbitrary docker image built from jupyter/base-notebook or any of the specialized notebooks that are base on this. For example:
FROM quay.io/jupyter/all-spark-notebook
RUN conda install --yes -c conda-forge git-lfs

Build and push the above image to Dockerhub. Then, update c.DockerSpawner.image as shown above.

To apply the configuration, restart the app using the Restart button.

Info:
Remove existing notebook containers
For the container image to take effect, you have to remove any existing docker notebook containers using the /app/code/remove_notebook_containers.py script. Notebook data will be intact despite deleting the container.