diff --git a/README.md b/README.md
index c040bd037d1dd6f70dc2175651245fd6e8f3f19d..b06fb4a7f5c874eff17184855e2edc9cd71b0c53 100644
--- a/README.md
+++ b/README.md
@@ -98,7 +98,8 @@ The results are saved in a folder called `aligned` created in the input director
 
 # Demo/Test
 
-To test this plugin, you can use our [demo dataset](./data/demo_input) proposed in this repository or directly on your own dataset.
+To test this plugin, you can use our [demo dataset](https://zenodo.org/records/15028169) proposed in a Zenodo repository or directly on your own dataset.
+If you use our demo dataset, please use the files [220817mdxGAKAYFPTMX1FDB_Movie-01-Scene-11-TR27.tif](https://zenodo.org/records/15028169/files/220817mdxGAKAYFPTMX1FDB_Movie-01-Scene-11-TR27_beforeAlignment.tif?download=1) and its [corresponding TrackMate file](https://zenodo.org/records/15028169/files/220817mdxGAKAYFPTMX1FDB_Movie-01-Scene-11-TR27.xml?download=1). Rename the movie to remove the `_beforeAlignement` mark.
 
 ## Steps
 The input of the process should be a folder containing your unaligned stacks with two channels (the fiber in transmitted light and the fluorescent channel) and the corresponding TrackMate files if you have already tracks on the unaligned movie (not mandatory).
@@ -110,7 +111,7 @@ Press `Ok` to launch it. If you selected the `visible` mode, images will pop up
 
 ## Output
 At the end of the process, you will get a folder named `aligned` in which all processed stacks will be saved (registered stacks) with the corresponding TrackMate files if you selected the option.
-You can see what you can obtain with our [demo dataset](./data/demo_output).
+You can see what you can obtain with our [demo dataset](https://zenodo.org/records/15028169). You can check the [segmented fiber movie](https://zenodo.org/records/15028169/files/220817mdxGAKAYFPTMX1FDB_Movie-01-Scene-11-TR27_toalignFiber2D.tif?download=1) used to calculate the registration, the [fiber channel after registration](https://zenodo.org/records/15028169/files/220817mdxGAKAYFPTMX1FDB_Movie-01-Scene-11-TR27_alignedFiber3D.tif?download=1) and the resulting [aligned movie](https://zenodo.org/records/15028169/files/220817mdxGAKAYFPTMX1FDB_Movie-01-Scene-11-TR27.tif?download=1).
 
 **Running time**: On a standard computer with one GPU, running the plugin on one image take typically around 50 min for a movie wit 12 Z-slices and 280 time frames.
 
@@ -118,4 +119,4 @@ You can see what you can obtain with our [demo dataset](./data/demo_output).
 
 Our plugin and trained neural networks are availabel freely and open-source under the BDS3-license. 
 
-If you use it, please cite our paper [Sarde et al.](https://www.biorxiv.org/content/10.1101/2025.03.13.643016v1). 
\ No newline at end of file
+If you use it, please cite our paper [Sarde et al., 2025](https://www.biorxiv.org/content/10.1101/2025.03.13.643016v1). 
\ No newline at end of file