zaccharieramzi commited on
Commit
d913147
1 Parent(s): dda1e26

added model card and model weights

Browse files
Files changed (2) hide show
  1. README.md +71 -0
  2. model_weights.h5 +3 -0
README.md ADDED
@@ -0,0 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # UNet-OASIS
2
+ ---
3
+ tags:
4
+ - TensorFlow
5
+ - MRI reconstruction
6
+ - MRI
7
+ datasets:
8
+ - OASIS
9
+ ---
10
+
11
+ This model can be used to reconstruct single coil OASIS data with an acceleration factor of 4.
12
+
13
+ ## Model description
14
+ For more details, see https://www.mdpi.com/2076-3417/10/5/1816.
15
+ This section is WIP.
16
+
17
+ ## Intended uses and limitations
18
+ This model can be used to reconstruct single coil brain retrospective data from the OASIS database at acceleration factor 4.
19
+ It cannot be used on multi-coil data.
20
+
21
+ ## How to use
22
+ This model can be loaded using the following repo: https://github.com/zaccharieramzi/fastmri-reproducible-benchmark.
23
+ After cloning the repo, `git clone https://github.com/zaccharieramzi/fastmri-reproducible-benchmark`, you can install the package via `pip install fastmri-reproducible-benchmark`.
24
+ The framework is TensorFlow.
25
+
26
+ You can initialize and load the model weights as follows:
27
+ ```python
28
+ from fastmri_recon.models.functional_models.unet import unet
29
+
30
+
31
+ model = unet(n_layers=4, layers_n_channels=[16, 32, 64, 128], layers_n_non_lins=2,)
32
+ model.load_weights('UNet-fastmri/model_weights.h5')
33
+
34
+ ```
35
+
36
+ Using the model is then as simple as:
37
+ ```python
38
+ model(zero_filled_recon)
39
+ ```
40
+
41
+ ## Limitations and bias
42
+ The limitations and bias of this model have not been properly investigated.
43
+
44
+ ## Training data
45
+ This model was trained using the [OASIS dataset](https://www.oasis-brains.org/).
46
+
47
+ ## Training procedure
48
+ The training procedure is described in https://www.mdpi.com/2076-3417/10/5/1816 for brain data.
49
+ This section is WIP.
50
+
51
+ ## Evaluation results
52
+ This model was evaluated using the [OASIS dataset](https://www.oasis-brains.org/).
53
+
54
+ - PSNR: 29.8
55
+ - SSIM: 0.847
56
+
57
+
58
+ ## Bibtex entry
59
+ ```
60
+ @article{ramzi2020benchmarking,
61
+ title={Benchmarking MRI reconstruction neural networks on large public datasets},
62
+ author={Ramzi, Zaccharie and Ciuciu, Philippe and Starck, Jean-Luc},
63
+ journal={Applied Sciences},
64
+ volume={10},
65
+ number={5},
66
+ pages={1816},
67
+ year={2020},
68
+ publisher={Multidisciplinary Digital Publishing Institute}
69
+ }
70
+ ```
71
+
model_weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7d9bb8fef30e0054ab25bb0477ea7960eebef34d1095e7eef7e14bcf463b6d46
3
+ size 5919696