File size: 2,312 Bytes
9e96356
ae2f7bc
fd700f4
9e96356
fd3a519
9e96356
6a1b31c
5bfba9e
8ffd27f
 
e2ca3e5
 
 
 
 
 
 
 
 
 
 
 
 
7a28a51
e2ca3e5
 
 
 
 
 
 
 
7a28a51
e2ca3e5
 
 
 
 
 
 
 
7a28a51
e2ca3e5
 
 
 
 
 
 
 
7a28a51
e2ca3e5
 
 
 
 
 
 
 
 
 
5bfba9e
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
# GenerateCT: Text-Guided 3D Chest CT Generation
Welcome to the official page of GenerateCT, a pioneering work in text-conditional 3D medical image generation with a particular focus on chest CT volumes. Here, you will find the code, all trained models for text-to-CT generation, and a dataset of generated CT volumes paired with text prompts, all freely accessible for researchers.

<p align="center">
  <img src="https://huggingface.co/generatect/GenerateCT/resolve/main/figures/examples.gif" width="100%">
</p>

## Results
Our models achieve the following performances:

<table>
  <thead>
    <tr style="text-align: center; background-color: #f8f8f8;">
      <th>Model</th>
      <th>Resolution (px)</th>
      <th>Dimension</th>
      <th>Text-Guided?</th>
      <th>FID Score (↓)</th>
      <th>FVD Score (↓)</th>
      <th>CLIP Score (↑)</th>
    </tr>
  </thead>
  <tbody>
    <tr style="text-align: center;">
      <td>CT-ViT</td>
      <td>128</td>
      <td>3D</td>
      <td>No</td>
      <td>73.4</td>
      <td>1817.4</td>
      <td>N/A</td>
    </tr>
    <tr style="text-align: center;">
      <td>Transformer</td>
      <td>128</td>
      <td>3D</td>
      <td>Yes</td>
      <td>104.3</td>
      <td>1886.8</td>
      <td>25.2</td>
    </tr>
    <tr style="text-align: center;">
      <td>Diffusion</td>
      <td>512</td>
      <td>2D</td>
      <td>Yes</td>
      <td>14.9</td>
      <td>409.8</td>
      <td>27.6</td>
    </tr>
    <tr style="font-weight: bold; text-align: center;">
      <td>GenerateCT</td>
      <td>512</td>
      <td>3D</td>
      <td>Yes</td>
      <td>55.8</td>
      <td>1092.3</td>
      <td>27.1</td>
    </tr>
  </tbody>
</table>

## License
Our work, including the codes, trained models, and generated data, is released under a [Creative Commons Attribution (CC-BY) license](https://creativecommons.org/licenses/by/4.0/). This means that anyone is free to share (copy and redistribute the material in any medium or format) and adapt (remix, transform, and build upon the material) for any purpose, even commercially, as long as appropriate credit is given, a link to the license is provided, and any changes that were made are indicated. This aligns with our goal of facilitating progress in the field by providing a resource for researchers to build upon.