Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -1,119 +1,105 @@
|
|
1 |
---
|
2 |
-
|
3 |
-
|
4 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5 |
---
|
6 |
-
Comming soon...!
|
7 |
-
# Dataset Card for Dataset Name
|
8 |
|
9 |
-
|
10 |
|
11 |
-
|
12 |
-
- **Repository:**
|
13 |
-
- **Paper:**
|
14 |
-
- **Leaderboard:**
|
15 |
-
- **Point of Contact:**
|
16 |
|
17 |
-
|
18 |
|
19 |
-
|
|
|
20 |
|
21 |
-
|
22 |
|
23 |
-
|
24 |
|
25 |
-
|
|
|
26 |
|
27 |
-
|
28 |
|
29 |
-
|
30 |
-
|
31 |
-
### Data Instances
|
32 |
|
33 |
-
|
34 |
|
35 |
-
|
|
|
|
|
36 |
|
37 |
-
|
38 |
|
39 |
-
|
40 |
|
41 |
-
|
|
|
|
|
42 |
|
43 |
## Dataset Creation
|
44 |
|
45 |
### Curation Rationale
|
46 |
|
47 |
-
|
|
|
|
|
48 |
|
49 |
### Source Data
|
50 |
|
51 |
-
|
52 |
-
|
53 |
-
[More Information Needed]
|
54 |
-
|
55 |
-
#### Who are the source language producers?
|
56 |
-
|
57 |
-
[More Information Needed]
|
58 |
|
59 |
### Annotations
|
60 |
|
61 |
-
|
62 |
-
|
63 |
-
[More Information Needed]
|
64 |
-
|
65 |
-
#### Who are the annotators?
|
66 |
-
|
67 |
-
[More Information Needed]
|
68 |
-
|
69 |
-
### Personal and Sensitive Information
|
70 |
|
71 |
-
|
72 |
|
73 |
-
|
|
|
|
|
|
|
74 |
|
75 |
-
|
76 |
|
77 |
-
|
78 |
|
79 |
-
|
80 |
|
81 |
-
[More Information Needed]
|
82 |
-
|
83 |
-
### Other Known Limitations
|
84 |
-
|
85 |
-
[More Information Needed]
|
86 |
-
|
87 |
-
## Additional Information
|
88 |
-
|
89 |
-
### Dataset Curators
|
90 |
-
|
91 |
-
[More Information Needed]
|
92 |
-
|
93 |
-
### Licensing Information
|
94 |
-
|
95 |
-
[More Information Needed]
|
96 |
-
|
97 |
-
### Citation Information
|
98 |
```
|
99 |
-
@inproceedings{
|
100 |
-
|
101 |
-
|
102 |
-
|
103 |
-
|
104 |
-
|
105 |
-
address = {New York, NY, USA},
|
106 |
-
url = {https://doi.org/10.1145/3592571.3592978},
|
107 |
-
doi = {10.1145/3592571.3592978},
|
108 |
-
booktitle = {Proceedings of the 4th ACM Workshop on Intelligent Cross-Data Analysis and Retrieval},
|
109 |
-
pages = {1–9},
|
110 |
-
numpages = {9},
|
111 |
-
keywords = {polyp segmentation, polyp generative model, generating synthetic data, diffusion model},
|
112 |
-
location = {Thessaloniki, Greece},
|
113 |
-
series = {ICDAR '23}
|
114 |
}
|
115 |
```
|
116 |
|
117 |
-
|
|
|
|
|
|
|
|
|
118 |
|
119 |
-
|
|
|
|
1 |
---
|
2 |
+
dataset: conditional-polyp-diffusion
|
3 |
+
annotations_creators:
|
4 |
+
- expert-generated
|
5 |
+
language:
|
6 |
+
- en
|
7 |
+
license: apache-2.0
|
8 |
+
multilinguality:
|
9 |
+
- monolingual
|
10 |
+
pretty_name: Conditional Polyp Diffusion
|
11 |
+
size_categories:
|
12 |
+
- 1K<n<10K
|
13 |
+
source_datasets:
|
14 |
+
- original
|
15 |
+
task_categories:
|
16 |
+
- image-generation
|
17 |
+
- image-segmentation
|
18 |
+
task_ids:
|
19 |
+
- image-to-image-translation
|
20 |
+
- semantic-segmentation
|
21 |
+
paperswithcode_id: mask-conditioned-latent-diffusion
|
22 |
---
|
|
|
|
|
23 |
|
24 |
+
# Dataset Card for Conditional Polyp Diffusion
|
25 |
|
26 |
+
## Dataset Summary
|
|
|
|
|
|
|
|
|
27 |
|
28 |
+
The **Conditional Polyp Diffusion** dataset provides synthetic gastrointestinal (GI) polyp images along with segmentation masks, generated using a two-stage diffusion modeling framework. The dataset is aimed at mitigating the challenges of data scarcity and privacy in medical imaging, especially for supervised polyp segmentation tasks.
|
29 |
|
30 |
+
- **Stage 1**: Improved diffusion model generates synthetic segmentation masks.
|
31 |
+
- **Stage 2**: Latent diffusion model generates corresponding realistic polyp images, conditioned on the masks.
|
32 |
|
33 |
+
This dataset enables training and benchmarking of polyp segmentation models, improving generalizability and reducing dependence on scarce annotated real data.
|
34 |
|
35 |
+
## Supported Tasks and Leaderboards
|
36 |
|
37 |
+
- **Image-to-Image Translation**: Generating realistic medical images from segmentation masks.
|
38 |
+
- **Semantic Segmentation**: Supervised training of segmentation models for polyp detection.
|
39 |
|
40 |
+
## Languages
|
41 |
|
42 |
+
The metadata and documentation are in English.
|
|
|
|
|
43 |
|
44 |
+
## Dataset Structure
|
45 |
|
46 |
+
Each sample includes:
|
47 |
+
- A synthetic GI polyp image.
|
48 |
+
- A corresponding segmentation mask.
|
49 |
|
50 |
+
The images are generated to mimic the distribution of Kvasir-SEG masks and HyperKvasir polyp appearances.
|
51 |
|
52 |
+
## Data Splits
|
53 |
|
54 |
+
The dataset contains:
|
55 |
+
- 1,000 synthetic masks
|
56 |
+
- 1,000 corresponding synthetic polyp images
|
57 |
|
58 |
## Dataset Creation
|
59 |
|
60 |
### Curation Rationale
|
61 |
|
62 |
+
Due to privacy and annotation constraints in medical imaging, the dataset addresses:
|
63 |
+
- Lack of large-scale annotated datasets for polyp segmentation.
|
64 |
+
- Need for diverse, high-fidelity training data for robust CAD systems.
|
65 |
|
66 |
### Source Data
|
67 |
|
68 |
+
The improved diffusion model is trained on the Kvasir-SEG dataset’s segmentation masks. The conditional polyp generator is trained using these generated masks to create realistic polyp images.
|
|
|
|
|
|
|
|
|
|
|
|
|
69 |
|
70 |
### Annotations
|
71 |
|
72 |
+
- Masks are generated via diffusion models conditioned on prior distributions.
|
73 |
+
- No manual annotations are provided; instead, generated masks are verified for similarity and diversity.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
74 |
|
75 |
+
## Usage
|
76 |
|
77 |
+
The dataset is intended for research in:
|
78 |
+
- Medical image generation
|
79 |
+
- Semi-supervised and supervised segmentation
|
80 |
+
- Evaluation of synthetic data utility in clinical tasks
|
81 |
|
82 |
+
## Evaluation
|
83 |
|
84 |
+
Three segmentation models (UNet++, FPN, DeepLabv3+) were trained with various combinations of real and synthetic data. Results demonstrated that using synthetic data can improve model performance, particularly with DeepLabv3+ achieving a micro-imagewise IoU of 0.7751.
|
85 |
|
86 |
+
## Citation
|
87 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
88 |
```
|
89 |
+
@inproceedings{machacek2023mask,
|
90 |
+
title={Mask-conditioned latent diffusion for generating gastrointestinal polyp images},
|
91 |
+
author={Macháček, Roman and Mozaffari, Leila and Sepasdar, Zahra and Parasa, Sravanthi and Halvorsen, Pål and Riegler, Michael A and Thambawita, Vajira},
|
92 |
+
booktitle={Proceedings of the 4th Workshop on Intelligent Cross-Data Analysis and Retrieval (ICDAR '23)},
|
93 |
+
year={2023},
|
94 |
+
doi={10.1145/3592571.3592978}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
95 |
}
|
96 |
```
|
97 |
|
98 |
+
## License
|
99 |
+
|
100 |
+
Apache License 2.0
|
101 |
+
|
102 |
+
## Dataset URL
|
103 |
|
104 |
+
- Dataset: [https://huggingface.co/datasets/deepsynthbody/conditional-polyp-diffusion](https://huggingface.co/datasets/deepsynthbody/conditional-polyp-diffusion)
|
105 |
+
- Code: [https://github.com/simulamet-host/conditional-polyp-diffusion](https://github.com/simulamet-host/conditional-polyp-diffusion)
|