saadwazir/MedCAGD-Dataset-Collection

MedCAGD: Context-Aware Gated Decoder for Efficient Medical Image Segmentation (ECCV 2026) - Medical Image Segmentation Datasets

This repo contains multiple publicly available medical image segmentation (Semantic Segmentation) datasets used for training and evaluation of the segmentation models. Each subfolder corresponds to a specific dataset and follows its original structure or a standardized format used in this project. These datasets are used for benchmarking segmentation performance across multiple medical imaging modalities including CT, MRI, dermoscopy, endoscopy, ultrasound, fundus imaging, and microscopy.

The segmentation benchmarks included in this dataset collection correspond to the evaluations reported in the paper (MedCAGD). Detailed benchmark results, evaluation protocols, and methodological descriptions are available in the publication.

Citation

If you use this, please cite the following papers

@inproceedings{wazir2026medcagdcontextawaregateddecoder,
      title={MedCAGD: Context-Aware Gated Decoder for Efficient Medical Image Segmentation}, 
      author={Saad Wazir, Patrick Dominique Vibild, Dinh Phu Tran, Seongah Kim, Daeyoung Kim},
      year={2026},
      eprint={2607.00409},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2607.00409}, 
}

@inproceedings{wazir2025rethinking,
  title={Rethinking decoder design: Improving biomarker segmentation using depth-to-space restoration and residual linear attention},
  author={Wazir, Saad and Kim, Daeyoung},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={30861--30871},
  year={2025},
  doi = {10.48550/arXiv.2506.18335},
  url = {https://doi.org/10.48550/arXiv.2506.18335}
}

MedCAGD: Context-Aware Gated Decoder for Efficient Medical Image Segmentation - ECCV 2026

GitHub | Paper

MCADS-Decoder - Rethinking decoder design: Improving Biomarker Segmentation Using Depth-to-Space Restoration and Residual Linear Attention - CVPR 2025

GitHub | Paper

---

MedCAGD-Dataset-Viewer

https://huggingface.co/spaces/saadwazir/MedCAGD-Dataset-Viewer

Included Datasets:

ACDC-2D-Slices 2D cardiac MRI slices from the ACDC dataset used for cardiac structure segmentation.

Synapse Multi-Organ Segmentation Dataset-8 abdominal organs-2D-Slices
2D slices from the Synapse multi-organ CT segmentation dataset containing annotations for eight abdominal organs.

ThyroidXL
Thyroid ultrasound dataset for thyroid nodule segmentation.

ISIC2017
Skin lesion segmentation dataset from the ISIC 2017 challenge.

ISIC2018
Skin lesion segmentation dataset from the ISIC 2018 challenge.

BKAI
Gastrointestinal polyp segmentation dataset released by BKAI.

ClinicDB
Colonoscopy polyp segmentation dataset from the CVC-ClinicDB benchmark.

ColonDB
Colonoscopy polyp segmentation dataset commonly used for evaluating polyp detection models.

ETIS
ETIS-Larib polyp dataset containing challenging colonoscopy images with pixel-level annotations.

Kvasir
Kvasir-SEG dataset for gastrointestinal polyp segmentation.

BUSI
Breast ultrasound dataset used for tumor segmentation.

DRIVE
Retinal vessel segmentation dataset from the DRIVE challenge.

FIVES
Fundus Image Vessel Segmentation dataset for retinal blood vessel analysis.

CHASEDB Retinal vessel segmentation dataset from the CHASE_DB1 benchmark.

LES-AV
Dataset for retinal artery and vein segmentation.

STARE
Retinal vessel segmentation dataset from the STARE benchmark.

UOA-DR
Diabetic retinopathy dataset with retinal lesion annotations.

CellSeg
Microscopy cell segmentation dataset used for cell instance or semantic segmentation.

---

Benchmark Results

Table: 1 - Comprehensive performance comparison across 9 medical image segmentation benchmarks. Average Dice scores are reported.

Method	Params ↓	FLOPs ↓	Skin		Polyp		Fundus		Neoplasm		Cell	All
			ISIC17	ISIC18	ETIS	ColonDB	DRIVE	FIVES	BUSI	ThyroidXL	CellSeg	Avg
U-Net	34.53 M	65.53 G	83.07	86.67	76.85	83.95	71.20	75.77	74.04	71.16	71.52	77.14
AttnUNet	34.88 M	66.64 G	83.66	87.05	76.84	86.46	71.68	75.99	74.48	72.50	72.64	77.92
DeepLabv3+	39.76 M	14.92 G	83.84	88.64	90.73	91.92	69.59	75.12	76.81	73.46	71.90	80.22
UNet++	09.16 M	34.65 G	82.98	87.46	77.40	87.88	72.94	85.74	74.46	83.94	78.30	81.23
nnU-Net	31.29 M	55.26 G	83.23	88.53	80.13	91.63	75.43	76.10	76.46	86.08	83.53	82.34
PraNet	32.55 M	06.93 G	83.03	88.56	83.84	89.16	75.21	84.57	75.14	85.51	79.07	82.68
TransUNet	105.32 M	38.52 G	85.00	89.16	87.79	91.63	74.98	83.54	78.30	85.77	79.08	83.92
Swin-Unet	27.17 M	06.20 G	83.97	89.26	85.10	89.27	74.93	84.17	77.38	85.80	78.84	83.19
UCTransNet	65.60 M	56.70 G	83.27	89.18	87.35	91.65	75.42	84.74	79.53	85.82	79.33	84.03
UNeXt	1.470 M	0.570 G	82.74	87.78	74.03	83.84	74.77	76.60	74.71	84.46	75.71	79.40
VM-UNet	27.43 M	04.12 G	85.99	87.05	85.52	88.71	73.25	83.51	74.69	78.31	74.94	81.33
Swin-UMamba	60.00 M	68.00 G	83.40	87.62	86.63	87.97	73.32	82.66	73.38	84.96	75.56	81.72
EMCAD	26.76 M	05.60 G	85.95	90.96	92.29	92.31	77.15	82.51	80.25	83.33	79.13	84.87
MCADS	50.90 M	61.89 G	84.14	91.01	92.24	91.37	78.42	76.05	80.03	86.33	86.68	85.14
Ours	30.60 M	05.00 G	86.61	91.56	93.47	93.27	81.63	87.50	83.47	88.02	86.61	88.01
AutoSam*	41.56 M	25.11 G	-	-	79.70	83.00	-	-	-	-	-	-
Medical SAM3*	840.0 M	-	-	-	86.10	-	55.80	-	-	-	-	-

Table: 2 - Performance comparison with SOTA methods on the Synapse multi-organ dataset.

Method	Dice ↑	IoU ↑	HD95 ↓	Aorta	GB	KL	KR	Liver	PC	SP	SM
U-Net	70.11	59.39	44.69	84.00	56.70	72.41	62.64	86.98	48.73	81.48	67.96
AttnUNet	71.70	68.09	26.01	84.04	66.42	57.26	84.53	81.28	73.87	66.06	60.17
UNet++	72.39	68.82	25.61	83.65	67.66	57.26	84.53	81.34	73.87	68.97	61.85
nnU-Net	75.33	71.47	19.34	77.06	73.27	76.34	84.53	79.98	73.34	77.62	60.52
PraNetV2	83.75	74.81	17.77	88.69	72.79	85.41	82.91	95.82	68.47	93.09	85.85
TransUNet	77.61	67.32	26.90	86.56	60.43	80.54	78.53	94.33	58.47	87.06	75.00
Swin-Unet	77.58	66.88	27.32	81.76	65.95	82.32	79.22	93.73	53.81	88.04	75.79
UCTransNet	79.08	75.41	15.59	83.06	81.35	77.24	78.23	85.76	74.77	81.89	70.31
UNETR*	78.35	-	18.59	89.80	56.30	85.60	84.52	94.57	60.47	85.00	70.46
MISSFormer*	81.96	-	18.20	86.99	68.65	85.21	82.00	94.41	65.67	91.92	80.81
U-Mamba	78.63	74.87	16.19	83.77	78.70	79.40	82.37	83.86	74.78	79.77	66.41
VM-UNet	73.39	71.61	27.97	63.57	72.62	77.98	92.59	79.44	70.80	55.58	74.55
EMCAD	83.63	74.65	15.68	88.14	68.87	88.08	84.10	95.26	68.51	92.17	83.92
MCADS	85.03	81.71	11.11	90.81	86.07	86.77	83.24	87.66	83.55	85.74	76.38
Ours	87.00±0.2	83.77	14.39	92.28	90.31	89.72	87.21	91.02	82.08	86.91	76.51
Self-Prompt SAM*	86.74	-	-	91.99	69.95	85.65	85.40	97.39	79.18	94.38	89.94

Table: 3 - Performance comparison with SOTA methods on the ACDC dataset.

Method	Dice ↑	IoU ↑	HD95 ↓	RV	Myo	LV
U-Net	81.56	73.41	6.9854	76.99	80.28	87.43
AttnUNet	82.37	73.94	6.1684	78.13	81.08	87.89
UNet++	81.97	73.92	6.4724	77.74	80.73	87.44
nnU-Net	82.66	74.27	6.1663	79.00	81.01	87.97
PraNetV2	83.74	76.13	6.3719	79.61	83.10	88.51
TransUNet	83.07	74.85	5.7578	79.16	81.65	88.41
Swin-Unet	82.61	74.59	6.1244	78.94	80.17	88.73
UCTransNet	84.89	77.57	5.6995	80.94	84.11	89.62
U-Mamba	84.18	76.47	5.8501	80.90	83.24	88.40
VM-UNet	81.02	72.74	7.0025	76.75	79.40	86.90
EMCAD	85.07	77.73	5.2472	81.58	84.23	89.42
MCADS	84.51	76.92	5.5595	81.16	83.27	89.09
Ours	87.54±0.3	80.96	4.4057	85.27	86.23	91.11

Acknowledgement

We gratefully acknowledge the prior contributions of the research community, which have provided the foundation for our framework.