LoDoInd: A Benchmark Low-dose Industrial CT Dataset - 3 of 3

Summary

This dataset accompanies the paper "LoDoInd: Introducing A Benchmark Low-dose Industrial CT Dataset and Enhancing Denoising with 2.5D Deep Learning Techniques". We are releasing the dataset with five different dose levels, including a reference set. All datasets are pre-registered, making them immediately suitable for deep learning applications in industrial CT.

Description

The uploaded content includes reconstructed images for noise levels 5 and reference. Each level comprises 4000 slices, with each slice being 1250x1250 pixels. Due to the 50 GB space limitation per submission on Zenodo, the rest of the dataset is available through separate links listed below:

Noise1 and Noise2 https://zenodo.org/records/10356955

Noise3 and Noise4 https://zenodo.org/records/10391277

Noise5 and Reference (this one) https://zenodo.org/records/10391412

The scanning parameters for all noise levels are summarized in the table below:

Averaged Projs
Exposure Time/ms
Scan Time/min
Voltage/kV
Current/uA

Reference
6
333
59.3
140
180

Noise Level 1
1
333
9.9
140
180

Noise Level 2
1
333
9.9
140
90

Noise Level 3
1
333
9.9
140
45

Noise Level 4
1
333
9.9
140
23

Noise Level 5
1
333
9.9
140
12

Additional link

The code for supervised learning-based denoising is available code .

Acknowledgment

This research was co-financed by the European Union H2020-MSCA-ITN-2020 under grant agreement no. 956172 (xCTing).