HEPS中文
MOCUPY is a CT reconstruction software based CUDA for GPU-accelerated, which written by Python and developed by the team of HEPS TXM beamline. MOCUPY contains modules such as Image import, Preprocessing, Shift align, Phase retrieval, Reconstruction and data analysis modules, which can realize the fast reconstruction of offline data. So as to meet the data processing requirements of synchrotron radiation users for "massive" X-ray CT.
- l The core algorithms achieve 100% GPU acceleration. Based on multi-threading to achieve fast reading of data and data processing with IO, CPU and GPU as the same time.
- l Rotation axis correction automatically for multi-scale and ring artifacts removal based on wavelet transform.
- l Provide manual and auto alignment modes for nano-CT images jitter correct, and the image correction accuracy can reach sub-pixel.
- l Realize the phase retrieval of propagation-based X-ray phase-contrast imaging in single ditance and multiple ditances acquisition modes. The main algorithms include tie, ctf, tiehom_Rytov, tiehom_Paganin, hioer, etc. It takes only 3 minutes for phase retrieval of 1E22 projections.
- l A variety of reconstruction algorithms such as FBP, Grid and EM are provided. It only takes several seconds for tomographic datasets with image size of 2k*2k to reconstruct based on grid algorithm.
Figure 1 Image jitter correction：reconstructed slice of nano-CT without and with alignment.
Figure 2 Phase retrieval: direct reconstruction slice and slice after phase retrieval in single distance mode of bone sample
The data is provided by the group of Professor Biao Deng in SSRF.
SYSTEM REQUIREMENTS
The algorithm of MOCUPY is based on CUDA Libraries，which requires Turing/Ampere cores or newer NVIDIA graphics/computing cards(https://www.techpowerup.com/gpu-specs/?mfgr=NVIDIA&sort=name).
In addition, in order to achieve higher reconstruction speed, please use the NVMe SSD (such as WD:SN770/SN570、KBG40ZNS) to read and write data.
The recommended computer graphics card performance (or higher) is as follows：
Figure 3 Recommended computer performance
Software Update Log
？ V2.0 —— July, 2025（2nd update）
Core optimization: Upgraded the libtiff library for faster read/write performance.
New functionality & UI: Modules added and the interface refined.
？ V1.0 —— April 28, 2023（1st update）
New Functionality:
---Automatic and manual jitter correction for nano-CT.
---Phase-retrieval module for propagation-based X-ray phase-contrast imaging.
UI Improvement: Dialog now warns when available memory is insufficient.
？ V0 —— September 28, 2022（Original release）
System: All modules tested and integrated to realize the CT reconstruction function.
TEAM MEMBER
Kai Zhang, Yan Wang, Jin Zhang, Chenpeng Zhou, Qingxi Yuan
CONTACT US
Jin Zhang，zhangjin2016@ihep.ac.cn
DOWNLOAD
SOFTWARE(.rar 804MB) | User Guide(PDF)
TestDATA01_TY(.rar 7.86GB) | TestDATA02_bone1_SDPhase(.rar 2.64GB)
Test data description: TestData01 is provided by the group of professor Gang Li in BSRF, which is for CT reconstruction test. TestData02 is provided by the group of Professor Biao Deng in SSRF, which is a test data for phase retrieval of propagation-based X-ray phase-contrast imaging in single distance.