Electron Microscopy Images as Set of Fragments for Mitochondrial Segmentation
Authors: Naisong Luo, Rui Sun, Yuwen Pan, Tianzhu Zhang, Feng Wu
AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experimental results on the challenging Mito EM, Lucchi, and AC3/AC4 benchmarks demonstrate the effectiveness of the proposed method. We conduct extensive experiments on three 3D EM image benchmarks: Mito EM (Wei et al. 2020), Lucchi (Lucchi et al. 2011) and AC3/AC4 (Arganda-Carreras, Turaga, and Berger 2015). |
| Researcher Affiliation | Academia | Naisong Luo1*, Rui Sun1*, Yuwen Pan1*, Tianzhu Zhang1, 2 , Feng Wu1, 2 1Deep Space Exploration Laboratory/School of Information Science and Technology, University of Science and Technology of China 2Institute of Artificial Intelligence, Hefei Comprehensive National Science Center {lns6, issunrui, panyw}@mail.ustc.edu.cn, {tzzhang, fengwu}@ustc.edu.cn |
| Pseudocode | No | The paper provides detailed descriptions of the model architecture and processes through text and figures (e.g., Figure 2: 'The framework of our proposed Frag Vi T'), but it does not include a specific pseudocode or algorithm block labeled as such. |
| Open Source Code | No | The paper does not contain any explicit statement about making its source code publicly available, nor does it provide a link to a code repository for the methodology described. |
| Open Datasets | Yes | To demonstrate the effectiveness of our proposed model, we conduct extensive experiments on three 3D EM image benchmarks: Mito EM (Wei et al. 2020), Lucchi (Lucchi et al. 2011) and AC3/AC4 (Arganda-Carreras, Turaga, and Berger 2015). |
| Dataset Splits | Yes | Mito EM is a mitochondria instance dataset. It is divided into two subsets, Mito EM-R (rat tissue) and Mito EM-H (human tissue), whose resolution is anisotropic 30 8 8 nm. The size of the training set is 400 4096 4096 and the validation set is 100 4096 4096 voxels. Following the SNEMI3D challenge, we use the top 80 slices of AC4 as the training set and the rest of AC4 as the validation set. The top 100 slices of AC3 are testing set. |
| Hardware Specification | Yes | During training, our model is trained with a batch size of 2, using the Adam optimizer with an initial learning rate of 0.0001 on two NVIDIA RTX 3090 GPUs for 160,000 iterations. |
| Software Dependencies | No | The paper mentions the use of the Adam optimizer, but it does not specify any particular software libraries, frameworks (e.g., PyTorch, TensorFlow), or their corresponding version numbers required to reproduce the experiment. |
| Experiment Setup | Yes | In the fragment encoder, the number of layers is {2, 2, 2, 2} in each stage. The size of the input volume is anisotropic (18, 160, 160). We set λ1 = 1 and λ2 = 0.5. During training, our model is trained with a batch size of 2, using the Adam optimizer with an initial learning rate of 0.0001 on two NVIDIA RTX 3090 GPUs for 160,000 iterations. |