DuSSS: Dual Semantic Similarity-Supervised Vision-Language Model for Semi-Supervised Medical Image Segmentation
Authors: Qingtao Pan, Wenhao Qiao, Jingjiao Lou, Bing Ji, Shuo Li
AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental results demonstrate that our Du SSS achieves outstanding performance with Dice of 82.52%, 74.61% and 78.03% on three public datasets (Qa Ta-COV19, BM-Seg and Mo Nu Seg). |
| Researcher Affiliation | Academia | 1School of Control Science and Engineering, Shandong University, Jinan, China 2Key Laboratory of Machine Intelligence and System Control, Ministry of Education, China 3Department of Computer and Data Science and Department of Biomedical Engineering, Case Western Reserve University, USA EMAIL, EMAIL |
| Pseudocode | No | The paper describes the methodology using textual explanations and mathematical formulas, but it does not contain any structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | Code https://github.com/Qingtao Pan/Du SSS/ |
| Open Datasets | Yes | Extensive experiments are conducted on three public medical image segmentation datasets: chest infection area segmentation, bone metastases segmentation, and nuclei instances segmentation. Qa Ta-COV19 (Degerli et al. 2022) ... BM-Seg (Afnouch et al. 2023) ... Mo Nu Seg (Kumar et al. 2017) |
| Dataset Splits | Yes | Qa Ta-COV19 (Degerli et al. 2022) contains 9258 COVID-19 chest X-ray radiographs. (Li et al. 2024b) provided text annotations and split 7145 samples for training and 2113 samples for testing. BM-Seg (Afnouch et al. 2023) consists of 23 CT-scans from 23 patients, totaling 1517 slices from different skeletal views. 270 slices from a single view are selected, allocating 200 for training and 70 for testing. Mo Nu Seg (Kumar et al. 2017) includes 44 images, and the image size is 1000 1000. The training dataset contains 30 images and the testing dataset contains 14 images. |
| Hardware Specification | Yes | Our method is implemented using Pytorch. The operating system is Ubuntu 20.04.4 LTS with 24GB V100 GPU. |
| Software Dependencies | No | The paper states 'Our method is implemented using Pytorch.' but does not provide specific version numbers for Pytorch or any other software dependencies. |
| Experiment Setup | Yes | The learning rate is set to 3e-4 for both the Qa Ta-COV19 and BM-Seg datasets, and 1e-3 for the Mo Nu Seg dataset. Early stopping is implemented if the model s performance does not improve after 20 epochs based on its current performance. The batch size is 32 for the Qa Ta-COV19 dataset, 16 for the BM-Seg dataset, and 4 for the Mo Nu Seg dataset. |