SpeHeaTal: A Cluster-Enhanced Segmentation Method for Sperm Morphology Analysis
Authors: Yi Shi, Yun-Kai Wang, Xu-Peng Tian, Tie-Yi Zhang, Bing Yao, Hui Wang, Yong Shao, Cen-Cen Wang, Rong Zeng
AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiment We evaluate the effectiveness of our proposed method, SPEHEATAL, on a sperm dataset obtained from clinical practice. The visual results demonstrate that SPEHEATAL provides superior segmentation of overlapping sperm tails compared to existing methods. Additional experimental settings and results are provided in the supplementary materials. |
| Researcher Affiliation | Academia | 1 School of Artificial Intelligence, Nanjing University, Nanjing, China 2 State Key Laboratory for Novel Software Technology, Nanjing University, Nanjing, China 3 Jinling Hospital, Affiliated Hospital of Medical School, Nanjing University, Nanjing, China 4 Jiangsu Provincial Medical Key Discipline Cultivation Unit, Nanjing, China |
| Pseudocode | No | The paper describes the CON2DIS and SPEHEATAL methodologies using textual descriptions and mathematical formulas (e.g., Eq. 1-12) but does not include any explicitly labeled pseudocode or algorithm blocks. |
| Open Source Code | Yes | Codes are available at https://www.github.com/shiy19/SpeHeaTal. |
| Open Datasets | No | Given that existing sperm datasets do not adequately capture the characteristics of sperm overlap and dye impurities observed in clinical practice, we collaborated with several prominent hospitals to compile a dataset comprising approximately 2,000 unlabeled sperm images for model calibration, along with an additional 240 expert-annotated images for rigorous model evaluation, as depicted in Figure 1(b). |
| Dataset Splits | Yes | For those models requiring labeled samples for training, we divided the 240 labeled images into a 3:1 ratio, with 180 images used for their training and the remaining 60 images reserved for performance evaluation. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used to run the experiments, such as GPU models, CPU types, or memory specifications. |
| Software Dependencies | No | Our method utilizes the SAM model in everything mode to segment microscopic sperm images. The paper mentions the use of the SAM model but does not provide its specific version number or any other software dependencies with version details. |
| Experiment Setup | Yes | According to the parameter settings described in the methodology section, the shape filtering threshold α = 0.25, color filtering threshold β = 0.4, and point allocation threshold γ = 5 pixels. The distance threshold λ1 = 30 pixels and angle threshold λ2 = 35 are determined to effectively splice different parts of each sperm. |