A Novel Approach for Sperm Morphology Classification Using Generative Adversarial Network and Attention Network
Traitement du Signal · 30.04.2026 · 43 (2) · 839-850
DOI: 10.18280/ts.430223 ↗Kurz gefasst
Fachartikel in der Zeitschrift Traitement du Signal, online erschienen am 30.04.2026. Verfasst von Chellammal Palanichamy Thamil Selvi und 3 weiteren Autor:innen.
Abstract (Originalsprache)
Sperm morphology classification plays a major role in fertility assessments. The accurate identification of different sperm morphology classes is needed for clinical diagnosis and treatment planning. In this work, an innovative approach is proposed for sperm classification. For data augmentation, a Modified Generative Adversarial Network (M-GAN) is proposed. This network generates high-quality sperm images using a morphology-based conditioning mechanism. To extract useful features, the Biomorphological Attention Feature Network (BMAF-Net) is introduced. BMAF-Net uses attention mechanisms and learnable Gabor filters to extract both morphological and texture features. To further enhance performance, a Modified Waterwheel Plant Algorithm (M-WWPA) is used for feature optimisation. For classification, a Gradient Boosted Random Forest (GBRF) is proposed. GBRF combines the robustness of Random Forest (RF) with the boosting mechanism of Gradient Boosting to reduce classification errors. The proposed model is validated on three different datasets of HusHem, EVISAN, and SCIAN-SpermSegGS. Also, the explainable artificial intelligence technique is used for better model interpretation. Results show that the model achieves an overall accuracy of 96%.
Zitieren
APA
Selvi, C. P. T., Subramaniyan, S., Ramakrishnan, N., & Sheela, K. S. S. (2026). A Novel Approach for Sperm Morphology Classification Using Generative Adversarial Network and Attention Network. Traitement du Signal, 43(2), 839-850. https://doi.org/10.18280/ts.430223
Vancouver
Selvi CPT, Subramaniyan S, Ramakrishnan N, Sheela KSS. A Novel Approach for Sperm Morphology Classification Using Generative Adversarial Network and Attention Network. Traitement du Signal. 2026;43(2):839-850. doi:10.18280/ts.430223
BibTeX
@article{Selvi2026,
author = {Selvi, Chellammal Palanichamy Thamil and Subramaniyan, Saravanan and Ramakrishnan, Narmadha and Sheela, Krishna Samy Selva},
title = {A Novel Approach for Sperm Morphology Classification Using Generative Adversarial Network and Attention Network},
journal = {Traitement du Signal},
year = {2026},
volume = {43},
number = {2},
pages = {839--850},
doi = {10.18280/ts.430223}
}
EUVIMED