ArticleOpen Access http://dx.doi.org/10.26855/acc.2025.12.003
Non-destructive Pre-incubation Egg Fertilization Detection Using Low-field MRI and Efficientnetv2
Tanyu Lin1, Kaiwen Huang2, Shuyu Cheng3, Zhuoheng Tang4, Zhoucai Ou1, Jiye Zeng5,*, Yuanyang Mao6,*
1Guangzhou Yiguang Technology Co., Ltd., Guangzhou 510000, Guangdong, China.
2College of Engineering, South China Agricultural University, Guangzhou 510642, Guangdong, China.
3Southern Medical University, Guangzhou 510515, Guangdong, China.
4South China Agricultural University, Guangzhou 510642, Guangdong, China.
5Central University for Nationalities, Industrial Development Promotion Center of the Ministry of Industry and Information Technology, Beijing 100081, China.
6The Pearl River Hospital of Southern Medical University, School of Mathematics and Information of South China Agricultural University, South China Tropical Smart Agricultural Technology Key Laboratory of the Ministry of Agriculture and Rural Affairs, Guangzhou 510000, Guangdong, China.
*Corresponding author: Jiye Zeng, Yuanyang Mao
Published: December 12,2025
Abstract
To address the challenge of non-destructive assessment of pre-incubation egg fertilization status, this study proposes an early screening method for infertile eggs combining low-field MRI with deep learning. A rapid spin echo (FSE) acquisition protocol was developed on a 0.3 T MRI system to obtain 3,200 pre-incubation egg images. The pre-trained EfficientNetV2 backbone network was enhanced with Spatial Enhancement (SE) attention and Fused MBConv modules, employing progressive input scaling and regularization scheduling for end-to-end training. GradCAM validation confirmed that the model’s discrimination of key regions like the blastodisc and yolk aligns with MRI contrast mechanisms. Experimental results demonstrated 98.37% accuracy, 98.32% F1 score, and approximately 14 FPS inference speed. Thermal maps revealed the model’s focus on high-signal annular regions around the blastodisc, consistent with pre-incubation morphology. Compared to light imaging, ultrasound, and spectral methods, low-field MRI shows stable imaging of eggshell color/thickness and internal microstructures. This study establishes a practical new paradigm for non-destructive fertilization assessment of hatching eggs, while providing insights for addressing industry challenges such as early sex determination in the hatching sector.
Keywords
Magnetic resonance imaging; low-field MRI; EfficientNetV2; SE attention; Fused-MBConv; Grad-CAM
References
[1] Zhou Z, Chen Z, Shan J, Lin J, Li X. Monitoring brain development of chick embryos in vivo using 3.0 T MRI: subdivision volume change and preliminary structural quantification using DTI. BMC Dev Biol. 2015;15:29.
[2] Liu CK, Liu Y, Wang QH, et al. Advances in technologies and equipment for detecting fertilization and sex determination in hatching eggs without loss of information. Nongye Jixie Xuebao/Journal of Agricultural Machinery. 2024;55(5):1-20.
[3] Marques JP, Simonis FFJ, Webb AG. Low-field MRI: An MR physics perspective. J Magn Reson Imaging. 2019;49(6):1528-42.
[4] Wald LL, McDaniel PC, Witzel T, Stockmann JP, Cooley CZ. Low-cost and portable MRI. J Magn Reson Imaging. 2020;52(3):686-96.
[5] Sarracanie M, LaPierre CD, Salameh N, Waddington DEJ, Witzel T, Rosen MS. Low-cost high-performance MRI. Sci Rep. 2015;5:15177.
[6] Litjens G, Kooi T, Bejnordi BE, Setio AAA, Sánchez CI. A survey on deep learning in medical image analysis. Med Image Anal. 2017;42:60-88.
[7] Tan M, Le Q. EfficientNetV2: Smaller Models and Faster Training. In: Proceedings of the 38th International Conference on Machine Learning (ICML); 2021. p. 10096-106.
[8] Hu J, Shen L, Sun G, Albanie S. Squeeze-and-excitation networks. IEEE Trans Pattern Anal Mach Intell. 2017;99.
[9] Loshchilov I, Hutter F. Decoupled weight decay regularization. arXiv 2017.
[10] Szegedy C, Vanhoucke V, Ioffe S, Shlens J, Wojna Z. Rethinking the Inception Architecture for Computer Vision. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR); 2016. p. 2818-26.
[11] Selvaraju RR, Cogswell M, Das A, Vedantam R, Parikh D, Batra D. Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization. Int J Comput Vis. 2020;128:336-59.
[12] Haacke EMB, Brown RW, Thompson MR, Venkatesan RV. Magnetic Resonance Imaging: Physical Principles and Sequence Design. 2nd ed. Hoboken, NJ: Wiley-Blackwell; 2014.
How to cite this paper
Non-destructive Pre-incubation Egg Fertilization Detection Using Low-field MRI and Efficientnetv2
How to cite this paper: Tanyu Lin, Kaiwen Huang, Shuyu Cheng, Zhuoheng Tang, Zhoucai Ou, Jiye Zeng, Yuanyang Mao. (2025) Non-destructive Pre-incubation Egg Fertilization Detection Using Low-field MRI and Efficientnetv2. Advances in Computer and Communication, 6(5), 274-279.
DOI: http://dx.doi.org/10.26855/acc.2025.12.003