Journals / Erciyes Üniversitesi Fen Bilimleri Enstitüsü Fen Bilimleri Dergisi / 2026 / Cilt: 42 Sayı: 3

Determining Phenotypic Relationships among Flax (Linum usitatissimum L.) Genotypes Using Machine Learning and Deep Learning Guided by Phenotypic Core-Based Binary Seed Image Classification

Fenotipik Çekirdek Rehberli İkili Tohum Görüntüsü Sınıflandırması ile Makine Öğrenmesi ve Derin Öğrenme Kullanılarak Keten (Linum usitatissimum L.) Genotipleri Arasındaki Fenotipik İlişkilerin Belirlenmesi

Sevim Akçura(ÇANAKKALE ONSEKİZ MART ÜNİVERSİTESİ)0000-0002-7699-748X
Pages
—
Publication Date
2026-10-27
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

DenseNet121 achieved the highest performance in six of the ten binary comparisons, while MobileNetV2 produced the highest overall F1 score (0.951).

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Background/Aim: Accurate identification of flax (Linum usitatissimum L.) genotypes is essential for germplasm management, breeding programmes and seed certification.
📊
Key Findings & Data:Although image-based phenotyping and deep learning have shown considerable potential for genotype classification, the relationships between phenotypic similarity, within-genotype variation and classification performance remain poorly understood.
💡
Conclusion & Impact:DenseNet121 achieved the highest performance in six of the ten binary comparisons, while MobileNetV2 produced the highest overall F1 score (0.951).
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Dijital fenotipleme · fenotipik çekirdek · derin özellik uzayı

Abstract

Background/Aim: Accurate identification of flax (Linum usitatissimum L.) genotypes is essential for germplasm management, breeding programmes and seed certification. Although image-based phenotyping and deep learning have shown considerable potential for genotype classification, the relationships between phenotypic similarity, within-genotype variation and classification performance remain poorly understood. This study aimed to investigate binary genotype classification using phenotypic core seed sets and to examine the relationships among deep feature space, phenotypic distance and classification performance. Methods: Single-seed images representing five flax genotypes were used to construct phenotypic core datasets by selecting representative individuals from each genotype. Binary classification was performed using four convolutional neural network architectures. Deep feature space extracted by DenseNet121 was further used to quantify phenotypic distances, evaluate within-genotype variation, construct phenotypic similarity networks and visualise genotype relationships using hierarchical clustering and UMAP. Results: Significant differences were detected among genotypes in morphological, colour and texture characteristics (P < 0.001). Binary classification performance varied among genotype pairs, with F1 scores ranging from 0.706 to 0.951. DenseNet121 achieved the highest performance in six of the ten binary comparisons, while MobileNetV2 produced the highest overall F1 score (0.951). Phenotypic distances calculated in the deep feature space showed a significant positive correlation with CNN classification performance (Pearson r = 0.710, P = 0.021), whereas no significant relationship was observed for conventional machine learning models. Phenotypic similarity networks, hierarchical clustering and UMAP projections consistently visualised phenotypic relationships among genotypes and highlighted differences in within-genotype homogeneity. Conclusions: The findings demonstrate that image-based deep phenotyping can provide biologically interpretable information beyond classification accuracy. Integrating phenotypic core selection with deep feature space analysis offers a complementary framework for evaluating genotype relationships, within-genotype variation and representative germplasm, supporting future applications in flax breeding and digital phenotyping.

Özet

Amaç: Keten (Linum usitatissimum L.) genotiplerinin doğru şekilde tanımlanması; genetik kaynakların etkin yönetimi, ıslah programlarının yürütülmesi ve tohum sertifikasyonu açısından büyük önem taşımaktadır. Görüntü tabanlı fenotipleme ve derin öğrenme yöntemleri genotip sınıflandırmasında başarılı sonuçlar vermesine rağmen, fenotipik benzerlik, genotip içi varyasyon ve sınıflandırma başarısı arasındaki ilişkiler henüz yeterince açıklığa kavuşturulamamıştır. Bu çalışmada, fenotipik çekirdek tohum veri setleri kullanılarak ikili genotip sınıflandırması gerçekleştirilmiş ve derin özellik uzayı, fenotipik uzaklık ile sınıflandırma performansı arasındaki ilişkilerin ortaya konulması amaçlanmıştır. Yöntemler: Beş farklı keten genotipine ait tek tohum görüntüleri kullanılarak, her genotipi en iyi temsil eden bireylerin seçilmesiyle fenotipik çekirdek veri setleri oluşturulmuştur. İkili genotip sınıflandırmaları dört farklı evrişimsel sinir ağı mimarisi kullanılarak gerçekleştirilmiştir. Ayrıca, DenseNet121 modelinden elde edilen derin özellik uzayı kullanılarak genotipler arasındaki fenotipik uzaklıklar hesaplanmış, genotip içi varyasyon değerlendirilmiş, fenotipik benzerlik ağları oluşturulmuş ve genotipler arasındaki ilişkiler hiyerarşik kümeleme ile UMAP yöntemleri kullanılarak görselleştirilmiştir. Bulgular: Genotipler arasında morfolojik, renk ve doku özellikleri bakımından istatistiksel olarak anlamlı farklılıklar belirlenmiştir (P < 0,001). İkili sınıflandırma performansı genotip çiftlerine göre değişmiş ve F1 skorları 0,706 ile 0,951 arasında bulunmuştur. DenseNet121 modeli on ikili karşılaştırmanın altısında en yüksek performansı gösterirken, en yüksek genel F1 skoru (0,951) MobileNetV2 modeli ile elde edilmiştir. Derin özellik uzayında hesaplanan fenotipik uzaklıklar ile evrişimsel sinir ağı modellerinin sınıflandırma performansı arasında anlamlı ve pozitif bir ilişki saptanmıştır (Pearson r = 0,710; P = 0,021). Buna karşılık, geleneksel makine öğrenmesi modellerinde benzer bir ilişki belirlenmemiştir. Fenotipik benzerlik ağları, hiyerarşik kümeleme ve UMAP analizleri, genotipler arasındaki genotipler arasındaki fenotipik ilişkileri tutarlı biçimde görselleştirmiş ve genotip içi homojenlik düzeylerindeki farklılıkları açıkça göstermiştir. Sonuç: Araştırma bulguları, görüntü tabanlı derin fenotipleme yaklaşımının yalnızca yüksek sınıflandırma doğruluğu sağlamakla kalmayıp, aynı zamanda genotipler arasındaki biyolojik ilişkilerin yorumlanmasına da önemli katkılar sunduğunu göstermektedir. Fenotipik çekirdek seçiminin derin özellik uzayı analizi ile birlikte kullanılması; genotipler arasındaki fenotipik ilişkilerin, genotip içi varyasyonun ve temsil edici genetik materyalin değerlendirilmesine yönelik tamamlayıcı ve güçlü bir yaklaşım sunmaktadır. Bu yaklaşımın keten ıslahı, genetik kaynakların karakterizasyonu ve dijital fenotipleme çalışmalarında gelecekte önemli uygulama alanları bulabileceği değerlendirilmektedir.

🔖

Cite This Article

Akçura, S. (2026). Determining Phenotypic Relationships among Flax (Linum usitatissimum L.) Genotypes Using Machine Learning and Deep Learning Guided by Phenotypic Core-Based Binary Seed Image Classification. Erciyes Üniversitesi Fen Bilimleri Enstitüsü Fen Bilimleri Dergisi, Cilt: 42 Sayı: 3. https://doi.org/10.65520/erciyesfen.1980963

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (30)

  1. Diederichsen, A., & Raney, J. P. (2006). Seed colour, seed weight and seed oil content in Linum usitatissimum accessions held by Plant Gene Resources of Canada. Plant Breeding, 125(4), 372–377. https://doi.org/10.1111/j.1439-0523.2006.01231.x
  2. Walkowiak, M., Spasibionek, S., & Krótka, K. (2022). Variation and genetic analysis of fatty acid composition in flax (Linum usitatissimum L.). Euphytica, 218, Article 2. https://doi.org/10.1007/s10681-021-02941-6
  3. De Beukelaer, H., Davenport, G. F., & Fack, V. (2018). Core Hunter 3: Flexible core subset selection. BMC Bioinformatics, 19, 203. https://doi.org/10.1186/s12859-018-2209-z
  4. Franco-Duran, J., Crossa, J., Chen, J., & Hearne, S. J. (2019). The impact of sample selection strategies on genetic diversity and representativeness in germplasm bank collections. BMC Plant Biology, 19, 520. https://doi.org/10.1186/s12870-019-2142-y
  5. Deng, Q., Zhao, S., Guan, C., Fu, G., Yu, S., Lin, S., Wang, Z., Fu, H., Lu, X., & Cheng, S. (2024). Construction and evaluation of pepper core collection based on phenotypic traits and SSR markers. Applied Sciences, 14(17), 7473. https://doi.org/10.3390/app14177473
  6. Haralick, R. M., Shanmugam, K., & Dinstein, I. (1973). Textural features for image classification. IEEE Transactions on Systems, Man, and Cybernetics, SMC-3(6), 610–621. https://doi.org/10.1109/TSMC.1973.4309314
  7. Becht, E., McInnes, L., Healy, J., Dutertre, C. A., Kwok, I. W. H., Ng, L. G., Ginhoux, F., & Newell, E. W. (2019). Dimensionality reduction for visualizing single-cell data using UMAP. Nature Biotechnology, 37, 38–44. https://doi.org/10.1038/nbt.4314
  8. Cortes, C., & Vapnik, V. (1995). Support-vector networks. Machine Learning, 20, 273–297. https://doi.org/10.1007/BF00994018
  9. Breiman, L. (2001). Random forests. Machine Learning, 45, 5–32. https://doi.org/10.1023/A:1010933404324
  10. Chen, T., & Guestrin, C. (2016). XGBoost: A scalable tree boosting system. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (pp. 785–794). https://doi.org/10.1145/2939672.2939785
  11. He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 770–778). https://doi.org/10.1109/CVPR.2016.90
  12. Huang, G., Liu, Z., van der Maaten, L., & Weinberger, K. Q. (2017). Densely connected convolutional networks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 4700–4708). https://doi.org/10.1109/CVPR.2017.243
  13. Sandler, M., Howard, A., Zhu, M., Zhmoginov, A., & Chen, L.-C. (2018). MobileNetV2: Inverted residuals and linear bottlenecks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 4510–4520). https://doi.org/10.1109/CVPR.2018.00474
  14. Tan, M., & Le, Q. V. (2019). EfficientNet: Rethinking model scaling for convolutional neural networks. In Proceedings of the 36th International Conference on Machine Learning (pp. 6105–6114). PMLR.
  15. Wiesnerová, D., & Wiesner, I. (2008). Computer image analysis of seed shape and seed color for flax cultivar description. Computers and Electronics in Agriculture, 61(2), 126–135. https://doi.org/10.1016/j.compag.2007.10.001
  16. Smykalova, I., Grillo, O., Bjelkova, M., Pavelek, M., & Venora, G. (2013). Phenotypic evaluation of flax seeds by image analysis. Industrial Crops and Products, 47, 232–238. https://doi.org/10.1016/j.indcrop.2013.03.001
  17. Kang, Q. H., Sun, Z. Y., Qiu, C. S., Wang, Y. F., Zhu, X., Song, X. X., Chen, X. Y., Yao, D. D., Jiang, W. D., Yuan, H. M., Tang, L. L., Liu, D. D., & Yao, Y. B. (2024). The study progress on breeding of flax and industrial hemp in China. Journal of Natural Fibers, 21(1), 2389159. https://doi.org/10.1080/15440478.2024.2389159
  18. Lu, J., Wu, H., Wang, F., Li, J., Wang, Y., Zhao, Q., Wang, Y., Wang, X., Lei, X., Sun, R., Zhang, J., Xiong, A., Deyholos, M. K., & Zhang, J. (2025). Telomere-to-telomere flax (Linum usitatissimum L.) genome assembly unlocks insights beyond fatty acid metabolism pathways. Horticulture Research, 12, uhaf127. https://doi.org/10.1093/hr/uhaf127
  19. Xu, P., Yang, R., Zeng, T., Zhang, J., Zhang, Y., & Tan, Q. (2021). Varietal classification of maize seeds using computer vision and machine learning techniques. Journal of Food Process Engineering, 44(11), e13846. https://doi.org/10.1111/jfpe.13846
  20. Taheri-Garavand, A., Nasiri, A., Fanourakis, D., Fatahi, S., Omid, M., & Nikoloudakis, N. (2021). Automated in situ seed variety identification via deep learning: A case study in chickpea. Plants, 10(7), 1406. https://doi.org/10.3390/plants10071406
  21. Yang, H., Ni, J., Gao, J., Han, Z., & Luan, T. (2021). A novel method for peanut variety identification and classification by improved VGG16. Scientific Reports, 11, 15756. https://doi.org/10.1038/s41598-021-95240-y
  22. Akbarpour, O., Akbari, S., Fanourakis, D., & Taheri-Garavand, A. (2026). Deep learning-based seed variety classification: A case study in maize. BMC Plant Biology, 26, 105. https://doi.org/10.1186/s12870-025-07919-3
  23. Wei, J., Dai, Z., Zhang, Q., Yang, L., Zeng, Z., Zhou, Y., Liu, J., & Chen, B. (2025). Seed multispectral imaging combined with machine learning algorithms for distinguishing different varieties of lettuce (Lactuca sativa L.). Food Chemistry: X, 27, 102399. https://doi.org/10.1016/j.fochx.2025.102399
  24. Yang, Y., Harrison, R. C., Zhang, D., Shen, B., Yan, Y., & Kang, D. (2024). Effect of genetic distances of different genotypes of maize on the authenticity of single seeds detected by NIR spectroscopy. Frontiers in Plant Science, 15, 1361328. https://doi.org/10.3389/fpls.2024.1361328
  25. Lim, S., Park, S., Baek, I., Botkin, J. R., Jang, J. H., Hong, S. M., Irish, B., Kim, M. S., Meinhardt, L., Curtin, S. J., & Ahn, E. (2025). Integrative analysis of seed morphology, geographic origin, and genetic structure in Medicago with implications for breeding and conservation. BMC Plant Biology, 25, 274. https://doi.org/10.1186/s12870-025-06304-4
  26. Lipiński, S. (2025). Statistical analysis of seeds morphology and texture for interspecific similarity assessment across taxonomic levels. Biuletyn Instytutu Hodowli i Aklimatyzacji Roślin, 303/304, 19–33. https://doi.org/10.37317/biul-2025-0006
  27. Wang, Y., Zhang, Y., Song, A., Guo, M., Zhang, C., Wang, C., Zhao, Y., Huang, G., Men, Q., Zhao, C., & Guo, X. (2025). Precision profiling of seed coat phenotypes in maize: 3D surface morphology, color, texture traits for the construction of phenotyping interaction networks. Current Plant Biology, 43, 100493. https://doi.org/10.1016/j.cpb.2025.100493
  28. Taghavi Namin, S., Esmaeilzadeh, M., Najafi, M., Brown, T. B., & Borevitz, J. O. (2018). Deep phenotyping: Deep learning for temporal phenotype/genotype classification. Plant Methods, 14, 66. https://doi.org/10.1186/s13007-018-0333-4
  29. Akçura, S. (2026). Image-based multimodal seed phenotyping using morphological, colorimetric, and textural traits for soybean genotype classification. Genetic Resources and Crop Evolution, 73(6), 232. https://doi.org/10.1007/s10722-026-02864-y
  30. Akçura, S. (2026). Explainable multimodal learning for genotype classification using image-based and biochemical features: A case study on black cumin (Nigella sativa L.). Food Analytical Methods, 19(10), 291. https://doi.org/10.1007/s12161-026-03202-9