Molecular Plant Breeding 2024, Vol.15, No.5, 233-246 http://genbreedpublisher.com/index.php/mpb 241 associated with traditional methods. The integration of these imaging techniques with high-throughput platforms enables the rapid screening of large populations, making it a valuable tool for modern plant breeding programs. 7.2 Use of hyperspectral and multispectral imaging to detect early disease symptoms Hyperspectral and multispectral imaging technologies have shown great potential in detecting early disease symptoms in wheat. These imaging techniques capture a wide range of wavelengths, providing detailed spectral information that can be used to identify subtle changes in plant physiology caused by pathogen infection. For example, hyperspectral imaging combined with machine learning has been used to monitor plant phenotypes under salt stress, demonstrating its ability to detect physiological and biochemical changes non-destructively (Feng et al., 2020). This approach can be adapted to detect early disease symptoms in wheat, allowing for timely and accurate disease management. A study comparing UAV-based RGB and multispectral imaging for phenotyping wheat stay-green traits found that multispectral indices, particularly those containing red-edge or near-infrared bands, were more effective in detecting early disease symptoms than RGB indices (Cao et al., 2021). This highlights the importance of selecting appropriate spectral bands for disease detection. The use of hyperspectral imaging to evaluate wheat chlorophyll content under drought stress further supports the utility of this technology in monitoring plant health and detecting early signs of disease (Yang et al., 2023). 7.3 Application of machine learning models for automated disease scoring and resistance evaluation The integration of machine learning models with high-throughput phenotyping platforms has significantly advanced the automated scoring of disease resistance in wheat. Machine learning algorithms, such as support vector machines (SVM) and deep learning, can analyze complex datasets generated by imaging technologies to identify disease symptoms and quantify resistance levels accurately. For instance, SVM classification was used to identify and quantify powdery mildew in barley, demonstrating the potential of machine learning in automating disease scoring (Kuska et al., 2018; Wu, 2024). In another study, deep learning approaches were applied to hyperspectral imaging data to segment plant and leaf regions accurately, enabling the precise measurement of physiological traits affected by salinity stress (Feng et al., 2020). This method can be extended to disease resistance screening, where machine learning models can be trained to recognize disease-specific spectral signatures and automate the evaluation process. Furthermore, the use of ensemble feature selection methods has been shown to improve the accuracy of hyperspectral data analysis by identifying the most informative spectral features for plant phenotyping (Moghimi et al., 2018). This approach can enhance the performance of machine learning models in disease resistance screening by reducing data dimensionality and focusing on relevant features. The combination of hyperspectral imaging and machine learning provides a powerful tool for high-throughput, non-destructive disease resistance evaluation in wheat, facilitating the rapid identification of resistant genotypes and accelerating breeding programs. The integration of novel image-based quantification approaches, hyperspectral and multispectral imaging, and machine learning models has transformed the landscape of wheat disease resistance screening. These high-throughput phenotyping methods offer unprecedented accuracy, speed, and efficiency, enabling the early detection of disease symptoms and the automated evaluation of resistance, ultimately contributing to the development of more resilient wheat varieties. 8 Future Directions 8.1 Potential advancements in HTP technology to improve accuracy and scalability in disease resistance screening High-throughput phenotyping (HTP) technology has made significant strides in recent years, but there is still considerable room for improvement, particularly in the context of disease resistance screening in wheat. One of the primary areas for advancement is the enhancement of sensor technologies. Current HTP platforms utilize a variety of sensors, including RGB cameras, hyperspectral sensors, and computed tomography, which provide
RkJQdWJsaXNoZXIy MjQ4ODYzMg==