Finding trees and crops that are resistant to pathogens is key in preparing for current and future disease threats. In this dissertation, we analyzed the potential of using hyperspectral imaging (HSI) to find infection while identifying issues and strategies addressing differences over time, in spatial resolution, the application of machine learning and data science, and plant susceptibility to diseases. Similar problems arise in many other applications in ecology and remote sensing. First, we monitored infection with white pine blister rust in seedlings of southwestern white pine from different seed-source families. A support vector machine was able to automatically detect infection with a classification accuracy of 87% over 16 image collection dates. HSI only missed 4% of infected seedlings that were impacted in terms of vigor according to expert’s assessments. Classification accuracy per family was highly correlated with plant susceptibility. The most informative hyperspectral features were identified using a simple and fast ‘new search algorithm’ that combines both the p-value of a 2-sample t-test and the Bhattacharyya distance. This method could be used for future developments of less expensive and more data-parsimonious multispectral cameras. Simultaneously, we assessed model selection and the impact of the timing of data acquisition, showing that using models retrospectively is less accurate than is their use on data collected after the calibration data set. We assessed two metrics of performance for selection of classification model: the predicted accuracy (PA); and the area under the receiver operating curve (AUC), both from a 10-fold cross-validation. AUC was more successful at identifying suitable classifiers to predict infection on other dates. Moreover, heterogeneous ensembles using either AUC or PA to select multiple models was even more accurate, with a 1.8 percentage point increase for AUC and 3 for PA. Classification accuracy was affected by the size of the time gap between training and testing dates as well as the choice of training and test date. Next, we used narrowband-hyperspectral data to select wavelength regions that can be exploited to identify wheat infected with soilborne mosaic virus. A model trained on the full spectrum of leaf scans resulted in nearly 70% accuracy, while a subset of broad band wavelengths (selected from the full spectrum) achieved an accuracy of 71.4%. The upscaling from leaf to field scans was successful at finding infection in the field using classifiers trained on the entire spectrum of the hyperspectral data acquired in a lab setting.