Latent tuberculosis infection (LTBI) is a precursor to active tuberculosis, a leading cause of death globally. The century-old tuberculin skin test (TST) and the recently recommended Mycobacterium tuberculosis(Mtb) antigen-based skin tests (TBST) are low-cost methods for screening for LTBI. The Mantoux method of reading these tests rely on tactile cues by clinicians to read the size of the induration formed after the skin tests. This leads to subjectivity in the interpretation of the readings as the boundaries of the induration are typically subdermal. Hyperspectral imaging (HSI) is an emerging modality for management of skin conditions like skin cancer and has potential in LTBI diagnosis. This chapter introduces a novel application of HSI for the segmentation and visualization of the subdermal induration boundaries to address the subjectivity of the Mantoux method of reading indurations. The segmentation implemented in this study is based on principal component analysis (PCA) features generated from the hyperspectral images of 20 human subjects. The features were used to develop a machine learning classification model. The classification results showed a cross validation mean accuracy of 86.67% and predictive accuracy of 80%. Thus, this study demonstrates the ability of HSI, coupled with PCA, to optically capture subdermal induration information that could be useful for LTBI detection.