Efficient identification and recycling of textiles containing Lycra is crucial due to its adverse environmental impacts, notably marine pollution. This study evaluates the use of Attenuated Total Reflection Fourier Transform Infrared (ATR-FTIR) spectroscopy combined with thermal pre-treatment and deep learning-enhanced microscopic imaging for detecting Lycra in blended textiles. Initial ATR-FTIR spectroscopy measurements on untreated fabrics revealed limitations due to shallow penetration depth and spectral overlap. Thermal pre-treatment significantly improved detection in cotton and polyamide blends but remained ineffective for polyester blends due to overlapping spectral bands. Microscopic images analyzed through a U-Net deep learning architecture provided effective segmentation and identification of Lycra fibers, overcoming spectroscopic limitations. These findings highlight the complementary role of deep learning-based microscopic imaging in textile recycling strategies, enhancing the feasibility of identifying Lycra-containing materials for sustainable waste management.