Raman spectroscopy has shown great promise as a method to discriminate between cancerous and normal tissue/cells for a range of oncology applications using microscopy and tissue interrogation instruments such as handheld probes and needles. Here we are presenting preliminary steps toward the development of a practical handheld macroscopic Raman spectroscopy instrument, demonstrating its capabilities to discriminate between different biological tissue types during ex vivo porcine experiments. The novel probe design can image a field of view of 25 mm2 with a spatial resolution <100 μm and an average spectral resolution of 95 cm-1, covering the fingerprint region between 450 to 1750 cm-1. The ability of the system to produce tissue maps based on molecular characteristics is demonstrated using a neural network machine learning technique.