The disclosure relates to measurement and classification of tissue structures in samples using a combination of light imaging and spectroscopy, in particular although not necessarily exclusively for detection of tumors such as basal cell carcinoma or breast tumors in tissue samples. Embodiments disclosed include a method of automatically identifying tissue structures in a sample, the method comprising the steps of: measuring (1702, 1703) a response of an area of the sample to illumination with light; identifying (1704) regions within the area having a measured response within a predetermined range; determining (1705) locations within the identified regions; performing (1706) spectroscopic analysis of the sample at the determined locations; and identifying (1707) a tissue structure for each region from the spectroscopic analysis performed on one or more locations therein.