Data sample for "Spatial-Spectral Deep Learning for Prostate Cancer Tissue Classification in Infrared Spectroscopy"

Sample data corresponding to Figure 6, Figure 8, and Supporting Figure S5 of the paper "Spatial-Spectral Deep Learning for Prostate Cancer Tissue Classification in Infrared Spectroscopy" in Analytical Chemistry: https://doi.org/10.1021/acs.analchem.5c04765 Included are four FTIR TMA cores of prostate tissue: s10_c048, s17_c015, s33_c059, s33_c070. Each core is provided with preprocessed spectra (no suffix), or as raw spectra (_raw suffix). Each core has an associated annotation file, tissue mask file, chemical image file generated as the area under the amide I band, and an aligned H&E image from a serial section. Wavenumbers corresponding to channels in the hyperspectral images are uploaded as vectors. Hyperspectral images are in hdf5 format, and can be loaded using e.g. h5py for Python. Included also are model weights corresponding to each of the ten classifiers used in the paper. Implementations of these classifiers as well as usage examples can be found at https://github.com/lyyraaa/spatial-spectral-prostate. Please direct any questions about the data or paper to: [e-mail] UPDATE 2026/03/05: The full dataset of 1045 unprocessed cores, including the annotated subset used in the paper, is now available at https://zenodo.org/records/18671069.