Interpolation of Imaging\nMass Spectrometry Data by\na Window-Based Adversarial Autoencoder Method

Imaging mass spectrometry (IMS) is a technique for simultaneously\nacquiring the expression and distribution of molecules on the surface\nof a sample, and it plays a crucial role in spatial omics research.\nIn IMS, the time cost and instrument load required for large data\nsets must be considered, as IMS typically involves tens of thousands\nof pixels or more. In this study, we developed a high-resolution method\nfor IMS data reconstruction using a window-based Adversarial Autoencoder\n(AAE) method. We acquired IMS data from partial cerebellum regions\nof mice with a pitch size of 75 μm and then down-sampled the\ndata to a pitch size of 150 μm, selecting 22 <i>m</i>/<i>z</i> peak intensity values per pixel. We established\nan AAE model to generate three pixels from the surrounding nine pixels\nwithin a window to reconstruct the image data at a pitch size of 75\nμm. Compared with two alternative interpolation methods, Bilinear\nand Bicubic interpolation, our window-based AAE model demonstrated\nsuperior performance on image evaluation metrics for the validation\ndata sets. A similar model was constructed for larger mouse kidney\ntissues, where the AAE model achieved high image evaluation metrics.\nOur method is expected to be valuable for IMS measurements of large\nanimal organs across extensive areas.