In this seminar, I would like to present my current research in including confidence measure in the spage2vec method. Spage2vec is an unsupervised segmentation‐free approach for decrypting the spatial transcriptomic heterogeneity of complex tissues at subcellular resolution. According to the paper 'Machine learning for cell classification and neighborhood analysis in glioma tissue' by L.Solorzano et al., it can be seen that some of the cells were really ’forced’ into becoming a cell type that they were probably not, resulting in false niches discovery. The low confidence cells were also in lower agreement with the manual annotations. Additionally, I will present a comparison to other methods for finding local neighbourhoods.