Single-cell RNA-seq (scRNA-seq) has provided researchers with an unprecedented opportunity to investigate the heterogeneities of cell populations. Recently popularized droplet-based technologies greatly increased the number of cells that can be profiled in each experiment. However, this type of data suffers from even higher technical noise and dropout rate, presenting new computational and analytical challenges. Here we try to identify and characterize a particular group of cells, tissue-resident memory T cells (TRM), in a breast cancer infiltrating T cell population sequenced with 10X Genomics droplet-based scRNA-seq platform. We performed a combined analysis including clustering, differential expression and trajectory analysis. Particularly we made vital use of imputation to overcome the high dropout rate of the data. A newly developed model, DECENT, was used to perform differential expression analysis on the imputed pre-dropout data. We identified 10 unique clusters with distinct gene expression profiles. One cluster showed enrichment for a list of TRM markers and the following differential expression analysis showed concordance with bulk RNA-seq experiments, suggesting that we were able to identify the TRM population unbiasedly with gene expression measurement. We also discovered a small cluster enriched in both TRM markers and mitotic genes, showing that TRM is undergoing active cell division and therefore potentially an active population in breast cancer immunosurveillance.