A comparative analysis of six in situ gene expression profiling technologies was conducted using publicly available mouse brain datasets. The study highlights the challenges of comparing these technologies due to differences in panel composition and off-target molecular artifacts. Standard sensitivity metrics, such as the number of unique molecules detected per cell, are not directly comparable across datasets. The researchers explored various sources of molecular artifacts, developed novel metrics to control for them, and used these metrics to evaluate and compare different in situ technologies. They found that molecular false positives can seriously confound spatially-aware differential expression analysis, requiring caution in the interpretation of downstream results. The study compared six in situ technologies, including three commercial ones (Xenium, MERSCOPE, Molecular Cartography) and three academic ones (MERFISH, STARmap PLUS, EEL FISH). The analysis revealed that MERSCOPE exhibited the best performance, with the optimal trade-off between sensitivity and specificity. Xenium also performed well, but its segmentations were more lenient, leading to a higher number of molecules detected per cell at the expense of incorrect molecular assignments. The Molecular Cartography dataset showed impressive sensitivity on a per-gene basis, while the MERFISH dataset demonstrated strong performance with a large panel size. The EEL FISH dataset had the lowest sensitivity. The study also examined how spatial location drives heterogeneity within an individual cell type, which is particularly well-suited to imaging datasets. They found that many of the top differentially expressed genes in in situ datasets are highly expressed markers of other cell types and are not expressed in scRNA-seq astrocytes from either the cortex or thalamus. This highlights the danger of non-specific molecular signals not only for comparative benchmarking but also for downstream molecular analyses. The researchers emphasize the importance of considering both sensitivity and specificity when evaluating in situ technologies. They also highlight the need for improved segmentation methods and computational approaches to address non-specific biases in downstream analyses. The study provides a framework for assessing new technologies, datasets, computational methods, and tissue preparation techniques in the field of spatial transcriptomics.A comparative analysis of six in situ gene expression profiling technologies was conducted using publicly available mouse brain datasets. The study highlights the challenges of comparing these technologies due to differences in panel composition and off-target molecular artifacts. Standard sensitivity metrics, such as the number of unique molecules detected per cell, are not directly comparable across datasets. The researchers explored various sources of molecular artifacts, developed novel metrics to control for them, and used these metrics to evaluate and compare different in situ technologies. They found that molecular false positives can seriously confound spatially-aware differential expression analysis, requiring caution in the interpretation of downstream results. The study compared six in situ technologies, including three commercial ones (Xenium, MERSCOPE, Molecular Cartography) and three academic ones (MERFISH, STARmap PLUS, EEL FISH). The analysis revealed that MERSCOPE exhibited the best performance, with the optimal trade-off between sensitivity and specificity. Xenium also performed well, but its segmentations were more lenient, leading to a higher number of molecules detected per cell at the expense of incorrect molecular assignments. The Molecular Cartography dataset showed impressive sensitivity on a per-gene basis, while the MERFISH dataset demonstrated strong performance with a large panel size. The EEL FISH dataset had the lowest sensitivity. The study also examined how spatial location drives heterogeneity within an individual cell type, which is particularly well-suited to imaging datasets. They found that many of the top differentially expressed genes in in situ datasets are highly expressed markers of other cell types and are not expressed in scRNA-seq astrocytes from either the cortex or thalamus. This highlights the danger of non-specific molecular signals not only for comparative benchmarking but also for downstream molecular analyses. The researchers emphasize the importance of considering both sensitivity and specificity when evaluating in situ technologies. They also highlight the need for improved segmentation methods and computational approaches to address non-specific biases in downstream analyses. The study provides a framework for assessing new technologies, datasets, computational methods, and tissue preparation techniques in the field of spatial transcriptomics.