Fig. 4
From: Detection of early-stage lung cancer in sputum using automated flow cytometry and machine learning
Heuristics-guided Singlets Gating. A In some cases, cells with intermediate FVS510 signal and high SSC-A throw off singlets gating on the full viable cell population (e.g., lower left corner below zero or top left point higher than bottom right). B A temporary gate is set on SSC-A to exclude problematic events above 5 × 104. C A singlets gate is fitted to the restricted population (red polygon) and adjusted to include the upper diagonal by setting the upper right corner to 2.5 × 105 on both axes (navy dashed polygon). D Temporary gates are removed and the tweaked singlets gate (red polygon) applied to the full viable population (compare the adjusted polygon in D to the original red polygon in A. E In some cases, a population representing > 10% of singlets (red oval) lies between 2.5 (logicle scale) and the viability threshold. F Population mixture analysis highlights the difference in signal distribution of the rightmost population identified by the oval in E (blue curve) relative to the bulk of the events left from the oval in E (black curve) and suggests a natural cutoff at 2.5 (dashed red line) in these unusual cases. G The adjusted viability cutoff (red line) replaces the one found by automated tail gating (dashed black line). H Finally, a new singlets gate for the refined viable cell population is calculated (red polygon). A–D are from a different patient sample than E, F to illustrate the heuristics applied in singlets gating