. Author manuscript; available in PMC: 2015 Aug 28.

Published in final edited form as: J Immunol Methods. 2014 Nov 13;416:84–93. doi: 10.1016/j.jim.2014.11.004

Table 1.

Comparison of performance of tracking T cells

Qualitative characteristics	Tool Name ^a	Number of tracks	Mean cell density (mm⁻²)	Mean speed (μm/min)	Mean track - length (μm)	SFDA ^b	SFDA ^b	ATA ^b
Qualitative characteristics	Tool Name ^a	Number of tracks	Mean cell density (mm⁻²)	Mean speed (μm/min)	Mean track - length (μm)	SFDA ^b	(thresholded)
Experiment 1 Antigen-induced motility of CD45RA+ve human CD8 T cells; 33.3 seconds between frames; 40x objective with 1.3 NA; Pixel dimensions were 0.439 μm; 225 μm square field; 100 frames	Ground-truth	125	1243.85	5.49	115.10
	TIAM	136	1171.16	5.1	87.32	0.636	0.945	0.596
	DYNAMIK	214	715.06	5.76	44.94	0.638	0.888	0.302
	Imaris (DIC)	144	1196.84	5.87	100.64	0.412	0.678	0.508
	Icy	240	1150.62	5.19	47.8	0.574	0.882	0.380
	Imaris	147	1070.22	5.40	77.12	0.566	0.837	0.459
	Volocity	184	1212.84	5.77	80.96	0.490	0.755	0.452

Experiment 2 Chemokine-induced fast amoeboid migration of CD45RA+ve human CD8 T cells; 20 seconds between frames; 25x objective with 0.8 NA; Pixel dimensions were 0.664 μm; 340 μm square field;100 frames	Ground-truth	198	923.18	9.40	158.38
	TIAM	247	906.05	9.65	118.37	0.497	0.869	0.545
	DYNAMIK	481	497.49	11.07	37.80	0.488	0.783	0.252
	Imaris (DIC)	431	926.21	12.67	82.15	0.330	0.526	0.304
	Icy	471	989.01	9.87	66.67	0.386	0.619	0.316
	Imaris	252	946.54	9.59	127.68	0.406	0.659	0.475
	Volocity	323	988.67	9.44	102.66	0.358	0.586	0.441

Ground-truth was established using DIC image series. Tracking on DIC image series was performed using TIAM, DYNAMIK and Imaris. Tracking was also performed on fluorescent image series of the same field collected in parallel with DIC. Icy, Imaris and Volocity were chosen for fluorescent particle tracking. Detection and tracking parameters used for each tool are described in Supplementary methods section. Tracks shorter than 5 frames were discarded. The same ground-truth was applied to both DIC and fluorescence image series as there is excellent registration between them (Figure S9). Similar tracking performance was ob served within each tool over a range of values of maximum allowed distance for linking cells. Nonetheless, optimal values obtained for each tool is reported.

Uniform box-sizes were considered around centroids of cells as results from Imaris and Volocity do not provide outlines of cells. Results shown here were with a box-size of 20 pixels. For thresholded performance metrics, Jaccard Similarity of 0.4 or more is considered as 1 (see Supplementary methods). Thus, minor localization inaccuracy is not penalized.