Test the ability of temporal diffusion spectroscopy to detect lymphocyte fractions in cancer cell samples.HM7, a human and MC38, a mouse adenocarcinoma cell line were expanded in RPMI + 10% FCS. Lymphocytes were isolated from C57BL/6 mice (n=2) activated for 48 hours with CD3/CD28 activation beads and cultured for 7 days in T-cell media (RPMI, 10% FCS, 1% Pen/Strep, 2 mM GlutaMax, 1 mM Pyruvate, 50 µM 2-Mercaptoethanol, 50 IU/ml murine Interleukin 2). HM7 and MC38 cells were detached using Trypsin and cell concentrations and sizes were measured (Beckman Coulter Vi-cell). HM7 and MC38 cells were mixed at defined ratios with activated lymphocytes, re-suspended in 2% PFA, transferred to small tubes and centrifuged at 500g for 15 minutes. Diffusion weighted images of sample tubes embedded in 2% agar were acquired on a 9.4T horizontal bore scanner (Agilent, Santa Clara, CA) with a gradient insert (max. 100 G/cm). To cover short diffusion times, a fast spin echo sequence with cosine shaped diffusion gradients was used: FOV 38x38x1 mm³, Matrix 256x256, TR 4000 ms, effective TE 60 ms, ETL 8, 4 signal averages, max. diffusion gradient 49 G/cm, δ/Δ 25/31.5 ms, oscillation frequency: 40, 80, 120, 160, 200, or 240 Hz, 6 b-values (27, 41, 56, 71, 85 and 100% of max. diffusion gradient), diffusion gradient applied along frequency encode direction, max. b-value: 400 s/mm² except for 200 and 240 Hz were max. b-values of 291 and 210 s/mm² were used respectively. For diffusion times greater than 10 ms, a stimulated echo sequence with trapezoid diffusion gradients was used: TR 4000 ms, TE 10.5 ms, max. diffusion gradient 29 G/cm, δ/Δ 1.8/(15, 30, or 45) ms, 6 b-values (27, 41, 56, 71, 85 and 100% of max. diffusion gradient), diffusion gradient applied along frequency encode direction, max. b-value: 400 s/mm². ADC maps were generated by voxel based log-linear fits using Matlab R2013a (The Mathworks, Natick, MA). Equal sized region of interests were used to estimate average ADC values for different cell pellets. T₂ weighted fast spin echo images were acquired to generate T₂ maps: TR 4000 ms, TE 6-60 ms with 8 steps. All experiments were repeated once.Adenocarcinoma cells MC38 and HM7 were larger compared to activated lymphocytes with respective diameters of 14.4±1.0, 12.4±0.1 and 10.2±0.1 µm. This difference in cell size led to contrasting diffusion characteristics by which cell types could be clearly identified (Figure 1). The gradient slew rate limited the maximum b-value that could be obtained for oscillation frequencies above 160 Hz. Higher variability due to lower b-values reduced the ability to distinguish ADCs from different cell types at these frequencies (Figure 2). Diffusion spectra from cell pellets containing 90% HM7 and 10% activated lymphocytes or pure HM7 were indistinguishable (Figure 3). A mixture of 80% HM7 and 20% activated lymphocytes was slightly different from pure HM7 while a mixture containing 40% activated lymphocytes was clearly separable (Figure 3). Separating a 40% lymphocyte / 60% MC38 mixture from pure MC38 was easier due to higher average ADCs for MC38 cells increasing the difference between them and activated lymphocytes (Figure 4). Cell pellets had similar T₂ values ranging from 39-56 ms.Cell pellets containing more than 20% activated lymphocytes could be separated from cell pellets containing only adenocarcinoma cells using temporal diffusion spectroscopy. The sensitivity of this approach increases with the initial size difference of respective cell types. Based on these results a substantial number of lymphocytes would be required for detection. However, the sensitivity to detect inflammation might be enhanced due to the large number of monocytes and macrophages present in inflamed tissues.Temporal diffusion spectroscopy would require a substantial number of lymphocytes for detection but the sensitivity may be sufficient to detect inflammation in highly inflamed tissues.