Blood stream infections are challenging infections to rapidly diagnose. The current clinical diagnostic methods for blood stream infections require culturing the blood sample prior to identifying the bacteria and any resistance the bacteria may contain. Removing the culturing step from the bacterial identification process of a blood stream infection provides a significant reduction in the processing time. However, eliminating the culturing step shifts the difficulty from processing time to concentration, since clinical concentration levels can be as low as 10 CFU/mL in blood. This dissertation developed and evaluated many aspects of the process required to identify bacteria from a blood stream infection without culturing the bacteria. Two new methods of separating the bacteria from the blood cells were developed: inducing clotting using a centrifugal-sedimentation on a hollow disk, and filtering whole blood. Inducing clotting achieved 69\% bacterial recovery from 7 mLs of whole blood in 117 s. Filtering whole blood achieved 100\% bacterial removal from 5 mLs of whole blood in $\approx 90$ s, but the bacteria were difficult to remove from the filter. Bacterial removal from the filter after blood filtration was also investigated. At a very low bacterial concentration of 200 CFU/mL, a blood lysis solution of 3\% Tween 80 followed by a 3\% Pluronic F108 backflush solution achieved 60\% removal of the bacteria from the filter. In addition to developing two new methods, a previously developed technique using centrifugal-sedimentation on a hollow disk underwent a stability analysis in order to decrease the occurrence of mixing. This analysis yielded the development of the analytical solution to the Navier-Stokes equations for a two-fluid flow with a moving wall boundary and a free surface. The analysis also experimentally identified a stability boundary that was found to be in good agreement with the Kelvin-Helmholtz instability model. After exploring the methods to recover bacteria from blood, experiments were performed to identify a bacterial lysing solution that could lyse \textit{E. coli}, \textit{E. cloacae} and \textit{K. pneumoniae} bacteria. The best bacterial lysing solution consisted of incubating the bacteria with 1 mg/mL lysozyme for 10 min followed by the addition of 6 M GHCl and 1\% SDS. This solution obtained a 46\% DNA recovery. The DNA were then fragmented by ultrasound to reduce the segment length for DNA labelling. In addition to lysing and fragmenting the DNA, a microfluidic device was prototyped and tested for incorporating the lysing, capturing, releasing, and fragmenting of the DNA all on a single device. Whole experiments were performed which extracted the bacteria from the blood, removed and collected the DNA from the bacteria, and fragmented the DNA. The best overall recovery from an experiment performing the whole process was 26.8\%. The 26.8\% recovery was achieved with a 68\% recovery of the bacteria from spinning and a 54.1\% removal of bacteria from off of the filter and a 72.9\% recovery of the DNA from the bacteria.





Date Submitted


Document Type





bacteria, blood, instability, filtering, DNA extraction, antibiotic resistance, microfluidic device



Included in

Engineering Commons