Sepsis is a life-threatening condition affecting an estimated 49 million people annually and causing approximately 20% of the deaths worldwide. Survival in septic shock decreases by about 8% per hour of delayed or inappropriate treatment, making rapid diagnosis critical. Consequently, patient outcomes strongly depend on the rapid initiation of appropriate antimicrobial therapy, making sepsis diagnosis a fight against time. However, current blood culture-based workflows typically require several days to deliver actionable results, largely due to the extremely low bacterial concentrations in blood (1-100 CFU/mL) that necessitate time-consuming culture steps. There is therefore a pressing need for diagnostic protocols capable of rapidly isolating and characterizing bacteria directly from blood.

In this work, we address a key limitation in current diagnostic pipelines by developing and evaluating culture-free approaches for the rapid isolation, detection, species identification, and antimicrobial susceptibility testing (AST) of bacteria directly from blood, starting at clinically relevant concentrations. Two centrifugation-based sample preparation strategies were developed: one compatible with samples drawn into blood culture bottles and one designed for whole blood. Using the first approach, we demonstrated the isolation and identification of five common sepsis-causing bacteria within 12 hours. This approach relies solely on standard laboratory equipment, which facilitates direct translation to clinical laboratories. The whole-blood approach combines centrifugation and selective blood cell lysis with microfluidic trapping and deep learning-based automated detection, and enables culture-free detection of bacteria from blood within 2 h. Building on this foundation, the workflow was further extended to integrate real-time automated detection, single-cell phenotypic AST, and species identification by fluorescence in situ hybridization (FISH), directly from uncultured blood in under 7 h. 

The sample preparation steps of these protocols rely heavily on manual processing. Therefore, subsequent work focused on the development of a one-step centrifuge device that automates bacterial isolation and up-concentration while maintaining compatibility with downstream processes, such as subculturing, mass spectrometry-based identification, and microfluidic single-cell analysis. All approaches were evaluated using healthy human blood samples spiked with common sepsis-causing bacteria, and serve as proof-of-concept of rapid, culture-free bacterial characterization directly from blood.

Together, the results presented in this thesis demonstrate the potential of combining centrifugation-based sample preparation, microfluidics, and automated image analysis to substantially shorten diagnostic turnaround times for sepsis and bloodstream infections. Although further validation with clinical samples and increased automation are required, this work provides experimental and methodological advances toward faster sepsis diagnostics.

Sepsis är ett livshotande tillstånd som drabbar uppskattningsvis 49 miljoner människor årligen och står för ungefär 20% av alla dödsfall världen över. Överlevnad i septisk chock minskar med cirka 8% för varje timme som behandling fördröjs eller hanteras fel, varför en snabb diagnos är kritisk. Patientens utfall är därför strängt beroende av en snabb initiering av en korrekt behandling, vilket gör sepsisdiagnostik till en kamp mot klockan. Nuvarande arbetsflöden med blododling kräver dock vanligtvis flera dagar innan man får resultat som går att agera på, till stor del beror detta på extremt låga koncentrationer av bakterier i blodprover (1-100 CFU/ml) som kräver tidsödande odlingssteg. Det finns därför ett tryckande behov av diagnostiska protokoll som klarar att snabbt isolera och identifiera bakterier direkt från blod.

I det här arbetet addresserar vi centrala begränsningar i nuvarande diagnostiska flöden genom att utveckla och utvärdera odlingsfria metoder för snabb isolering, detektion och artidentifiering, samt antimikrobiell susceptibilitetstestning (AST) av bakterier direkt från blod, testad vid kliniskt relevanta koncentrationer. Två centrifug-baserade metoder för blodprovsberedning utvecklades: den ena kompatibel med prover som är tagna i blododlingsflaskor och den andra utvecklad för helblod. Genom den första metoden demonstrerar vi isolering och identifiering av fem vanliga bakterier som orsakar sepsis inom 12 timmar. Denna metod utnyttjar enbart standardiserad labbutrustning vilket möjliggör direkt translation hos kliniska laboratorier. Metoden för helblod kombinerar centrifugering och selektiv lysering av blod med mikrofluidiska fällor och deep learning baserad automatisk detektion, och möjliggör odlingsfri detektion av bakterier i blod inom 2 timmar. Med detta som bas vidareutvecklades arbetsflödet till att också inkludera automatisk realtidsdetektion, singel-cell fenotyp AST och artidentifiering genom fluorescens in situ hybridisering (FISH), direkt från blod på under 7 timmar. 

Provberedningsstegen i dessa protokoll är väldigt beroende av manuella processer. Således fokuserade efterföljande arbete på utvecklingen av en device för enstegs-centrifugering som automatiserar isolering av bakterier och uppkoncentrering och som samtidigt bibehåller kompatibilitet med processer nedströms, såsom subodlingar, mass-spektrometri-baserad identifikation, samt mikrofluidisk singelcells-analys. Alla metoder utvärderades med hjälp av friska humana blodprover som spikats med vanliga sepsisorsakande bakterier och tjänar som ett proof-of-concept för snabb, odlingsfri bakteriell karakterisering direkt från blod. 

Tillsammans demonstrerar resultaten i denna avhandling potentialen i att kombinera centrifugeringsbaserad provberedning, mikrofluidik och automatiserad bildanalys för att avsevärt förkorta diagnostiska svarstider för sepsis och infektioner i blodbanan. Även om vidare arbete med validering med kliniska prover och ökad automatisering krävs, bidrar detta arbete med experimentella och metodologiska framsteg mot snabbare sepsisdiagnostik. 

Approximately 50 million people suffer from sepsis yearly, and 13 million die from it. For every hour a patient with septic shock is untreated, their survival rate decreases by 8%. Therefore, rapid detection and antibiotic susceptibility profiling of bacterial agents in the blood of sepsis patients are crucial for determining appropriate treatment. Here, we introduce a method to isolate bacteria from whole blood with high separation efficiency through Smart centrifugation , followed by microfluidic trapping and subsequent detection using deep learning applied to microscopy images. We detected, within 2 h, E. coli , K. pneumoniae , or E. faecalis from spiked samples of healthy human donor blood at clinically relevant concentrations as low as 9, 7 and 32 colony-forming units per ml of blood, respectively. However, the detection of S. aureus remains a challenge. This rapid isolation and detection represents a significant advancement towards culture-free detection of bloodstream infections.

Sepsis has an incidence of 50 million cases per year and represents a significant cause of morbidity and mortality worldwide. Current diagnostic methods rely on blood drawn directly into blood culture media in hemoculture bottles, followed by culturing, often taking days to yield results and failing to meet urgent clinical needs. We present here a protocol for isolating and identifying bacteria from blood within 12 h after sampling, bypassing prior hemocultures. Starting from blood added into blood culture media, according to standard hospital sampling practice, we isolated up to 85% of bacteria at clinically-relevant concentrations in less than 15 min using an optimized centrifugation protocol. Subsequent overnight culture of the isolated bacteria on chromogenic agar plates enabled species identification of five of the most prevalent sepsis-causing bacteria. The rapidity and simplicity of the protocol may accelerate the diagnostic pipeline for sepsis patients. Moreover, the use of standard laboratory equipment may enable direct translation to clinical praxis and concatenation with downstream assays for antibiotic susceptibility testing.

Sepsis is a time‐critical condition causing over 13 million deaths annually, with each hour of treatment delay in patients with septic shock increasing mortality by 8%. Rapid pathogen identification is crucial, yet current workflows depend on multiple culture steps that delay pathogen identification and targeted treatment by days. A plug‐and‐play, fully automated centrifuge tube is presented that isolates and concentrates bacteria directly from blood or blood culture using only conventional lab centrifuges. Each tube can process 7.5 ml of sample and yields, within 40 min, a 0.7 mL clear suspension with greater than threefold enhanced bacteria concentration and 99.9% blood cell rejection, ready for downstream detection. It is demonstrated that this approach supports key diagnostic workflows, including 1) a novel isolate‐then‐culture strategy detecting bacterial concentrations as low as 10 CFU/mL; 2) direct matrix‐assisted laser desorption ionization time‐of‐flight (MALDI‐TOF) identification, bypassing subculturing, and; 3) microfluidic single‐cell detection. This fully automated platform is compatible with existing centrifuges, is anticipated to facilitate broader adoption in routine clinical practice, while its ability to enable rapid, same‐workshift bacterial enhancement can reduce diagnostic time by about one day in the context of time‐critical sepsis diagnostics.

We report a novel workflow for the isolation of bacteria from whole blood samples. We detected E. coli from 50 μL of whole blood in concentrations as low as 400 CFU/mL in less than 1 h using a microfluidic device. The method presented consists of reducing the number of blood cells found in whole blood by selectively lysing the blood cells with a solution of saponin and sodium cholate. We demonstrate that the bacteria grow during the time of loading in the microchip and can continue to be grown afterwards on the chip. Additionally, we show our approach is compatible with some of the most common sepsis-causing bacteria. The method can half the bacterial culture time needed for antibiotic susceptibility testing (AST) of bloodstream infections.