Urinary tract infection (UTI) is a common disorder. Its diagnosis can be made by microscopic examination of voided urine for markers of infection. This manual technique is technically difficult, time-consuming and prone to inter-observer errors. The application of computer vision to this domain has been slow due to the lack of a clinical image dataset from UTI patients. We present an open dataset containing 300 images and 3,562 manually annotated urinary cells labelled into seven classes of clinically significant cell types. It is an enriched dataset acquired from the unstained and untreated urine of patients with symptomatic UTI using a simple imaging system. We demonstrate that this dataset can be used to train a Patch U-Net, a novel deep learning architecture with a random patch generator to recognise urinary cells. Our hope is, with this dataset, UTI diagnosis will be made possible in nearly all clinical settings by using a simple imaging system which leverages advanced machine learning techniques.
Chronic urinary tract infection (UTI) presents with protracted lower urinary tract symptoms and elevated urinary leukocyte counts, but its bacterial etiological agents remain obscure. In this cross-sectional investigation, we aimed to unravel the role of the bladder microbiota in chronic UTI pathogenesis by studying the host immune response. Urine samples were collected from healthy controls (HT), chronic UTI patients who had not initiated treatment (PT) and those undergoing treatment (OT), then sorted into white blood cell (WBC) and epithelial cell (EPC) fractions. Bacteria associated with both fractions were identified by chromogenic agar culture coupled with mass spectrometry and 16S rRNA sequencing. Distinct WBC-exclusive bacteria were observed in the healthy population, but this pattern was less obvious in patients, plausibly due to epithelial shedding and breaching of the urothelial barrier. We also described a bacterial fingerprint guided by Escherichia that was able to stratify patients based on symptom severity. Clustering analyses of mean rank changes revealed highly statistically significant upward and downward ecological shifts in communities of bacteria between the healthy and diseased populations. Interestingly, many of the most abundant genera identified in sequencing remained stable when compared between the study cohorts. We concluded that reshuffling of the urinary microbiome, rather than the activity of a single known urinary pathogen, could drive chronic UTI.
Urinary tract infection (UTI) is a common disorder. Its diagnosis can be made by microscopic examination of voided urine for cellular markers of infection. This manual technique is technically difficult, time-consuming and prone to inter-observer errors. The application of computer vision to this domain has been slow due to the lack of a clinical image dataset from UTI patients. We present an open dataset containing 300 images and 3,562 manually annotated urinary cells labelled into seven classes of clinically significant cell types. It is an enriched dataset acquired from the unstained and untreated urine of patients with symptomatic UTI using a simple imaging system. We demonstrate that this dataset can be used to train a Patch U-Net, a novel deep learning architecture with a random patch generator to recognise urinary cells. Our hope is that with this dataset UTI diagnosis will be made possible in nearly all clinical settings by using a simple imaging system which leverages advanced machine learning techniques.