In a previous canine study, we demonstrated that volatile organic compounds specific to bladder cancer are present in urine headspace, subsequently showing that up to 70% of tumours can be correctly classified using an electronic nose. This study aimed to evaluate the sensitivity and specificity which can be achieved by a group of four trained dogs. In a series of 30 double-blind test runs, each consisting of one bladder cancer urine sample placed alongside six controls, the highest sensitivity achieved by the best performing dog was 73% (95% CI 55-86%), with the group as a whole correctly identifying the cancer samples 64% (95% CI 55-73%) of the time. Specificity of the dogs individually ranged from 92% (95% CI 82-97%) for urine samples obtained from healthy, young volunteers down to 56% (95% CI 42-68%) for those taken from older patients with non-cancerous urological disease. Odds ratio comparisons confirmed a significant decrease in performance as the extent of urine dipstick abnormality and/or pathology amongst the control population increased. Importantly, however, statistical analysis indicated that covariates such as smoking, gender and age, as well as blood, protein and /or leucocytes in the urine did not significantly alter the odds of response to the cancer samples. Our results provide further evidence that volatile biomarkers for bladder cancer exist in urine headspace, and that these have the potential to be exploited for diagnosis.

False negatives are recorded in every chemical detection system, but when animals are used as a scent detector, some false negatives can arise as a result of a failure in the link between detection and the trained alert response, or a failure of the handler to identify the positive alert. A false negative response can be critical in certain scenarios, such as searching for a live person or detecting explosives. In this study, we investigated whether the nature of sniffing behavior in trained detection dogs during a controlled scent-detection task differs in response to true positives, true negatives, false positives, and false negatives. A total of 200 videos of 10 working detection dogs were pseudorandomly selected and analyzed frame by frame to quantify sniffing duration and the number of sniffing episodes recorded in a Go/No-Go single scent-detection task using an eight-choice test apparatus. We found that the sniffing duration of true negatives is significantly shorter than false negatives, true positives, and false positives. Furthermore, dogs only ever performed one sniffing episode towards true negatives, but two sniffing episodes commonly occurred in the other situations. These results demonstrate how the nature of sniffing can be used to more effectively assess odor detection by dogs used as biological detection devices.

We conducted a short study investigating the pressure patterns produced by cancer detection dogs via a caninecentered interface while searching samples of amyl acetate. We advance previous work by providing further insights into the potential of the approach for supporting and partly automating the practice of cancer detection with dogs.

Prostate cancer is the second leading cause of cancer death in men in the developed world. A more sensitive and specific detection strategy for lethal prostate cancer beyond serum prostate specific antigen (PSA) population screening is urgently needed. Diagnosis by canine olfaction, using dogs trained to detect cancer by smell, has been shown to be both specific and sensitive. While dogs themselves are impractical as scalable diagnostic sensors, machine olfaction for cancer detection is testable. However, studies bridging the divide between clinical diagnostic techniques, artificial intelligence, and molecular analysis remains difficult due to the significant divide between these disciplines. We tested the clinical feasibility of a cross-disciplinary, integrative approach to early prostate cancer biosensing in urine using trained canine olfaction, volatile organic compound (VOC) analysis by gas chromatography-mass spectroscopy (GC-MS) artificial neural network (ANN)-assisted examination, and microbial profiling in a double-blinded pilot study. Two dogs were trained to detect Gleason 9 prostate cancer in urine collected from biopsy-confirmed patients. Biopsy-negative controls were used to assess canine specificity as prostate cancer biodetectors. Urine samples were simultaneously analyzed for their VOC content in headspace via GC-MS and urinary microbiota content via 16S rDNA Illumina sequencing. In addition, the dogs’ diagnoses were used to train an ANN to detect significant peaks in the GC-MS data. The canine olfaction system was 71% sensitive and between 70–76% specific at detecting Gleason 9 prostate cancer. We have also confirmed VOC differences by GC-MS and microbiota differences by 16S rDNA sequencing between cancer positive and biopsy-negative controls. Furthermore, the trained ANN identified regions of interest in the GC-MS data, informed by the canine diagnoses. Methodology and feasibility are established to inform larger-scale studies using canine olfaction, urinary VOCs, and urinary microbiota profiling to develop machine olfaction diagnostic tools. Scalable multi-disciplinary tools may then be compared to PSA screening for earlier, non-invasive, more specific and sensitive detection of clinically aggressive prostate cancers in urine samples.

We report on participatory design research where interaction designers, and canine behavioral specialists, together with their cancer detection dogs, teamed up to better support the dogs' life-saving work. We discuss interspecies communication challenges in cancer detection training, requiring the dogs to use human signaling conventions that perturb their detection work. We describe our effort to develop a technology that could resolve those challenges, and how in the process our design focus gradually shifted from a human-centered to a caninecentered interaction model. The resulting interface, based on honest signaling, re-centers cancer detection practices on the dogs themselves, enabling them to better express their potential as cancer detection workers; it also provides a model for re-thinking human-computer interactions.