Young, R. J. | bioRxiv Science

bioRxiv2021

Snapshot of the Atlantic Forest canopy: surveying arboreal mammals in a biodiversity hotspot

The Atlantic Forest of South America supports some of the greatest terrestrial biodiversity on our planet but is now reduced to only a small extent of its original forest cover. It hosts a large number of endemic mammalian species but our knowledge on arboreal mammal ecology and conservation has been hindered by the challenges of observing arboreal species from ground level. Camera trapping has proven to be an effective tool in terrestrial mammal monitoring, but the technique has rarely been used for arboreal species. Here we generated data on the arboreal mammal community based on canopy camera trapping for the first time in the Atlantic Forest, focusing on Caparao National Park, Brazil. We placed 24 infrared camera traps in the forest canopy distributed in seven areas within the Park, operating continuously from January 2017 to June 2019. In this time, they accumulated 4,736 camera-days and generated 2,256 sets of pictures and 30 second videos of vertebrates. Canopy camera traps were able to detect arboreal mammals spanning a wide variety of body sizes. The local mammal assemblage comprised of 15 identifiable species, including the critically endangered northern muriqui (Brachyteles hypoxanthus) and the buffy-headed marmoset (Callithrix flaviceps), and other rare, nocturnal and inconspicuous species. For the first time, we confirmed the occurrence of the thin-spined porcupine (Chaetomys subspinosus) in the Park. Species richness varied across sampling areas and forest types. Our findings demonstrate the potential of canopy camera trapping for future surveying efforts to better inform conservation strategies for mammals.

zoology↗

bioRxiv2021

Deep Learning Achieves Neuroradiologist-Level Performance in Detecting Hydrocephalus

PurposeIn large clinical centers a small subset of patients present with hydrocephalus that requires surgical treatment. We aimed to develop a screening tool to detect such cases from the head MRI with performance comparable to neuroradiologists. MethodsWe leveraged 496 clinical MRI exams collected retrospectively at a single clinical site from patients referred for any reason. This diagnostic dataset was enriched to have 259 hydrocephalus cases. A 3D convolutional neural network was trained on 16 manually segmented exams (ten hydrocephalus) and subsequently used to automatically segment the remaining 480 exams and extract volumetric anatomical features. A linear classifier of these features was trained on 240 exams to detect cases of hydrocephalus that required treatment with surgical intervention. Performance was compared to four neuroradiologists on the remaining 240 exams. Performance was also evaluated on a separate screening dataset of 451 exams collected from a routine clinical population to predict the consensus reading from four neuroradiologists using images alone. The pipeline was also tested on an external dataset of 31 exams from a 2nd clinical site. ResultsThe most discriminant features were the Magnetic Resonance Hydrocephalic Index (MRHI), ventricle volume, and the ratio between ventricle and brain volume. At matching sensitivity, the specificity of the machine and the neuroradiologists did not show significant differences for detection of hydrocephalus on either dataset (proportions test, p > 0.05). ROC performance compared favorably with the state-of-the-art (AUC 0.90-0.96), and replicated in the external validation. ConclusionHydrocephalus cases requiring treatment can be detected automatically from MRI in a heterogeneous patient population based on quantitative characterization of brain anatomy with performance comparable to that of neuroradiologists.

bioinformatics↗