Zsom, A. | bioRxiv Science

bioRxiv2020

Automated identification of multinucleated germ cells with U-Net

Phthalic acid esters (phthalates) are male reproductive toxicants, which exert their most potent toxicity during a critical window of sensitivity in fetal development. In the fetal rat, exposure to phthalates reduces testosterone biosynthesis, alters the development of seminiferous cords and other male reproductive tissues, and induces the formation of abnormal multinucleated germ cells (MNGs). Identification of MNGs is a time-intensive process, and it requires specialized training to identify MNGs in histological sections. As a result, MNGs are not routinely quantified in phthalate toxicity experiments. In order to speed up and standardize this process, we have developed an improved method for automated detection of MNGs. Using hand-labeled histological section images with human-identified MNGs, we trained a convolutional neural network with a U-Net architecture to identify MNGs on unlabeled images. With unseen hand-labeled images not used in model training, we assessed the performance of the model, using five different configurations of the data. On average, the model reached near human accuracy, and in the best model, it exceeded it. The use of automated image analysis will allow data on this histopathological endpoint to be more readily collected for analysis of phthalate toxicity. Our trained model application code is available for download at github.com/brown-ccv/mngcount.

pharmacology and toxicology↗

bioRxiv2019

Automated data extraction from historical city directories the rise and fall of mid-century gas stations in Providence, RI

The location of defunct environmentally hazardous businesses like gas stations has many implications for modern American cities. To track down these locations, we present the directoreadr code (github.com/brown-ccv/directoreadr). Using scans of Polk city directories from Providence, RI, directoreadr extracts and parses business location data with a high degree of accuracy. The image processing pipeline ran without any human input for 94.4% of the pages we examined. For the remaining 5.6%, we processed them with some human input. Through hand-checking a sample of three years, we estimate that ~94.6% of historical gas stations are correctly identified and located, with historical street changes and non-standard address formats being the main drivers of errors. As an example use, we look at gas stations, finding that gas stations were most common early in the study period in 1936, beginning a sharp and steady decline around 1950. We are making the dataset produced by directoreadr publicly available. We hope it will be used to explore a range of important questions about socioeconomic patterns in Providence and cities like it during the transformations of the mid-1900s.

ecology↗