You are here:
Automated identification of multinucleated germ cells with U-Net
Citation:
Bell, S., A. Zsom, J. Conley, AND D. Spade. Automated identification of multinucleated germ cells with U-Net. PLOS ONE . Public Library of Science, San Francisco, CA, 15(7):e0229967, (2020). https://doi.org/10.1371/journal.pone.0229967
Impact/Purpose:
This manuscript is a product of an on-going collaboration between USEPA researchers (Earl Gray, Justin Conley) and researchers at the Brown University Department of Pathology and Laboratory Medicine (Kim Boekelheide, Dan Spade). This collaborative research involves investigation of the molecular mechanisms and pathological and teratogenic effects of in utero exposure to phthalates on male reproductive tissues. Phthalate exposure is nearly ubiquitous as many structures within the chemical class are utilized in the manufacturing of consumer products. Further, there are currently multiple phthalates on the TSCA High-priority contaminants list. This manuscript does not report toxicity data, rather it reports the development and performance of an automated method for enumerating multi-nucleated germ cells in late-gestation fetal rat testis, which is a consistent adverse effect of in utero exposure to active phthalates, but is rarely measured due the time consuming nature and expertise required. This method will be useful for future collaborative research with our group and those at Brown University, as well as other researchers in the field.
Description:
Phthalic acid esters (phthalates) are male reproductive toxicants, which exert their most potent toxicity during 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.
