Kazemi, E. | Search bioRxiv

bioRxiv2018

Robust Automated Assessment of Human Blastocyst Quality using Deep Learning

Morphology assessment has become the standard method for evaluation of embryo quality and selecting human blastocysts for transfer in in vitro fertilization (IVF). This process is highly subjective for some embryos and thus prone to human bias. As a result, morphological assessment results may vary extensively between embryologists and in some cases may fail to accurately predict embryo implantation and live birth potential. Here we postulated that an artificial intelligence (AI) approach trained on thousands of embryos can reliably predict embryo quality without human intervention.\n\nTo test this hypothesis, we implemented an AI approach based on deep neural networks (DNNs). Our approach called STORK accurately predicts the morphological quality of blastocysts based on raw digital images of embryos with 98% accuracy. These results indicate that a DNN can automatically and accurately grade embryos based on raw images. Using clinical data for 2,182 embryos, we then created a decision tree that integrates clinical parameters such as embryo quality and patient age to identify scenarios associated with increased or decreased pregnancy chance. This IVF data-driven analysis shows that the chance of pregnancy varies from 13.8% to 66.3%.\n\nIn conclusion, our AI-driven approach provides a novel way to assess embryo quality and uncovers new, potentially personalized strategies to select embryos with an improved likelihood of pregnancy outcome.

bioinformatics↗

bioRxiv2017

Deep Convolutional Neural Networks Enable Discrimination of Heterogeneous Digital Pathology Images

Pathological evaluation of tumor tissue is pivotal for diagnosis in cancer patients and automated image analysis approaches have great potential to increase precision of diagnosis and help reduce human error.\n\nIn this study, we utilize various computational methods based on convolutional neural networks (CNN) and build a stand-alone pipeline to effectively classify different histopathology images across different types of cancer. In particular, we demonstrate the utility of our pipeline to discriminate between two subtypes of lung cancer, four biomarkers of bladder cancer, and five biomarkers of breast cancer. In addition, we apply our pipeline to discriminate among four immunohistochemistry (IHC) staining scores of bladder and breast cancers.\n\nOur classification pipeline utilizes a basic architecture of CNN, Googles Inceptions within three training strategies, and an ensemble of two state-of-the-art algorithms, Inception and ResNet. These strategies include training the last layer of Googles Inceptions, training the network from scratch, and fine-tunning the parameters for our data using two pre-trained version of Googles Inception architectures, Inception-V1 and Inception-V3.\n\nWe demonstrate the power of deep learning approaches for identifying cancer subtypes, and the robustness of Googles Inceptions even in presence of extensive tumor heterogeneity. Our pipeline on average achieved accuracies of 100%, 92%, 95%, and 69% for discrimination of various cancer types, subtypes, biomarkers, and scores, respectively. Our pipeline and related documentation is freely available at https://github.com/ih-lab/CNN_Smoothie.

bioinformatics↗