RESUMO
INTRODUCTION: Glaucoma and non-arteritic anterior ischemic optic neuropathy (NAION) are optic neuropathies that can both lead to irreversible blindness. Several studies have compared optical coherence tomography angiography (OCTA) findings in glaucoma and NAION in the presence of similar functional and structural damages with contradictory results. The goal of this study was to use a deep learning system to differentiate OCTA in glaucoma and NAION. MATERIAL AND METHODS: Sixty eyes with glaucoma (including primary open angle glaucoma, angle-closure glaucoma, normal tension glaucoma, pigmentary glaucoma, pseudoexfoliative glaucoma and juvenile glaucoma), thirty eyes with atrophic NAION and forty control eyes (NC) were included. All patients underwent OCTA imaging and automatic segmentation was used to analyze the macular superficial capillary plexus (SCP) and the radial peripapillary capillary (RPC) plexus. We used the classic convolutional neural network (CNN) architecture of ResNet50. Attribution maps were obtained using the "Integrated Gradients" method. RESULTS: The best performances were obtained with the SCP + RPC model achieving a mean area under the receiver operating characteristics curve (ROC AUC) of 0.94 (95% CI 0.92-0.96) for glaucoma, 0.90 (95% CI 0.86-0.94) for NAION and 0.96 (95% CI 0.96-0.97) for NC. CONCLUSION: This study shows that deep learning architecture can classify NAION, glaucoma and normal OCTA images with a good diagnostic performance and may outperform the specialist assessment.
RESUMO
While pap test is the most common diagnosis methods for cervical cancer, their results are highly dependent on the ability of the cytotechnicians to detect abnormal cells on the smears using brightfield microscopy. In this paper, we propose an explainable region classifier in whole slide images that could be used by cyto-pathologists to handle efficiently these big images (100,000x100,000 pixels). We create a dataset that simulates pap smears regions and uses a loss, we call classification under regression constraint, to train an efficient region classifier (about 66.8% accuracy on severity classification, 95.2% accuracy on normal/abnormal classification and 0.870 KAPPA score). We explain how we benefit from this loss to obtain a model focused on sensitivity and, then, we show that it can be used to perform weakly supervised localization (accuracy of 80.4%) of the cell that is mostly responsible for the malignancy of regions of whole slide images. We extend our method to perform a more general detection of abnormal cells (66.1% accuracy) and ensure that at least one abnormal cell will be detected if malignancy is present. Finally, we experiment our solution on a small real clinical slide dataset, highlighting the relevance of our proposed solution, adapting it to be as easily integrated in a pathology laboratory workflow as possible, and extending it to make a slide-level prediction.