A Systematic Search over Deep Convolutional Neural Network Architectures for Screening Chest Radiographs.

Chest radiographs are primarily employed for the screening of pulmonary and cardio-/thoracic conditions. Being undertaken at primary healthcare centers, they require the presence of an on-premise reporting Radiologist, which is a challenge in low and middle income countries. This has inspired the development of machine learning based automation of the screening process. While recent efforts demonstrate a performance benchmark using an ensemble of deep convolutional neural networks (CNN), our systematic search over multiple standard CNN architectures identified single candidate CNN models whose classification performances were found to be at par with ensembles. Over 63 experiments spanning 400 hours, executed on a 11.3 FP32 TensorTFLOPS compute system, we found the Xception and ResNet-18 architectures to be consistent performers in identifying co-existing disease conditions with an average AUC of 0.87 across nine pathologies. We conclude on the reliability of the models by assessing their saliency maps generated using the randomized input sampling for explanation (RISE) method and qualitatively validating them against manual annotations locally sourced from an experienced Radiologist. We also draw a critical note on the limitations of the publicly available CheXpert dataset primarily on account of disparity in class distribution in training vs. testing sets, and unavailability of sufficient samples for few classes, which hampers quantitative reporting due to sample insufficiency.

Immunohistochemical evaluation of prime molecules in cervical lesions towards assessment of malignant potentiality.

OBJECTIVE: A comparative immunohistochemical evaluation of p63, CD105, and E-cadherin expression pattern in histopathologically confirmed normal cervical epithelium (NCM), dysplastic cervical epithelium (DYS) and squamous cell carcinoma (SCC) of uterine cervix towards assessing malignant potentiality of the precancerous condition. MATERIALS AND METHODS: The biopsies from cervical mucosa (normal, dysplasia, and cancer) were studied by routine hematoxylin and eosin (H and E) and by immunohistochemistry for p63, E-cadherin, and CD105 expression. The expressions of these molecules were assessed in a semiquantitative way by (i) counting p63 cell population and distribution, (ii) intensity scoring of E-cadherin along the expression path, and (iii) measuring CD105 expression density. RESULT: p63+ cells were highest in carcinomas followed by dysplasia and normal. An abrupt increase in CD105 expression was observed through change of normal to dysplasia and cancer. A decrease in membranous E-cadherin expression was noticed in the transformation from normal to precancer and cancers. CONCLUSION: The malignant potential of the dysplastic conditions is likely to be correlated with upregulation in p63 and CD105 expression and a simultaneous downregulation of membranous E-cadherin.