Rajasekaran Subramanian

Abstract

Epithelium tissue covers and lines all internal organs of the body. Breast cancer carcinomas arise from the epithelial cells of the breast. The epithelial component lines tubules, ducts, etc. and segmenting out this region from other tissues is important to detect breast cancer. This paper proposes application of deep learning technique U-Net, to segment epithelium tissue automatically from a whole slide image (WSI) image. It also implements U-Net, an image segmentation algorithm to automatically learn the features of epithlium components from the experts-annotated WSI image dataset and tested the results. Various image processing techniques such as thresholding are applied to improve the quality of dataset before and after the training of images by U-Net. The performance of the U-Net is measured by statistic parameters Sørensen–Dice coefficient and F1 score. The automatic system generated an epithelium segmentation of accuracy of 0.932. Context: Automatic, faster and accurate computational technique implementation for breast cancer diagnostic pathology. Aims: Automatic identification of carcinoma in WSI of Breast Epithelium Tissue by using deep learning network, U-Net. Settings and Design: Epithelium tissue covers and lines all internal organs of the body. Breast cancer carcinomas arise from the epithelial cells of the breast. The epithelial component lines tubules, ducts, etc. and segmenting out this region from other tissues is important to detect breast cancer. Methods and Material: This paper proposes application of deep learning technique U-Net, to segment epithelium tissue automatically from a whole slide image (WSI) image. It also implements U-Net, an image segmentation algorithm to automatically learn the features of epithlium components from the experts-annotated WSI image dataset and tested the results. Various image processing techniques such as thresholding are applied to improve the quality of dataset before and after the training of images by U-Net. Statistical analysis used: The performance of the U-Net is measured by statistic parameters Sørensen–Dice coefficient and F1 score. Results: The automatic system achieved an accuracy of 0.932 for the given dataset. Conclusions: The automatic system achieved an accuracy of 0.932 for 40 images. System should be trained for more images to realistically achieve more accuracy for higher complex digital pathology images. Key Messages: [S1] Automatic identification of carcinoma in WSI of Breast Epithelium Tissue by using deep learning network, U-Net. Keywords: Carcinoma, Breast Cancer, Epithelial Tissue, U-Net, Epithelium Segmentation, CNN, Diagnostic Pathology, Digital Pathology, WSI Images

View Article

Abstract

The evaluation of lymph nodes’ metastasis is an important component of Tumor, Node, Metastasis (TNM) breast cancer staging system for better clinical management and treatment. Assessing lymph node metastasis through histologic examination is the most accurate method. This paper proposes significantly advanced and faster image classification Convolutional Neural Network (CNN) model called Densenet-161 for lymph node metastasis. This paper uses pre-processing technique called image thresholding to improve the contrast intensities of the SLN images, which improves the performance of DenseNet. The experimental PCam dataset contains 327,680 patches extracted from 400 Haemotoxylin and Eosin (H&E) stained WSIs of breast cancer with sentinel lymph node sections. The proposed system has generated 94% accuracy for lymph node metastasis classification. Context: Automatic, faster and accurate computational technique implementation for breast cancer sentinel lymph metastases classification in cancer diagnostic pathology. Aims: Automatic sentinel lymph node metastases classification on breast carcinoma WSI through deep learning network, DenseNet-161. Settings and Design: Tumor cells are migrating from a primary metastasize to one or a few lymph nodes, before spreading to other lymph nodes. These few lymph nodes are called as ”sentinel” lymph nodes. The status of these sentinel lymph nodes would accurately predict the status of the remaining lymph nodes. Lymph node status assessment is considered to be one of the most important independent prognostic factors in breast cancer. Methods and Materials: Breast cancer metastasis spreads the tumor cells to other parts of the body, which is predominantly through sentinel lymph node. This paper uses an image classification model DenseNet-161 to classify metastaized and normal SLN images. The DenseNets are significantly advantaged over traditional CNNs, by reducing the vanishing-gradient problem, having feature reusage, strengthening feature propagation, having significant reduction in number of parameters and less computation time. The experimental dataset contains Statistical analysis used: The performance of the Densenet-161 is measured by statistic parameter F1 score, training and validation accuracy. Results: The proposed system has generated a training accuracy of 0.9477 and validation accuracy of 0.944, with an F1 score of 0.8406. Conclusions: This model involves extraction of complex information from the medical images dataset, which requires the removal of noise. Even after applying thresholding pre-processing method the noise persists, which requires additional pre-processing before training the model. And by increasing the dataset size through data-augmentation will also improve the accuracy considerably. Keywords: Sentinel Lymph Node, Axillary Lymph Node, Metastasis, Metastasized Breast Cancer, DenseNet, CNN, Diagnostic Pathology, Digital Pathology, WSI Images.

View Article