Architectures Based on Deep Learning for the Detection of Invasive Ductal Carcinoma
Isha Gupta, Sheifali Gupta, Swati Singh · ECS Transactions · 2022
Image processing techniques have improved dramatically in recent years to help pathologists identify cancer cells. Like convolutional neural networks (CNNs), deep learning methods are increasingly used for imaging processes and analysis in histopathology images. This research aims to demonstrate the detection of histopathological pictures linked with the prediction of invasive ductal carcinomas (IDC) and non-IDC in the breasts. A complex problem is the detection of IDC in histopathological images, as cancer includes minor entities with a variety of shapes that can readily be confused with other objects or facts in the picture. As a result, the suggested research recommends three distinct CNN architectures for identifying IDC using histopathological images, referred to as 10-layer, 19-layer, and 20-layer convolutional neural networks, respectively. Excellent values of sensitivity, precision, and low classification error rate have been achieved to detect IDC in histopathology images using deep layer-convolutional neural networks. Using 19 layer-convolutional neural networks, the efficiency obtained was 87%, revealing improved results for deep layer convolutional neural network architecture.