Breast Carcinoma Detection using WT Segmentation based on Soft Computing

Throughout the past ten years, new research directions have emerged in response to the growing demand for fighting breast cancer early detection and diagnosis. Breast cancer is the most common malignancy worldwide, affecting millions of female patients each year. It also accounts for most women'...

Full description

Saved in:

Bibliographic Details
Published in:	2023 International Conference on Applied Intelligence and Sustainable Computing (ICAISC) pp. 1 - 6
Main Authors:	Manohara, H T, Reddy, Kasireddy Umesh, Kumar, Nanubolu Ravi, Indu, Allu, Vamsi, Kamsala
Format:	Conference Proceeding
Language:	English
Published:	IEEE 16-06-2023
Subjects:	Breast cancer Computational modeling Convolutional Neural Network Feature extraction Filters Image resolution Image segmentation Mammographic images MATLAB Measurement Segmentation Training
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	Throughout the past ten years, new research directions have emerged in response to the growing demand for fighting breast cancer early detection and diagnosis. Breast cancer is the most common malignancy worldwide, affecting millions of female patients each year. It also accounts for most women's cancer-related deaths. When used to classify tumors using an image database, convolutional neural networks show promising results. The planned project primarily focuses on the identification of breast cancer using a soft computing approach that makes use of feature extractors that have been trained on millions of different images and that use pre-trained models as feature extractors. The proposed research examines how pre-trained models categorize breast tumors when used with mammographic pictures. Also, the proposed work suggests a shallow custom convolutional neural network that performs better than training models in a number of performance-related metrics. Using bespoke Wiener Threshold (WT) segmentation, the suggested model's impressive accuracy of 86.23% was attained in 36m25s, which is less time than the existing Gaussian threshold segmentation computation time.
DOI:	10.1109/ICAISC58445.2023.10200721