An Intelligent Model for Improved Breast Cancer Prognosis
C O Iloghalu, Eze U F, Adetokunbo MacGregor John-Otumu, Obi Chukwuemeka Nwokonkwo, Emmanuel Osaze Oshoiribhor, Jude Chukwuma Onyeakazi, Aliga Paul Aliga, Stanley Adiele Okolie, Euphemia Chioma Nwokorie, Leonard Chukwualuka Nnadi · International Journal of Electronics and Communication Engineering · 2023
This research suggests developing a deep learning model using customized CNN to categorize and predict breast cancer in a timely period. The model utilizes a large dataset of breast cancer images obtained from Kaggle, an online research repository. Pre-processing techniques were applied to the images to eliminate noise, such as shadows on the images, and resize the images to lessen the high computation cost. The dataset was separated into training set 80% (48, 852) and test set 20% (16, 284). CNN was employed to mine meaningful features from the images and to classify them based on predefined criteria, assessing the presence and severity of breast cancer. Additionally, the model could provide treatment recommendations depending on the patient's health account and other pertinent aspects. The model's performance was evaluated using a confusion matrix, revealing a 95% accuracy rate, 100% recall value, 90% precision value, and 95% F1 score. The classifier's AUC value was 88%, indicating high reliability for breast cancer prognosis. The proposed methodology may significantly increase diagnostic speed and accuracy, resulting in earlier detection and better patient outcomes.