Breast Cancer Classification from Histopathological Images Using Transfer Learning and Deep Neural Networks

Abstract

Early diagnosis of breast cancer is the most reliable and practical approach to managing cancer. Computer-aided detection or computeraided diagnosis is one of the software technology designed to assist doctors in detecting or diagnose cancer and reduce mortality via using the medical image analysis with less time. Recently, medical image analysis used Convolution Neural Networks to evaluate a vast number of data to detect cancer cells or image classification. In this thesis, we implemented transfer learning from pre-trained deep neural networks ResNet18, Inception-V3Net, and ShuffleNet in terms of binary classification and multiclass classification for breast cancer from histopathological images.We use transfer learning with the fine-tuned network results in much faster and less complicated training than a training network with randomly initialized weights from scratch. Our approach is applied to image-based breast cancer classification using histopathological images from public dataset BreakHis. The highest average accuracy achieved for binary classification of benign or malignant cases was 97.11% for ResNet 18, followed by 96.78% for ShuffleNet and 95.65% for Inception-V3Net. In terms of the multiclass classification of eight cancer classes, the average accuracies for pre-trained networks are as follows. ResNet18 achieved 94.17%, Inception-V3Net 92.76% and ShuffleNet 92.27%. iv AcknowledgementFirst and foremost, all praises to Allah for blessing, protecting, and guiding me throughout my studies. I could never have accomplished this without my faith.Second, I would like to express my sincere gratitude to my supervisor Professor Adam Krzyzak for his continuous support, immense knowledge, continuous motivation, and patience.His unique personality as a supervisor and friend is the main reason behind the success of this research. The objectives of the research would not have achieved without the professional and experienced guidance and support of my supervisor.I also extend my thanks to members of the examining committee, including Dr. Thomas Fevens and Dr. Tristan Glatard, for critically evaluating my thesis and providing me with valuable comments about my research. I would like to extend my thanks Halina Monkiewicz, the program advisor of Computer Science and Software Engineering, for her support, patience and motivation.Third, I would like to express my sincere gratitude to the financial support from the Government of Saudi Arabia under the scholarship of Saudi Electronic University, which enabled me to undertake my studies, and I was lucky to obtain this opportunity. I am forever thankful to the Saudi Cultural Bureau and the Embassy of Saudi Arabia in Canada for their continued support and motivation during my journey.A heartfelt deepest gratitude expresses to my first teachers (my dear mother and my father), my brothers, and my sisters for their unconditional love, prayers, and support. Without them, this journey would not have been possible, and to them, I dedicate this milestone.Last but never least, to my l...

Volumetric Model Genesis in Medical Domain for the Analysis of Multimodality 2-D/3-D Data Based on the Aggregation of Multilevel Features

“…In total, 660 Xray images were used to validate the method. Subsequently, Aloyayri et al [14] utilized the strength of transfer learning in breast cancer classification using histopathological images. Three different CNN architectures were trained to classify data samples as either benign or malignant.…”

“…Consequently, various state-of-the-art computer-aided diagnosis (CAD) methods have been proposed in the literature that utilize the power of AI in medical data analysis and enable effective diagnostic decisions [8]- [11]. Among the different AI methods, a subset of deep learning (DL) algorithms has received special attention owing to its remarkable performance, particularly in the case of visual data analysis [12]- [14]. Such DL-driven CAD methods mimic the processing of the human brain and deliver accurate diagnostic results, similar to those of medical experts.…”

Revolutionizing Breast Cancer Diagnosis: A Concatenated Precision through Transfer Learning in Histopathological Data Analysis

“…Their results demonstrate the effectiveness of pretrained models for accurate classification. The thesis [15] applied transfer learning using ResNet18, Inception-V3Net, and ShuffleNet to perform binary and multiclass classification of breast cancer from histopathological images. Transfer learning offers faster and simpler training compared to training networks from scratch.…”

“…Transfer learning offers faster and simpler training compared to training networks from scratch. The study used the BreakHis dataset, achieving high accuracy: 97.11% for ResNet18 in binary classification, and 94.17% for ResNet18 in multiclass classification, with other models also performing well [15].…”

Modality Specific CBAM-VGGNet Model for the Classification of Breast Histopathology Images via Transfer Learning

“…The last layers of the models are fined tuned and trained on the BreakHis images. Authors achieve the highest accuracy of 98.73% [31]. HM Ahmad et al used 240 training and 20 test images and classify them into four classes.…”