Going deeper: magnification‐invariant approach for breast cancer classification using histopathological images

Abstract

Breast cancer has the highest fatality for women compared with other types of cancer. Generally, early diagnosis of cancer is crucial to increase the chances of successful treatment. Early diagnosis is possible through physical examination, screening, and obtaining a biopsy of the dubious area. In essence, utilizing histopathology slides of biopsies is more efficient than using typical screening methods. Nevertheless, the diagnosing process is still tiresome and is prone to human error during slide preparation, such as when dyeing and imaging. Therefore, a novel method is proposed for diagnosing breast cancer into benign or malignant in a magnification-specific binary (MSB) classification. Besides, the introduced method classifies each type into four subclasses in a magnification-specific multicategory (MSM) fashion. The proposed method involves normalizing the hematoxylin and eosin stains to enhance colour separation and contrast. Then, two types of novel features-deep and shallow features-are extracted using two deep structure networks based on DenseNet and Xception. Finally, a multi-classifier method based on the maximum value is utilized to achieve the best performance. The proposed method is evaluated using the BreakHis histopathology data set, and the results in terms of diagnostic accuracy are promising, achieving 99% and 92% in terms of MSB and MSM, respectively, compared with recent state-of-the-art methods reported in the survey conducted by Benhammou on the BreakHis data set using deep learning and texture-based models. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

Deep Learning in Cancer Diagnosis and Prognosis Prediction: A Minireview on Challenges, Recent Trends, and Future Directions

“…As data augmentation methods, they used flipping, shifting, change of brightness, and rotation achieving 98.51% accuracy. The authors in [ 105 ] used a combination of DenseNet and Xception transfer learning architectures for benign/malignant binary and magnification-specific multiclass classification tasks. They used the BreakHis dataset achieving an accuracy of 99% and 92% on binary and multiclass classification tasks, respectively, while deploying stain normalization for preprocessing of images.…”

A Review of Computer-Aided Expert Systems for Breast Cancer Diagnosis

“…The malignant type of breast tumour consists of ductal carcinoma (DC), lobular carcinoma (LC), mucinous carcinoma (MC), and papillary carcinoma (PC). This dataset is the most used dataset by many researchers for CAD breast cancer in histopathology images [ 11 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 ]. This dataset can be obtained from (accessed on 16 March 2021).…”

“…Ho proposed a random subspace classifier, in which a random feature subset is picked up from the original dataset for training each classifier; a voting scheme is then applied to produce a unique output from the from all the outputs in the combined classifiers [ 132 ]. Alkassar et al applied an ensemble classifier that chooses the maximum score of prediction that includes a combination of decision tree, linear and quadratic discriminant, logistic regression, naive Bayes, SVM, and KNN [ 22 ].…”

“…The authors adopted this architecture that combines a bidirectional RNN (Bi-RNN) that handles two sources of information and LSTM for classifying breast cancer [ 28 ]. Alkassar et al used the Xception and DeseNet to perform extraction on shallow and deep features from breast histology images [ 22 ]. Araujo et al combined a CNN model to extract features and a SVM classifier to perform breast cancer classification [ 31 ].…”

“…In medical imaging diagnostic tests, sensitivity and specificity parameters are widely used to evaluate performance [ 40 ]. Other commonly used metrics for diagnosis evaluation are F1-measure [ 19 , 23 , 59 , 84 , 148 , 151 ], precision [ 19 , 59 , 60 , 148 ], accuracy [ 11 , 18 , 19 , 20 , 21 , 22 , 24 , 25 , 26 , 27 , 28 , 29 , 31 , 53 , 60 , 83 , 128 , 130 , 131 ] and receiver operating characteristics (ROC). The parameters are mostly calculated depending on the true positive (TP), true negative (TN), false positive (FP), and false negative (FN), which are the numbers of pixels corresponding to the parameters.…”