Publication | Closed Access
Classification of Histopathological Images for Early Detection of Breast Cancer Using Deep Learning
71
Citations
0
References
2021
Year
Convolutional Neural NetworkMedical Image SegmentationEngineeringMachine LearningDigital PathologyDiagnosisPathologyImage AnalysisPattern RecognitionBreast ImagingHistopathological ImagesEarly DetectionRadiologyDermoscopic ImageMedical ImagingHistopathologyComputational PathologyDeep LearningMedical Image ComputingComputer VisionBiomedical ImagingMultimodal ImagingComputer-aided DiagnosisBreast CancerClinical Image AnalysisDiagnostic AccuracyMedicineMedical Image Analysis
ABSTRACT Breast cancer is one of the most common and deadly types of cancer that develops in the breast tissue of women worldwide. This is why researchers and experts are interested in developing a computer-aided diagnostic system (CAD) for diagnosing histopathological images of breast cancer. CAD has contributed to increasing the diagnostic accuracy of the biopsy tissue using eosin stained and hematoxylin images. Most CAD systems have used traditional methods to extract handcrafted features, which are imprecise in diagnosis and time-consuming. The diagnostics by both CAD and the calculations are used to reduce the pathologist’s workload and improve accuracy. In this study, the proposed convolutional neural network (AlexNet) approach to extract the deepest features from the BreaKHis dataset to diagnose breast cancer as either benign or malignant. In the current proposal, the study performed four experiments according to a magnification factor (40X, 100X, 200X and 400X). Each experiment contains 1407 images. The network was trained and validated on 80 % tissue images and 20 % for testing. The proposed system obtained accuracy, sensitivity, specificity, and AUC, 95 %, 97 %, 90 % and 99.36 % respectively.