Hyperparameter Optimization of Deep Learning Networks for Classification of Breast Histopathology Images

Abstract

After tumor detection in their breasts, women typically fear mastectomy; this affects curative care outcomes. Most tumors are benign. After resection and pathological examination, because of advances in medicine and treatment, the success rate of early breast cancer treatment can reach 60 to 90%. An accurate assessment of tumor extent is essential. In this study, a novel method of hyperparameter optimization of deep learning networks was proposed to classify tumors as malignant or benign. When setting hyperparameters in deep learning networks, most users use trial and error to determine them. In our experiments, the Taguchi method was used to select the impact factors. The orthogonal table design was used to conduct experiments. Then, the best combination of parameters was determined and significant impact factors were analyzed. The Breast Cancer Histopathological Database was used for analysis. This database was built in collaboration with the P&D Laboratory and contains 2480 benign and 5429 malignant samples. The experimental results showed that the proposed method obtained a high accuracy of 83.19%.

Addressing Uncertainty in Imbalanced Histopathology Image Classification of HER2 Breast Cancer: An Interpretable Ensemble Approach With Threshold Filtered Single Instance Evaluation (SIE)

“…DL has been applied heavily to analyze tumor detection, grading, subtypes, predictions, etc [18]. Lin et al [19] acquired 77.50% accuracy using single-layer CNN. Darken et al [20] proposed a better LeNet-5 (RMSprop) method and achieved 82.58% accuracy.…”

Application of Deep Learning in Histopathology Images of Breast Cancer: A Review

“…Propose a new deep architecture based on self-integration to leverage semantic information from annotated images and explore information hidden in unlabeled data [140] Propose an analysis and synthesis model learning method with novel algorithms and search strategies to classify images more effectively [141][142][143][144][145][146][147][148][149][150] Propose a set of training techniques and use image processing techniques to improve the performance of CNN-based models in breast cancer classification [143,[151][152][153][154][155][156][157] Deep residual network (ResNet) Present a deep neural network which performs representation learning and cell nuclei recognition in an end-to-end manner [158] Propose an automatic multiclassification method for breast cancer histopathological images based on metastasis learning [159] Present a method that employs a convolutional neural network for detecting tumor on entire-slide images [59,130,136,160] Propose a breast cancer multiclassification method using a proposed deep learning model [106,113,137,[161][162][163][164][165][166][167][168] Thus, through research and analysis, we found that the classifier proposed by [110] first predicts the class label of each input patch by OPOD technique and then predicts the whole-image label by APOD technique. At the same time, the number o...…”

Understanding the Mechanism of Deep Learning Frameworks in Lesion Detection for Pathological Images with Breast Cancer

“…In recent years, computer-aided technologies have been widely adopted to achieve automated inspection for health care delivery, having a wide range of applications from scalp inspection (1) to lung nodule detection (2) in radiology and lesion classification (3) in histopathology. However, biomedical image analysis is a complex task that relies on highly skilled domain experts, such as radiologists and pathologists.…”