Enhancing deep convolutional neural network scheme for breast cancer diagnosis with unlabeled data

Sun, Wenqing; Tseng, Tzu-Liang; Zhang, Jianying; Qian, Wei

doi:10.1016/j.compmedimag.2016.07.004

Cited by 229 publications

(120 citation statements)

References 27 publications

Supporting

Mentioning

110

Contrasting

Unclassified

Order By: Relevance

“…Neural network based methods rely heavily on the support of big data. A semi-supervised algorithm has been proposed to deal with a large amount of unlabeled data with CNN approaches [167]. Their approaches using unlabeled data increased the overall accuracy, rather than just using labeled data.…”

Section: Computer Aided Diagnosismentioning

confidence: 99%

Radiological images and machine learning: Trends, perspectives, and prospects

Zhang

Sejdić

2019

Computers in Biology and Medicine

149

View full text Add to dashboard Cite

The application of machine learning to radiological images is an increasingly active research area that is expected to grow in the next five to ten years. Recent advances in machine learning have the potential to recognize and classify complex patterns from different radiological imaging modalities such as x-rays, computed tomography, magnetic resonance imaging and positron emission tomography imaging. In many applications, machine learning based systems have shown comparable performance to human decision-making. The applications of machine learning are the key ingredients of future clinical decision making and monitoring systems. This review covers the fundamental concepts behind various machine learning techniques and their applications in several radiological imaging areas, such as medical image segmentation, brain function studies and neurological disease diagnosis, as well as computeraided systems, image registration, and content-based image retrieval systems. Synchronistically, we will briefly discuss current challenges and future directions regarding the application of machine learning in radiological imaging. By giving insight on how take advantage of machine learning powered applications, we expect that clinicians can prevent and diagnose diseases more accurately and efficiently.

show abstract

Section: Computer Aided Diagnosismentioning

confidence: 99%

Radiological images and machine learning: Trends, perspectives, and prospects

Zhang

Sejdić

2019

Computers in Biology and Medicine

149

View full text Add to dashboard Cite

show abstract

“…In general, biopsy results can be used as benign/malignant labels. Unfortunately, most clinical samples generally do not have biopsy results . In clinical diagnosis, radiologists assign the breast ultrasound images a Breast Imaging, Reporting and Data System (BI‐RADS) rating, which expresses the malignancy probability of breast tumors, as displayed in Table .…”

Section: Introductionmentioning

confidence: 99%

“…Unfortunately, most clinical samples generally do not have biopsy results. [17][18][19] In clinical diagnosis, radiologists assign the breast ultrasound images a Breast Imaging, Reporting and Data System (BI-RADS) rating, which expresses the malignancy probability of breast tumors, as displayed in Table I. However, in general, only samples with high BI-RADS ratings will be biopsied.…”

Section: Introductionmentioning

confidence: 99%

Breast tumor classification through learning from noisy labeled ultrasound images

et al. 2019

View full text Add to dashboard Cite

Purpose To train deep learning models to differentiate benign and malignant breast tumors in ultrasound images, we need to collect many training samples with clear labels. In general, biopsy results can be used as benign/malignant labels. However, most clinical samples generally do not have biopsy results. Previous works have proposed generating benign/malignant labels according to Breast Imaging, Reporting and Data System (BI‐RADS) ratings. However, this approach will cause noisy labels, which means that the benign/malignant labels produced from BI‐RADS diagnoses may be inconsistent with the ground truths. Consequently, deep models will overfit the noisy labels and hence obtain poor generalization performance. In this work, we mainly focus on how to reduce the negative effect of noisy labels when they are used to train breast tumor classification models. Methods We propose an effective approach called noise filter network (NF‐Net) to address the problem of noisy labels when training breast tumor classification models. Specifically, to prevent deep models from overfitting the noisy labels, we propose incorporating two softmax layers for classification. Additionally, to strengthen the effect of clean labels, we design a teacher–student module for distilling the knowledge of clean labels. Results We conduct extensive comparisons with the existing works on addressing noisy labels. Our method achieves a classification accuracy of 73%, with a precision of 69%, recall of 80%, and F1‐score of 0.74. This result is significantly better than those of the existing state‐of‐the‐art works on addressing noisy labels. Conclusions This work provides a means to overcome the label shortage problem in training breast tumor classification models. Specifically, we can generate benign/malignant labels according to the BI‐RADS ratings. Although this approach will cause noisy labels, the design of NF‐Net can effectively reduce the negative effect of such labels.

show abstract

“…The results showed that the diagnostic accuracy was 82.43%. A semi-supervised deep convolutional neural network [23] was developed for breast cancer diagnosis, which used large amount of unlabeled data to improve the accuracy. SVM, KNN, C4.5 and NB models have been implemented and tested on 699 samples for breast cancer detection [24].…”

Section: A Principal Component Analysis (Pca)-cross Entropy Optimizatmentioning

confidence: 99%

“…The Convolutional Neural Network (CNN) [23] model is inspired from ANN and was implemented and tested over 3,158 data samples. The CNN model is composed of neurons with substantial weight and bias can be learned.…”

Section: A Principal Component Analysis (Pca)-cross Entropy Optimizatmentioning

confidence: 99%

A Modified Model Based on Flower Pollination Algorithm and K-Nearest Neighbor for Diagnosing Diseases

Gashti¹

2018

IIUMEJ

View full text Add to dashboard Cite

Exponential growth of medical data and recorded resources from patients with different diseases can be exploited to establish an optimal association between disease symptoms and diagnosis. The main issue in diagnosis is the variability of the features that can be attributed for particular diseases, since some of these features are not essential for the diagnosis and may even lead to a delay in diagnosis. For instance, diabetes, hepatitis, breast cancer, and heart disease, that express multitudes of clinical manifestations as symptoms, are among the diseases with higher morbidity rate. Timely diagnosis of such diseases can play a critical role in decreasing their effect on patients' quality of life and on the costs of their treatment. Thanks to the large data set available, computer aided diagnosis can be an advanced option for early diagnosis of the diseases. In this paper, using a Flower Pollination Algorithm (FPA) and K-Nearest Neighbor (KNN), a new method is suggested for diagnosis. The modified model can diagnose diseases more accurately by reducing the number of features. The main purpose of the modified model is that the Feature Selection (FS) should be done by FPA and data classification should be performed using KNN. The results showed higher efficiency of the modified model on diagnosis of diabetes, hepatitis, breast cancer, and heart diseases compared to the KNN models. ABSTRAK:Pertumbuhan eksponen dalam data perubatan dan sumber direkodkan daripada pesakit dengan penyakit berbeza boleh disalah guna bagi membentuk kebersamaan optimum antara simptom penyakit dan mengenal pasti gejala penyakit (diagnosis). Isu utama dalam diagnosis adalah kepelbagaian ciri yang dimiliki pada penyakit tertentu, sementara ciri-ciri ini tidak penting untuk didiagnosis dan boleh mengarah kepada penangguhan dalam diagnosis. Sebagai contoh, penyakit kencing manis, radang hati, barah payudara dan penyakit jantung, menunjukkan banyak klinikal simptom jelas dan merupakan penyakit tertinggi berlaku dalam masyarakat. Diagnosis tepat pada penyakit tersebut boleh memainkan peranan penting dalam mengurangkan kesan kualiti hidup dan kos rawatan pesakit. Terima kasih kepada set data yang banyak, diagnosis dengan bantuan komputer boleh menjadi pilihan maju menuju ke arah diagnosis awal kepada penyakit. Kertas ini menggunakan Algoritma Flower Pollination (FPA) dan K-Nearest Neighbor (KNN), iaitu kaedah baru dicadangkan bagi diagnosis. Model yang diubah suai boleh mendiagnosis penyakit lebih tepat dengan mengurangkan bilangan ciri-ciri. Tujuan utama model yang diubah suai ini adalah bagi Pemilihan Ciri (FS) perlu dilakukan menggunakan FPA and pengkhususan data perlu dijalankan menggunakan KNN. Keputusan menunjukkan model yang diubah suai lebih cekap dalam mendiagnosis penyakit kencing manis, radang hati, barah payudara dan penyakit jantung berbanding model KNN.

show abstract

Enhancing deep convolutional neural network scheme for breast cancer diagnosis with unlabeled data

Cited by 229 publications

References 27 publications

Radiological images and machine learning: Trends, perspectives, and prospects

Radiological images and machine learning: Trends, perspectives, and prospects

Breast tumor classification through learning from noisy labeled ultrasound images

A Modified Model Based on Flower Pollination Algorithm and K-Nearest Neighbor for Diagnosing Diseases

Contact Info

Product

Resources

About