Representation learning for mammography mass lesion classification with convolutional neural networks

Arévalo, John; González, Fabio A.; Ramos-Pollán, Raúl; Oliveira, José Luís; López, Miguel

doi:10.1016/j.cmpb.2015.12.014

Cited by 390 publications

(200 citation statements)

References 16 publications

Supporting

Mentioning

192

Contrasting

Unclassified

Order By: Relevance

“…Results using 9-fold cross validation on a private database of 33 mammograms reached an area under the ROC curve with a value of 0.83. Similarly in [38], the authors used a hybrid convolutional neural network (CNN) classifier along with an enhancement step to classify 426 benign and 310 malignant breast masses. On this dataset, they obtained 0.82 score for the area under the ROC curve (AUC).…”

Section: Deep-learning Methodsmentioning

confidence: 99%

DeepCAD: A Computer-Aided Diagnosis System for Mammographic Masses Using Deep Invariant Features

Abbas

2016

Computers

View full text Add to dashboard Cite

Abstract:The development of a computer-aided diagnosis (CAD) system for differentiation between benign and malignant mammographic masses is a challenging task due to the use of extensive pre-and post-processing steps and ineffective features set. In this paper, a novel CAD system is proposed called DeepCAD, which uses four phases to overcome these problems. The speed-up robust features (SURF) and local binary pattern variance (LBPV) descriptors are extracted from each mass. These descriptors are then transformed into invariant features. Afterwards, the deep invariant features (DIFs) are constructed in supervised and unsupervised fashion through multilayer deep-learning architecture. A fine-tuning step is integrated to determine the features, and the final decision is performed via softmax linear classifier. To evaluate this DeepCAD system, a dataset of 600 region-of-interest (ROI) masses including 300 benign and 300 malignant masses was obtained from two publicly available data sources. The performance of DeepCAD system is compared with the state-of-the-art methods in terms of area under the receiver operating characteristics (AUC) curve. The difference between AUC of DeepCAD and other methods is statistically significant, as it demonstrates a sensitivity (SN) of 92%, specificity (SP) of 84.2%, accuracy (ACC) of 91.5% and AUC of 0.91. The experimental results indicate that the proposed DeepCAD system is reliable for providing aid to radiologists without the need for explicit design.

show abstract

Section: Deep-learning Methodsmentioning

confidence: 99%

DeepCAD: A Computer-Aided Diagnosis System for Mammographic Masses Using Deep Invariant Features

Abbas

2016

Computers

View full text Add to dashboard Cite

show abstract

“…The scheme was evaluated on a total of 600 mammographic images from DDSM database and the results indicate superior performances for the two-stage scheme over other single stage models. Additional work that integrates CNN with linear SVM was discussed by Arevalo et.al [6]. Extreme Learning Machine (ELM) based model was presented by Xie et.al [7].…”

Section: Related Workmentioning

confidence: 99%

Breast Cancer Classification Enhancement Based on Entropy Method

Sama¹,

Baneamoon²

2017

IJEACS

View full text Add to dashboard Cite

Abstract-Breast cancer is the most common form of cancer diseases among women and the second leading cause of cancer deaths worldwide. However, due to some limitations in mammography, it is difficult to classify a suspicious mass in the breast as malignant (cancerous) or benign. This paper attempts to classify the mammographic masses with high accuracy by combining entropy method with evolutionary algorithm (EA) and fuzzy logic. EA and fuzzy logic are applied at training phase for parameters tuning, however, entropy method is applied at training phase and localization phase, where at training phase entropy enhances indicator of fuzziness, while at mass phase entropy enhances EA for the classification of masses in mammogram images. The proposed method is evaluated by experimenting a number of the benchmark Mini-MIAS databases and the results shows better and more effective identification.

show abstract

“…Furthermore, we plan to use a deep learning approach to find discriminant features and combine them with handcrafted features (e.g., local patterns from LQP operators). Although several studies [22][23][24][25][26][27][28][29] have investigated the use of deep learning in breast density classification, to the best of our knowledge, the separation was only based on two-or three-class classification (none of the deep learning based approaches has been applied to four-class classification). In fact, no study has been conducted combining non-handcrafted and handcrafted features in breast density classification.…”

Section: Future Workmentioning

confidence: 99%

“…Based on the 100 mammograms used as a testing dataset, the authors reported the correlation coefficient between the breast estimates by the CNN and those by the expert's manual measurement to be 0.96. Arevalo et al [26] proposed a hybrid CNN method to segment breast mass and reported 0.83 in the area under the curve (AUC) value. Qui et al [27] proposed three pairs of convolution-max-pooling layers that contain 20, 10, and five feature maps to predict short-term breast cancer risk and achieved 71.4%.…”

Section: Literature Reviewmentioning

confidence: 99%

Breast Density Classification Using Local Quinary Patterns with Various Neighbourhood Topologies

et al. 2018

View full text Add to dashboard Cite

This paper presents an extension of work from our previous study by investigating the use of Local Quinary Patterns (LQP) for breast density classification in mammograms on various neighbourhood topologies. The LQP operators are used to capture the texture characteristics of the fibro-glandular disk region (FGD roi ) instead of the whole breast area as the majority of current studies have done. We take a multiresolution and multi-orientation approach, investigate the effects of various neighbourhood topologies and select dominant patterns to maximise texture information. Subsequently, the Support Vector Machine classifier is used to perform the classification, and a stratified ten-fold cross-validation scheme is employed to evaluate the performance of the method. The proposed method produced competitive results up to 86.13% and 82.02% accuracy based on 322 and 206 mammograms taken from the Mammographic Image Analysis Society (MIAS) and InBreast datasets, which is comparable with the state-of-the-art in the literature.

show abstract

Representation learning for mammography mass lesion classification with convolutional neural networks

Cited by 390 publications

References 16 publications

DeepCAD: A Computer-Aided Diagnosis System for Mammographic Masses Using Deep Invariant Features

DeepCAD: A Computer-Aided Diagnosis System for Mammographic Masses Using Deep Invariant Features

Breast Cancer Classification Enhancement Based on Entropy Method

Breast Density Classification Using Local Quinary Patterns with Various Neighbourhood Topologies

Contact Info

Product

Resources

About