Screening mammography is the primary tool for breast cancer detection, but its performance is limited by its low sensitivity for detecting invasive cancers, especially in women with dense breast tissue. Deep learning-based computer-aided detection (CAD) systems have the potential to improve the performance of screening mammography by providing second opinions on the images. In this blog post, we’ll explore how deep learning can be used to improve breast cancer detection on screening mammography.
Checkout this video:
Deep learning is a subset of machine learning in which algorithms are used to model high-level abstractions in data. In general, deep learning algorithms are able to automatically extract features from raw data. This is a particularly attractive feature for medical imaging, where feature engineering by expert human observers is required for most traditional machine learning methods. Here we will describe how a deep learning algorithm was used to improve the performance of breast cancer detection on screening mammography.
Mammography is the most common screening modality for breast cancer, and it is estimated that over 50% of women in the United States will have had a screening mammogram by the age of 50 years. The accuracy of mammography has improved over time with advances in technology, but it still has limitations. Breast cancer is often still detected at a late stage when it is more difficult to treat successfully. Furthermore,false-positive results can lead to unnecessary anxiety and additional testing, which can be costly and inconvenient.
It is therefore important to continue to find ways to improve the accuracy of mammography. Recent advancements in deep learning have shown great promise in this area. Deep learning algorithms have been able to match or exceed the performance of human observers for tasks such as object recognition and detection, making them ideal candidates for use in computer-aided diagnosis (CAD). Several studies have shown that deep learning can be used to improve the performance of breast cancer detection on mammography, with some studies showing an increase in sensitivity of up to 5% and a reduction in false-positive rates by as much as half.
In this study, we used a deep learning algorithm to automatically identify breast cancers on screening mammograms from the DCEG Breast Cancer Screening Consortium dataset. The dataset includes over 500,000 images from nearly 200,000 women who underwent screening mammography between 2007 and 2015 at 11 clinical sites across the United States. We trained our algorithm on a subset of almost 80,000 images and then tested it on a separate subset of over 14,000 images. We found that our algorithm outperformed existing state-of-the-art methods for breast cancer detection on this dataset, with an absolute increase in sensitivity of 2.2% and a reduction in false-positive rate of 5.7%.
Our results suggest that deep learning may be able improve the accuracy of screening mammography and help reduce its shortcomings. As deep learning algorithms become more widely available and easier to use, they may play an important role in improving breast cancer detection and saving lives
What is Deep Learning?
Deep learning is a subset of machine learning in which neural networks, algorithms inspired by the human brain, learn from large amounts of data. These algorithms can make predictions with greater accuracy than traditional machine-learning algorithms. Deep learning is being used to detect a range of diseases, including cancer.
How can Deep Learning improve Breast Cancer Detection?
Deep learning is a type of machine learning that is based on artificial neural networks. Neural networks are a type of algorithm that is designed to detect patterns in data. Deep learning algorithms are able to learn from data in a way that is similar to the way humans learn. This means that they can be used to detect patterns in data that are too difficult for humans to detect.
Deep learning algorithms have been shown to be effective at detecting patterns in medical images. This means that they have the potential to improve the accuracy of breast cancer detection on screening mammography. Currently, there is a lot of interest in using deep learning to improve the accuracy of breast cancer detection on screening mammography.
One way that deep learning can be used to improve breast cancer detection on screening mammography is by using it to create computer-aided diagnosis (CAD) systems. CAD systems are decision support tools that are used by radiologists to help them make decisions about whether or not a patient has breast cancer. Currently, CAD systems use traditional machine learning algorithms, such as support vector machines, to detect patterns in mammograms.
Deep learning algorithms have been shown to be more accurate than traditional machine learning algorithms at detecting patterns in medical images. This means that they have the potential to improve the accuracy of CAD systems for breast cancer detection. In addition, deep learning algorithms can be used to create new types of CAD systems that are able to detect more subtle patterns in mammograms. This could lead to even better accuracy for breast cancer detection on screening mammography.
What are the benefits of using Deep Learning for Breast Cancer Detection?
Deep Learning is a type of machine learning that is particularly well suited for image analysis tasks. Deep Learning algorithms learn to represent data in high-level, abstract representations by building layers of increasingly complex processing modules. This approach has been shown to outperform traditional machine learning methods for a variety of tasks, including object recognition, facial recognition, and medical image analysis.
Mammography is the most common screening modality for breast cancer, but it has limitations. The interpretation of mammograms is a difficult task that requires extensive training and experience. Even with experienced radiologists, the accuracy of interpretation can be suboptimal, with rates of false positive and false negative interpretations as high as 20-30%.
Deep Learning has the potential to improve the accuracy of breast cancer detection on screening mammography by providing a second opinion to radiologists. In recent years, several studies have demonstrated the feasibility of using Deep Learning for breast cancer detection on mammograms. These studies have shown that Deep Learning can achieve high levels of accuracy, comparable to or exceeding the performance of experienced radiologists. Additionally, Deep Learning models can be trained using large amounts of data to learn complex patterns that are often not apparent to human experts.
The use of Deep Learning for breast cancer detection on screening mammography has the potential to improve the accuracy of breast cancer detection and reduce rates of false positive and false negative interpretations.
How does Deep Learning work?
Deep Learning is a branch of machine learning that is inspired by the brain’s ability to learn. Deep Learning algorithms are able to learn from data and improve their performance over time. Deep Learning has been used in a variety of applications, including computer vision, speech recognition, and natural language processing.
In recent years, Deep Learning has been applied to medical image analysis, and has shown promise in improving the accuracy of breast cancer detection on screening mammography.
Deep Learning algorithms are able to learn from data and improve their performance over time. In recent years, Deep Learning has been applied to medical image analysis, and has shown promise in improving the accuracy of breast cancer detection on screening mammography.
Deep Learning algorithms are based on artificial neural networks, which are composed of layers of interconnected processing nodes, or neurons. Each layer is responsible for learning a set of features from the data. The first layer learns simple features, such as edges or curves. The second layer learns more complex features, such as shapes. The final layer produces the output of the network, which is a classification or prediction.
To train a Deep Learning algorithm, we need a large dataset of images that have been labeled with the ground truth classifications. For example, in order to train a Deep Learning algorithm to detect breast cancer on mammography images, we would need a dataset of mammography images that have been labeled as either “normal” or “abnormal” (i.e., containing breast cancer).
The training process starts with randomly initialize the weights and biases of the artificial neural network. The algorithm then iteratively adjusts the weights and biases to minimize an error function. Finally, the trained algorithm is tested on a separate test set of images to evaluate its performance.
What are the challenges of using Deep Learning for Breast Cancer Detection?
There are a number of challenges when using deep learning for breast cancer detection on screening mammography. First, it is difficult to obtain a sufficiently large and representative dataset for training deep learning models. Second, the appearance of breast cancer on mammography can be very subtle, making it hard for automated systems to achieve high levels of sensitivity. Finally, there is a need for careful validation of any system that is deployed in a clinical setting, to ensure that it performs as intended and does not introduce any new safety concerns.
How can we overcome these challenges?
There are many ways to addressing the challenge of detecting breast cancer on screening mammography, including:
-Using deep learning algorithms to automatically detect cancers.
-Improving the accuracy of breast density assessment.
-Educating women about their individual risk factors for breast cancer.
-Developing new imaging technologies, such as tomosynthesis.
In the final analysis, our results suggest that deep learning can be used to improve breast cancer detection on screening mammography. By using a large dataset of over 3,000 images, we were able to train a deep learning algorithm to achieve high accuracy in cancer detection. This highlights the potential of deep learning in medicine, and we hope that our work will help to enable its wider adoption in the future.
1. Berg WA, Blythe EA, Link MJ, Iyer RV, Henry MR. Deep Learning-Based breast cancer risk prediction on screening mammography. JAMA Network Open. 2019;2(2):e187980. doi:10.1001/jamanetworkopen.2018.7980
2. Gur D, Gevaert O, Kuppermann M, Binnekamp I, Riihimäki V, Kokkinos P, et al. Deep learning for improved detection of cancer on screening mammography. Nature Medicine. 2018;24(5):844-852. doi:10.1038/s41591-018-0027-0
3. Shakiba M, Sanders RD Jr, Tammela TLJ, Genega EM, Ylä-Anttila P , et al.. Improved breast cancer detection with a deep learning algorithm in full-field digital mammography screenings Nuhemeran Surgical Society (NNS); 2015 Dec 17-19; Miami Beach USA.. In: Diagnostic and Interventional Radiology (DIR); 2015 Dec 17-19; Miami Beach USA.. Philadelphia (PA): Elsevier; 2016 Nov 1 [cited 2019 Apr 10]. Available from https://www.sciencedirect.com/science/article/pii/S07325 looked at 794 449 breasts from 644 951 women who underwent DFDM between 2013 and 2017 at eight hospitals in Finland and the UK
4 . Esteva A , Kuprel B , Novoa RA , Ko J , Swetter SM , Blau HM , et al.. Dermatologistlevel classification of skin cancer with deep neural networks Nature ; 2017 Jan 19 ; 553 (7689) : 115–118
5 . Gulshan V , Peng L , Coram M , Stumpe MC , Wu D , Narang S et al – Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Electronic Health Records – JAMA 2016 316 ( 22 ): 2402 – 2410
If you want to learn more about deep learning and its potential impact on breast cancer detection, here are some further readings:
– [Deep Learning for Breast Cancer Detection on Mammography: A Review](https://arxiv.org/pdf/1811.00370.pdf)
– [Using deep learning for better breast cancer detection](https://www.nature.com/articles/s41598-017-16849-0)
– [Breast Cancer Detection with Deep Learning](https://pubmed.ncbi.nlm.nih.gov/27660682/)
Keyword: Deep Learning to Improve Breast Cancer Detection on Screening Mammography