Improving Computer-aided Detection using Convolutional Neural Networks and Random View Aggregation

By Holger R. Roth, Le Lu, Jiamin Liu, Jianhua Yao, Ari Seff, Kevin M. Cherry, Lauren Kim, Ronald M. Summers. 2015

585 citations recorded by the source corpus.

Automated computer-aided detection (CADe) has been an important tool in clinical practice and research. State-of-the-art methods often show high sensitivities at the cost of high false-positives (FP) per patient rates. We design a two-tiered coarse-to-fine cascade framework that first operates a candidate generation system at sensitivities ~ 100% of but at high FP levels. By leveraging existing CADe systems, coordinates of regions or volumes of interest (ROI or VOI) are generated and function as input for a second tier, which is our focus in this study. In this second stage, we generate 2D (two-dimensional) or 2.5D views via sampling through scale transformations, random translations and rotations. These random views are used to train deep convolutional neural network (ConvNet) classifiers. In testing, the ConvNets assign class (e.g., lesion, pathology) probabilities for a new set of random views that are then averaged to compute a final per-candidate classification probability. This second tier behaves as a highly selective process to reject difficult false positives while preserving high sensitivities. The methods are evaluated on three data sets: 59 patients for sclerotic metastasis detection, 176 patients for lymph node detection, and 1,186 patients for colonic polyp detection. Experimental results show the ability of ConvNets to generalize well to different medical imaging CADe applications and scale elegantly to various data sets. Our proposed methods improve performance markedly in all cases. Sensitivities improved from 57% to 70%, 43% to 77%, and 58% to 75% at 3 FPs per patient for sclerotic metastases, lymph nodes and colonic polyps, respectively.

Subfields: Artificial Intelligence, Radiology, Nuclear Medicine and Imaging

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. P. Heng
2. Qi-Yan Dou
3. B. van Ginneken
4. Jing Qin
5. Hao Chen
6. Lequan Yu
7. Francesco Ciompi
8. Ronald M. Summers

In Radiology, Nuclear Medicine and Imaging

1. P. Heng
2. B. van Ginneken
3. Hao Chen
4. Qi-Yan Dou
5. Lequan Yu
6. Ronald M. Summers
7. Jing Qin
8. Dinggang Shen

Citations in the corpus

• A survey on deep learning in medical image analysis (2017, methodology): “The incorpora- ion of contextual or 3D information is also handled using multi- tream CNNs ( Section 2.4.2 , for example by Barbu et al., 2016 and oth et al., 2016b ).”
• Deep Learning for Health Informatics (2017, background): “Implicit features could determine fibroids and polyps [1], and characterize irregularities in tissue morphology such as tumors [2].”
• Deep learning in bioinformatics (2016, methodology): “[125] applied CNNs to three different CT image datasets to classify sclerotic metastases, lymph nodes and colonic polyps.”
• Deep learning in medical imaging and radiation therapy. (2018, background): “166,168 More recently, 3D CNNs that use 3D convolution kernels are successfully replacing 2D CNNs for volumetric data.”
• A review of deep learning in medical imaging: Imaging traits, technology trends, case studies with progress highlights, and future promises (2020, background): “Studies have assessed abdominal lymph nodes on CT in general and for specific diseases, such as prostate cancer [35], [199], [200].”
• Deep-Learning Convolutional Neural Networks Accurately Classify Genetic Mutations in Gliomas (2018, background): “24-26 Recent use of CNNs has started to yield promising results in multiple medical imaging disciplines, including the detection of pulmonary nodules, 27 colon cancer, 28 and cerebral microbleeds.”
• A deep learning framework for supporting the classification of breast lesions in ultrasound images (2017, unknown): “problem of abnormality detection (CADe)[56, 57, 58].”
• Road crack detection using deep convolutional neural network (2016, methodology): “Inspired by the method proposed in [11], the probability of a point can be calculated by averaging probability {P1, .”
• Road Crack Detection Using Deep Convolutional Neural Network and Adaptive Thresholding (2019, background): “Furthermore, some machine learning-based crack detection approaches [30]–[35] have been proposed in recent years, and the features produced by neural network are very likely to replace the local features utilized in traditional”
• Breast cancer detection using deep convolutional neural networks and support vector machines (2019, methodology): “There are many CNN architectures such as CiFarNet ( Krizhevsky, 2009 ; Roth et al., 2016 ), AlexNet ( Krizhevsky, Sutskever & Hinton, 2012 ), GoogLeNet ( Szegedy et al., 2015 ), The AlexNet architecture achieved significantly better performance over the other deep learning methods for ImageNet…”
• Automatic segmentation of nine retinal layer boundaries in OCT images of non-exudative AMD patients using deep learning and graph search. (2017, methodology): “In this paper, we use a modified Cifar-CNN [38,57] architecture.”
• GAN-based Synthetic Medical Image Augmentation for increased CNN Performance in Liver Lesion Classification (2018, methodology): “T HE greatest challenge in the medical imaging domain is how to cope with the small datasets and limited amount of annotated samples [1]–[5], especially when employing supervised machine learning algorithms that require labeled data and larger training examples.”
• Acute Lymphoblastic Leukemia Detection and Classification of Its Subtypes Using Pretrained Deep Convolutional Neural Networks (2018, methodology): “So lot of groups in medical imaging deployed DCNNs and they were able to achieve good accuracy.(10-15) As per our knowledge for leukemia detection only, Vogado et al(16) investigated a pretrained CNN for the features extraction and used ensemble classifiers for classification of normal and blast cel”
• Deep learning in radiology: an overview of the concepts and a survey of the state of the art with focus on MRI (2018, methodology): “thors performed detection of 3D anatomy in chest CT images by processing data slice by slice in one direction. Combining output from dierent planes was performed in several studies. Most of the them [102, 103, 104] used orthogonal planes of MRI and CT images performing detection in each direction se”