Nodule Candidates Research Articles

An attentive and adaptive 3D CNN for automatic pulmonary nodule detection in CT image

Automatic pulmonary nodule detection is the most crucial technology in early diagnosis of lung cancer and early treatment. However, the various nodule type, shape and size, especially the diameter of lung nodules (ranging from 3 mm to 30 mm), make the lung nodule detection results with high false positives. It significantly affects the detection performance of lung nodules. In this paper, an adaptive and attentive 3D Convolutional Neural Network (CNN) is proposed for automatic pulmonary nodule detection, which contains two parts: the candidate nodule detection and false positive reduction. In the first stage, the globule, spital and fine-grained information of focal nodules are scratched by the high-resolution fused attention module in the proposed method. In the second stage, an adaptive 3D CNN structure is designed to further reduce the false positives, which extracts the multilevel contextual information via an adaptive 3D convolution kernel. Extensive experiments are conducted on publicly available LUNA16. The results demonstrate that the proposed method can increase the sensitivity and decrease the false positives rate for automated pulmonary nodule detection effectively.

Detection and diagnosis of lung cancer using region based convolutional neural network model on computed tomography images

Computed tomography is one of the most sensitive imaging techniques for the segmentation of lung cancer. Since lung cancer has the high mortality rate among all the cancers so it becomes so important to detect and diagnose the cancer at beginning stage itself. Segmentation is an important step in medical image analysis and classification for radiological evaluation or computer aided diagnosis. The CAD (Computer Aided Diagnosis) of lung CT generally first segment the area of interest (lung) and then analyse the separately obtained area for nodule detection in order to diagnosis the disease. In this work, implemented a semantic segmentation method using region based convolutional neural network in order to segment the lung nodule part separately from the lung which will enhance the chances of detection of lung cancer and will optimize the extraction of candidate nodules. R-CNN is developed to be used to segment lung regions. The network consists of total 4 convolution layers and 2 max pooling layers. This network has provided an accuracy of 98.28%.

A improved detection method for lung nodule based on multi‐scale 3Dconvolutional neural network

AbstractLow detection sensitivity and high false positives are two main challenges from traditional lung nodule detection model. In order to improve reliability of detection model, we proposed a multi‐scale three‐dimensional convolutional neural network lung nodule detection method which can simultaneously detect lung nodules and reduce false positives. First, a 3D UNet++‐like architecture with an encoding‐decoding structure is regarded as a feature extraction network and combines it with the region proposal network. The multi‐feature fusion method can fully learn features at different levels through the cross‐layer connection between the encoder and the decoder. Second, the layers of the 3D UNet++‐like architecture are connected by residual blocks, which enhance feature reuse and speeds up the convergence speed of the network. Finally, the three input sizes are input into the 3D neural network, and their classification results are merged, thus to find the final nodule determination result. Based on the LUNA16 dataset, the experiment result shows that average sensitivity is 87.3% and be increased 7.8% than the UNet++ network in our model. And if the number of candidate nodules is 48, the sensitivity is as high as 96.2%, it increases 8.1% than the VGG16 network. Obviously, that the improved model can significantly improve detection sensitivity and reduce false positives, which can provide a valuable theoretical reference for clinical medicine.

A novel method for lung nodule detection in computed tomography scans based on Boolean equations and vector of filters techniques

This work presents a novel method that uses a Vector of Pre-processing Filters combined with simple relational and Boolean equations for pulmonary nodule detection. To isolate nodules from other lung structures, we propose a 16 filter scheme endowed with multiscale median masks, statistical-based thresholds, and 3-D morphological operations. In the Boolean False Positive Reduction stage, relational and Boolean equations select nodule candidates from pre-processing filters using our descriptor with 20 attributes. Finally, a Convolutional Neural Network (CNN) classifies the remaining structures into nodule or no nodule. The method reached 92.75% sensitivity for an average of 8 false positives per exam using all exams in the public Lung Image Database Consortium (LIDC) with a slice thickness of less than 2 mm that contains lesions larger than 3 mm marked “nodule” by at least 3 radiologists. For a less consensual gold standard, our method reaches the highest sensitivity levels among listed results.

Pulmonary nodules detection based on multi-scale attention networks

Pulmonary nodules are the main manifestation of early lung cancer. Therefore, accurate detection of nodules in CT images is vital for lung cancer diagnosis. A 3D automatic detection system of pulmonary nodules based on multi-scale attention networks is proposed in this paper to use multi-scale features of nodules and avoid network over-fitting problems. The system consists of two parts, nodule candidate detection (determining the locations of candidate nodules), false positive reduction (minimizing the number of false positive nodules). Specifically, with Res2Net structure, using pre-activation operation and convolutional quadruplet attention module, the 3D multi-scale attention block is designed. It makes full use of multi-scale information of pulmonary nodules by extracting multi-scale features at a granular level and alleviates over-fitting by pre-activation. The U-Net-like encoder-decoder structure is combined with multi-scale attention blocks as the backbone network of Faster R-CNN for detection of candidate nodules. Then a 3D deep convolutional neural network based on multi-scale attention blocks is designed for false positive reduction. The extensive experiments on LUNA16 and TianChi competition datasets demonstrate that the proposed approach can effectively improve the detection sensitivity and control the number of false positive nodules, which has clinical application value.

A radiomics approach for lung nodule detection in thoracic CT images based on the dynamic patterns of morphological variation.

To propose and evaluate a set of radiomic features, called morphological dynamics features, for pulmonary nodule detection, which were rooted in the dynamic patterns of morphological variation and needless precise lesion segmentation. Two datasets were involved, namely, university hospital (UH) and LIDC datasets, comprising 72 CT scans (360 nodules) and 888 CT scans (2230 nodules), respectively. Each nodule was annotated by multiple radiologists. Denoted the category of nodules identified by at least k radiologists as ALk. A nodule detection algorithm, called CAD-MD algorithm, was proposed based on the morphological dynamics radiomic features, characterizing a lesion by ten sets of the same features with different values extracted from ten different thresholding results. Each nodule candidate was classified by a two-level classifier, including ten decision trees and a random forest, respectively. The CAD-MD algorithm was compared with a deep learning approach, the N-Net, using the UH dataset. On the AL1 and AL2 of the UH dataset, the AUC of the AFROC curves were 0.777 and 0.851 for the CAD-MD algorithm and 0.478 and 0.472 for the N-Net, respectively. The CAD-MD algorithm achieved the sensitivities of 84.4% and 91.4% with 2.98 and 3.69 FPs/scan and the N-Net 74.4% and 80.7% with 3.90 and 4.49 FPs/scan, respectively. On the LIDC dataset, the CAD-MD algorithm attained the sensitivities of 87.6%, 89.2%, 92.2%, and 95.0% with 4 FPs/scan for AL1-AL4, respectively. The morphological dynamics radiomic features might serve as an effective set of radiomic features for lung nodule detection. • Texture features varied with such CT system settings as reconstruction kernels of CT images, CT scanner models, and parameter settings, and so on. • Shape and first-order statistics were shown to be the most robust features against variation in CT imaging parameters. • The morphological dynamics radiomic features, which mainly characterized the dynamic patterns of morphological variation, were shown to be effective for lung nodule detection.

Gauze blotting technique: a novel method to identify parathyroid glands during thyroid surgery without tissue damage.

Hypoparathyroidism is a major complication of thyroid surgery. To avoid this complication, visual identification of the parathyroid glands is essential. However, its effectiveness depends heavily on the surgeon's expertise. Here, we describe a novel method, the gauze blotting technique, to immunochemically identify the parathyroid glands during thyroid surgery. Twenty-three patients who underwent thyroid lobectomy were enrolled in this study; 16 and 7 had benign and malignant thyroid diseases, respectively. After visually identifying candidate nodules for the parathyroid gland, a piece of dry gauze (5 mm × 10 mm) was applied to each tissue until it was moistened by exudates from the tissue. Pieces of gauze were also applied to the thyroid gland and adipose tissue located away from the candidate nodules. The gauze was immersed in saline, and the intact PTH (i-PTH) level of the supernatant was measured. The median PTH level for the parathyroid glands was 1,060 pg/mL, which was significantly higher than that for the thyroid gland (34 pg/mL) and adipose tissue (28 pg/mL) (p < 0.001). The cut-off value to distinguish the parathyroid gland from other tissues was 68 pg/mL with a positive predictive value, negative predictive value, sensitivity, and specificity of 84.6%, 88.8%, 86.8%, and 86.7%, respectively. A value ≥250 pg/mL yielded a 100% positive predictive value. Our novel gauze blotting technique can identify the parathyroid glands without damaging tissues during thyroid surgery.

Pulmonary Nodule Detection Using 3-D Residual U-Net Oriented Context-Guided Attention and Multi-Branch Classification Network

Accurate detection of pulmonary nodules on chest computed tomography scans is crucial to early diagnosis of lung cancer. To address the thorn problems on low detection sensitivity and high false-positive rate caused by heterogeneity and morphological complexity of 3-D nodule features, a computer-aided detection system is developed to increase the detection sensitivity and classification accuracy of pulmonary nodules. The contributions include: (1) Nodule candidate detection: 3-D Residual U-Net model is improved to detect candidate nodules, which constructs 3-D context-guided module to extract local and global nodule features by setting the dilated convolution with different dilation rates. Furthermore, channel attention mechanism is used to dynamically adjust the channel features, which enhances the generalization and expression ability of the detection-network to effectively learn 3-D spatial context features. (2) False-positive reduction: multi-branch classification network is designed for multi-task learning. Image reconstruction task is performed to retain more microscopic nodules information from convolutional neural network (CNN) hierarchy. Moreover, each branch deals with the feature map at corresponding depth layers, and then all branches&#x2019; feature maps are combined together to perform nodule classification task. Numerous experimental results show that the proposed system is perfectly qualified for pulmonary nodules detection on Lung Nodules Analysis 2016 dataset, which achieves detection sensitivity up to 94.0&#x0025; and competition performance metric (CPM) score up to 0.959.

Accurate classification of nodules and non‐nodules from computed tomography images based on radiomics and machine learning algorithms

AbstractAccurate classification of lung nodules can lead to a more favorable diagnosis and treatment for lung cancer. In this study, an accurate classification of nodules and non‐nodules based on radiomics and machine learning algorithms has been presented. We validate our method on 4999 nodule candidates and use accuracy, recall, precision, f1‐score, and the area under receiver operating characteristic curve (AUC) as evaluation metrics. Experimental results manifest that for most classifiers, recursive feature elimination (RFE) has higher AUC values than chi‐square test and principal component analysis, selecting 15 features is better in AUC than 10 features and 20 features and the ratio between training data set and testing data set of 9:1 has the best predictive performance. When the feature selection method is RFE, the number of features is 15 and the ratio is 9:1, random forest has the highest AUC value (0.9536), accuracy (0.9580), recall (0.9893), precision (0.9392), and f1‐score (0.9636).

Interpretative computer-aided lung cancer diagnosis: From radiology analysis to malignancy evaluation

Background and Objective: Computer-aided diagnosis (CAD) systems promote accurate diagnosis and reduce the burden of radiologists. A CAD system for lung cancer diagnosis includes nodule candidate detection and nodule malignancy evaluation. Recently, deep learning-based pulmonary nodule detection has reached satisfactory performance ready for clinical application. However, deep learning-based nodule malignancy evaluation depends on heuristic inference from low-dose computed tomography (LDCT) volume to malignant probability, and lacks clinical cognition. Methods: In this paper, we propose a joint radiology analysis and malignancy evaluation network called R2MNet to evaluate pulmonary nodule malignancy via the analysis of radiological characteristics. Radiological features are extracted as channel descriptor to highlight specific regions of the input volume that are critical for nodule malignancy evaluation. In addition, for model explanations, we propose channel-dependent activation mapping (CDAM) to visualize features and shed light on the decision process of deep neural networks (DNNs). Results: Experimental results on the lung image database consortium image collection (LIDC-IDRI) dataset demonstrate that the proposed method achieved an area under curve (AUC) of 96.27% and 97.52% on nodule radiology analysis and nodule malignancy evaluation, respectively. In addition, explanations of CDAM features proved that the shape and density of nodule regions are two critical factors that influence a nodule to be inferred as malignant. This process conforms to the diagnosis cognition of experienced radiologists. Conclusion: The network inference process conforms to the diagnostic procedure of radiologists and increases the confidence of evaluation results by incorporating radiology analysis with nodule malignancy evaluation. Besides, model interpretation with CDAM features shed light on the focus regions of DNNs during the estimation of nodule malignancy probabilities.

An efficient multi-path 3D convolutional neural network for false-positive reduction of pulmonary nodule detection.

Considering that false-positive and true pulmonary nodules are highly similar in shapes and sizes between lung computed tomography scans, we develop and evaluate a false-positive nodules reduction method applied to the computer-aided diagnosis system. To improve the pulmonary nodule diagnosis quality, a 3D convolutional neural networks (CNN) model is constructed to effectively extract spatial information of candidate nodule features through the hierarchical architecture. Furthermore, three paths corresponding to three receptive field sizes are adopted and concatenated in the network model, so that the feature information is fully extracted and fused to actively adapting to the changes in shapes, sizes, and contextual information between pulmonary nodules. In this way, the false-positive reduction is well implemented in pulmonary nodule detection. Multi-path 3D CNN is performed on LUNA16 dataset, which achieves an average competitive performance metric score of 0.881, and excellent sensitivity of 0.952 and 0.962 occurs to 4, 8 FP/Scans. By constructing a multi-path 3D CNN to fully extract candidate target features, it accurately identifies pulmonary nodules with different sizes, shapes, and background information. In addition, the proposed general framework is also suitable for similar 3D medical image classification tasks.

Lung Cancer Detection using CT SCAN Images

Detection of pulmonary nodules has a crucial effect on the diagnosis of lung cancer, but the detection is a nontrivial task, not only because the appearance of pulmonary nodules varies in a wide range, but also because nodule densities have low contrast against adjacent vessel segments and other lung tissues. Computed tomography (CT) has been shown as the most popular imaging modality for nodule detection, because it has the ability to provide reliable image textures for the detection of small nodules. The development of lung nodule CADe systems using CT imaging modality has made good progress over the past decade. Generally, such CADe systems consist of three stages: 1) image pre-processing, 2) initial nodule candidates (INCs) identification, and 3) false positive (FP) reduction of the INCs with preservation of the true positives (TPs). In the pre-processing stage, the system aims to largely reduce the search space to the lungs, where a segmentation of the lungs from the entire chest volume is usually required. Because of the high image contrast between lung fields and the surrounding body tissue, image intensity-based simple thresholding is effective, and is currently the most commonly used technique for lung segmentation. This paper proposes an adaptive solution to mitigate the difficulty of thresholding-based method in lung segmentation. Sufficient detection power for nodule candidates is inevitably accompanied by many (obvious) FPs. A rule-based filtering operation is often employed to cheaply and drastically reduce the number of obvious FPs, so that their influence on the computationally more expensive learning process can be eliminated. In general, FP reduction using machine learning has been extensively studied in the literature. Compared with unsupervised learning that aims to find hidden structures in unlabelled data, supervised learning, which aims to infer a function from labelled training data, is more frequently used to design a CADe system. Compared with the existing approaches, the morphology based lung cancer detection can be an alternative with either comparable detection performance and less computational cost, or comparable cost and better detection performance.

High precision localization of pulmonary nodules on chest CT utilizing axial slice number labels

BackgroundReidentification of prior nodules for temporal comparison is an important but time-consuming step in lung cancer screening. We develop and evaluate an automated nodule detector that utilizes the axial-slice number of nodules found in radiology reports to generate high precision nodule predictions.Methods888 CTs from Lung Nodule Analysis were used to train a 2-dimensional (2D) object detection neural network. A pipeline of 2D object detection, 3D unsupervised clustering, false positive reduction, and axial-slice numbers were used to generate nodule candidates. 47 CTs from the National Lung Cancer Screening Trial (NLST) were used for model evaluation.ResultsOur nodule detector achieved a precision of 0.962 at a recall of 0.573 on the NLST test set for any nodule. When adjusting for unintended nodule predictions, we achieved a precision of 0.931 at a recall 0.561, which corresponds to 0.06 false positives per CT. Error analysis revealed better detection of nodules with soft tissue attenuation compared to ground glass and undeterminable attenuation. Nodule margins, size, location, and patient demographics did not differ between correct and incorrect predictions.ConclusionsUtilization of axial-slice numbers from radiology reports allowed for development of a lung nodule detector with a low false positive rate compared to prior feature-engineering and machine learning approaches. This high precision nodule detector can reduce time spent on reidentification of prior nodules during lung cancer screening and can rapidly develop new institutional datasets to explore novel applications of computer vision in lung cancer imaging.

Automatic detection and segmentation of lung nodules in different locations from CT images based on adaptive α‐hull algorithm and DenseNet convolutional network

AbstractAutomatic lung nodules detection and segmentation can assist doctors in better diagnosis and treatment for lung cancer. However, precise detection and segmentation are still challenging, because lung nodules can have different contours or locations and may be attached to other tissues, such as neighboring blood vessel and pleural surface. In this study, an automatic detection and segmentation method for lung nodules in different locations has been developed. First, we apply Otsu thresholding to segment lung parenchyma. Next, a morphological opening operation is carried out to remove blood vessels. Then α‐hull operation is proposed to correct lung contours and optimal α values can be acquired adaptively. Finally, DenseNet convolutional network is applied to classify true lung nodules from all nodule candidates. We select the intersection area of at least three radiologists' annotations as ground truth and validate our method on 466 nodules including well‐circumscribed, juxta‐vascular, juxta‐pleural, and pleural tail. Our study not only concentrates on false positive reduction but also evaluates segmentation performance. To give a more comprehensive quantitative evaluation of nodule segmentation, we use evaluation metrics including Jaccard index (JI), dice similar coefficient (DSC), Hausdorff distance, under‐segmentation rate, over‐segmentation rate, sensitivity, specificity, accuracy, and false positive rate. Overall results are 0.6385 ± 0.1309, 0.7710 ± 0.1057, 3.5123 ± 3.1251, 0.1769 ± 0.1308, 0.1848 ± 0.1463, 0.7936 ± 0.1417, 0.9998 ± 0.0003, 0.9997 ± 0.0003, and 0.0002 ± 0.0003, respectively. These metrics can demonstrate segmentation performance in multiple dimensions. As a general and automatic detection and segmentation framework for lung nodules in different locations, this study achieves better performance than previous approaches in JI and DSC.

3D multi-scale deep convolutional neural networks for pulmonary nodule detection.

With the rapid development of big data and artificial intelligence technology, computer-aided pulmonary nodule detection based on deep learning has achieved some successes. However, the sizes of pulmonary nodules vary greatly, and the pulmonary nodules have visual similarity with structures such as blood vessels and shadows around pulmonary nodules, which make the quick and accurate detection of pulmonary nodules in CT image still a challenging task. In this paper, we propose two kinds of 3D multi-scale deep convolution neural networks for nodule candidate detection and false positive reduction respectively. Among them, the nodule candidate detection network consists of two parts: 1) the backbone network part Res2SENet, which is used to extract multi-scale feature information of pulmonary nodules, it is composed of the multi-scale Res2Net modules of multiple available receptive fields at a granular level and the squeeze-and-excitation units; 2) the detection part, which uses a region proposal network structure to determine region candidates, and introduces context enhancement module and spatial attention module to improve detection performance. The false positive reduction network, also composed of the multi-scale Res2Net modules and the squeeze-and-excitation units, can further classify the nodule candidates generated by the nodule candidate detection network and screen out the ground truth positive nodules. Finally, the prediction probability generated by the nodule candidate detection network is weighted average with the prediction probability generated by the false positive reduction network to obtain the final results. The experimental results on the publicly available LUNA16 dataset showed that the proposed method has a superior ability to detect pulmonary nodules in CT images.

Adoption of computerized tomography images in detection of lung nodules and analysis of neuropeptide correlative substances under deep learning algorithm

To solve the problems of low sensitivity and high false positive rate of current lung nodule detection technology, and to explore the correlation between neuropeptide correlative substances and lung nodules, in the study, the deep learning-based computerized tomography (CT) images were used to detect lung nodules. Specifically, a multi-scale three-dimensional convolutional neural network (CNN) was established for lung nodule detection. First, high-quality candidate nodules were obtained by a two-dimensional CNN. Then, three-dimensional CNNs of different scales were constructed for candidate nodules of different sizes, and the fused model was used to classify candidate nodules. Finally, the detection results of the single network, the two-dimensional CNN, and the fused network were compared. Meanwhile, the serum levels of vasoactive intestinal peptide (VIP) and substance P (SP) were detected in patients with lung injury caused by lung nodules, so as to analyze the correlation between neuropeptide correlative substances and lung nodules. The results showed that the detection sensitivity of the fused network was 85.8% and 92.9% when the number of false positives was 1 and 4, respectively, which were higher than that of the single network and two-dimensional CNN. However, the content of VIP in the serum of patients with pulmonary nodules was significantly reduced, while the content of SP was significantly increased. Therefore, the algorithm proposed in this study can effectively improve the sensitivity of the pulmonary nodule detection system and reduce false positive rate; at the same time, neuropeptide correlative substances may be correlated with the lung injury caused by pulmonary nodules.

Classifier Ensemble Based on Computed Tomography Attenuation Patterns for Computer-Aided Detection System

Cancer is one of the leading causes of mortality worldwide, specifically lung cancer. Computer-Aided Detection (CADe) systems are being proposed to assist radiologists in the task of pulmonary nodule detection. In this paper, we propose a CADe system that uses Deep Convolutional Neural Network (DCNN). In the Nodule Candidate Detection (NCD) step, we used Mask Region-Convolutional Neural Network (Mask R-CNN) to detect bounding boxes in 2D slices of low-dose Computed Tomography (CT) scans. In the False Positive Reduction (FPR) step, we used a classifier ensemble based on CT attenuation patterns to boost 3D pulmonary nodule classification performance. The final confidence index generated by the CADe system to the pulmonary nodule candidates is the average of the prediction obtained with the NCD and FPR steps. The CADe system was validated on the publicly available LUng Nodule Analysis 2016 (LUNA16) challenge and obtained a sensitivity of 94.90% and an average of 1.0 False Positives per scan (FP/Scan), against 96.90% of the proposal that combines different existing CADe systems. To the best of our knowledge, our proposal has one of the best results of CADe systems, outperforming other state-of-the-art individual methods.

SANet: A Slice-Aware Network for Pulmonary Nodule Detection.

Lung cancer is the most common cause of cancer death worldwide. A timely diagnosis of the pulmonary nodules makes it possible to detect lung cancer in the early stage, and thoracic computed tomography (CT) provides a convenient way to diagnose nodules. However, it is hard even for experienced doctors to distinguish them from the massive CT slices. The currently existing nodule datasets are limited in both scale and category, which is insufficient and greatly restricts its applications. In this paper, we collect the largest and most diverse dataset named PN9 for pulmonary nodule detection by far. Specifically, it contains 8,798 CT scans and 40,439 annotated nodules from 9 common classes. We further propose a slice-aware network (SANet) for pulmonary nodule detection. A slice grouped non-local (SGNL) module is developed to capture long-range dependencies among any positions and any channels of one slice group in the feature map. And we introduce a 3D region proposal network to generate pulmonary nodule candidates with high sensitivity, while this detection stage usually comes with many false positives. Subsequently, a false positive reduction module (FPR) is proposed by using the multi-scale feature maps. To verify the performance of SANet and the significance of PN9, we perform extensive experiments compared with several state-of-the-art 2D CNN-based and 3D CNN-based detection methods. Promising evaluation results on PN9 prove the effectiveness of our proposed SANet. The dataset and source code is available at https://mmcheng.net/SANet/.

Deep convolutional neural networks for multiplanar lung nodule detection: Improvement in small nodule identification.

PurposeEarly detection of lung cancer is of importance since it can increase patients’ chances of survival. To detect nodules accurately during screening, radiologists would commonly take the axial, coronal, and sagittal planes into account, rather than solely the axial plane in clinical evaluation. Inspired by clinical work, the paper aims to develop an accurate deep learning framework for nodule detection by a combination of multiple planes.MethodsThe nodule detection system is designed in two stages, multiplanar nodule candidate detection, multiscale false positive (FP) reduction. At the first stage, a deeply supervised encoder–decoder network is trained by axial, coronal, and sagittal slices for the candidate detection task. All possible nodule candidates from the three different planes are merged. To further refine results, a three‐dimensional multiscale dense convolutional neural network that extracts multiscale contextual information is applied to remove non‐nodules. In the public LIDC‐IDRI dataset, 888 computed tomography scans with 1186 nodules accepted by at least three of four radiologists are selected to train and evaluate our proposed system via a tenfold cross‐validation scheme. The free‐response receiver operating characteristic curve is used for performance assessment.ResultsThe proposed system achieves a sensitivity of 94.2% with 1.0 FP/scan and a sensitivity of 96.0% with 2.0 FPs/scan. Although it is difficult to detect small nodules (i.e., <6 mm), our designed CAD system reaches a sensitivity of 93.4% (95.0%) of these small nodules at an overall FP rate of 1.0 (2.0) FPs/scan. At the nodule candidate detection stage, results show that the system with a multiplanar method is capable to detect more nodules compared to using a single plane.ConclusionOur approach achieves good performance not only for small nodules but also for large lesions on this dataset. This demonstrates the effectiveness of our developed CAD system for lung nodule detection.

No Surprises: Training Robust Lung Nodule Detection for Low-Dose CT Scans by Augmenting With Adversarial Attacks.

Detecting malignant pulmonary nodules at an early stage can allow medical interventions which may increase the survival rate of lung cancer patients. Using computer vision techniques to detect nodules can improve the sensitivity and the speed of interpreting chest CT for lung cancer screening. Many studies have used CNNs to detect nodule candidates. Though such approaches have been shown to outperform the conventional image processing based methods regarding the detection accuracy, CNNs are also known to be limited to generalize on under-represented samples in the training set and prone to imperceptible noise perturbations. Such limitations can not be easily addressed by scaling up the dataset or the models. In this work, we propose to add adversarial synthetic nodules and adversarial attack samples to the training data to improve the generalization and the robustness of the lung nodule detection systems. To generate hard examples of nodules from a differentiable nodule synthesizer, we use projected gradient descent (PGD) to search the latent code within a bounded neighbourhood that would generate nodules to decrease the detector response. To make the network more robust to unanticipated noise perturbations, we use PGD to search for noise patterns that can trigger the network to give over-confident mistakes. By evaluating on two different benchmark datasets containing consensus annotations from three radiologists, we show that the proposed techniques can improve the detection performance on real CT data. To understand the limitations of both the conventional networks and the proposed augmented networks, we also perform stress-tests on the false positive reduction networks by feeding different types of artificially produced patches. We show that the augmented networks are more robust to both under-represented nodules as well as resistant to noise perturbations.