Enhanced focal cortical dysplasia detection in pediatric frontal lobe epilepsy with asymmetric radiomic and morphological features.

Objective: Focal cortical dysplasia (FCD) is the most common pathological cause for pediatric epilepsy, with frontal lobe epilepsy (FLE) being the most prevalent in the pediatric population. We attempted to utilize radiomic and morphological methods on MRI and PET to detect FCD in children with FLE. Methods: Thirty-seven children with FLE and 20 controls were included in the primary cohort, and a five-fold cross-validation was performed. In addition, we validated the performance in an independent site of 12 FLE children. A two-stage experiments including frontal lobe and subregions were employed to detect the lesion area of FCD, incorporating the asymmetric feature between the left and right hemispheres. Specifically, for the radiomics approach, we used gray matter (GM), white matter (WM), GM and WM, and the gray-white matter boundary regions of interest to extract features. Then, we employed a Multi-Layer Perceptron classifier to achieve FCD lesion localization based on both radiomic and morphological methods. Results: The Multi-Layer Perceptron model based on the asymmetric feature exhibited excellent performance both in the frontal lobe and subregions. In the primary cohort and independent site, the radiomics analysis with GM and WM asymmetric features had the highest sensitivity (89.2 and 91.7%) and AUC (98.9 and 99.3%) in frontal lobe. While in the subregions, the GM asymmetric features had the highest sensitivity (85.6 and 79.7%). Furthermore, relying on the highest sensitivity of GM and WM asymmetric features in frontal lobe, when integrated with the subregions results, our approach exhibited overlaps with GM asymmetric features (55.4 and 52.4%), as well as morphological asymmetric features (54.4 and 53.8%), both in the primary cohort and at the independent site. Significance: This study demonstrates that a two-stage design based on the asymmetry of radiomic and morphological features can improve FCD detection. Specifically, incorporating regions of interest for GM, WM, GM, and WM, and the gray-white matter boundary significantly enhances the localization capabilities for lesion detection within the radiomics approach.

End-to-end automatic pathology localization for Alzheimer's disease diagnosis using structural MRI.

Structural magnetic resonance imaging (sMRI) is an essential part of the clinical assessment of patients at risk of Alzheimer dementia. One key challenge in sMRI-based computer-aided dementia diagnosis is to localize local pathological regions for discriminative feature learning. Existing solutions predominantly depend on generating saliency maps for pathology localization and handle the localization task independently of the dementia diagnosis task, leading to a complex multi-stage training pipeline that is hard to optimize with weakly-supervised sMRI-level annotations. In this work, we aim to simplify the pathology localization task and construct an end-to-end automatic localization framework (AutoLoc) for Alzheimer's disease diagnosis. To this end, we first present an efficient pathology localization paradigm that directly predicts the coordinate of the most disease-related region in each sMRI slice. Then, we approximate the non-differentiable patch-cropping operation with the bilinear interpolation technique, which eliminates the barrier to gradient backpropagation and thus enables the joint optimization of localization and diagnosis tasks. Extensive experiments on commonly used ADNI and AIBL datasets demonstrate the superiority of our method. Especially, we achieve 93.38% and 81.12% accuracy on Alzheimer's disease classification and mild cognitive impairment conversion prediction tasks, respectively. Several important brain regions, such as rostral hippocampus and globus pallidus, are identified to be highly associated with Alzheimer's disease.

Improved motor imagery classification using adaptive spatial filters based on particle swarm optimization algorithm.

Background: As a typical self-paced brain-computer interface (BCI) system, the motor imagery (MI) BCI has been widely applied in fields such as robot control, stroke rehabilitation, and assistance for patients with stroke or spinal cord injury. Many studies have focused on the traditional spatial filters obtained through the common spatial pattern (CSP) method. However, the CSP method can only obtain fixed spatial filters for specific input signals. In addition, the CSP method only focuses on the variance difference of two types of electroencephalogram (EEG) signals, so the decoding ability of EEG signals is limited. Methods: To make up for these deficiencies, this study introduces a novel spatial filter-solving paradigm named adaptive spatial pattern (ASP), which aims to minimize the energy intra-class matrix and maximize the inter-class matrix of MI-EEG after spatial filtering. The filter bank adaptive and common spatial pattern (FBACSP), our proposed method for MI-EEG decoding, amalgamates ASP spatial filters with CSP features across multiple frequency bands. Through a dual-stage feature selection strategy, it employs the Particle Swarm Optimization algorithm for spatial filter optimization, surpassing traditional CSP approaches in MI classification. To streamline feature sets and enhance recognition efficiency, it first prunes CSP features in each frequency band using mutual information, followed by merging these with ASP features. Results: Comparative experiments are conducted on two public datasets (2a and 2b) from BCI competition IV, which show the outstanding average recognition accuracy of FBACSP. The classification accuracy of the proposed method has reached 74.61 and 81.19% on datasets 2a and 2b, respectively. Compared with the baseline algorithm, filter bank common spatial pattern (FBCSP), the proposed algorithm improves by 11.44 and 7.11% on two datasets, respectively (p < 0.05). Conclusion: It is demonstrated that FBACSP has a strong ability to decode MI-EEG. In addition, the analysis based on mutual information, t-SNE, and Shapley values further proves that ASP features have excellent decoding ability for MI-EEG signals and explains the improvement of classification performance by the introduction of ASP features. These findings may provide useful information to optimize EEG-based BCI systems and further improve the performance of non-invasive BCI.

Automatic Localization of Seizure Onset Zone Based on Multi-Epileptogenic Biomarkers Analysis of Single-Contact from Interictal SEEG.

Successful surgery on drug-resistant epilepsy patients (DRE) needs precise localization of the seizure onset zone (SOZ). Previous studies analyzing this issue still face limitations, such as inadequate analysis of features, low sensitivity and limited generality. Our study proposed an innovative and effective SOZ localization method based on multiple epileptogenic biomarkers (spike and HFOs), and analysis of single-contact (MEBM-SC) to address the above problems. We extracted contacts epileptic features from signal distributions and signal energy based on machine learning and end-to-end deep learning. Among them, a normalized pathological ripple rate was designed to reduce the disturbance of physiological ripple and enhance the performance of SOZ localization. Then, a feature selection algorithm based on Shapley value and hypothetical testing (ShapHT+) was used to limit interference from irrelevant features. Moreover, an attention mechanism and a focal loss algorithm were used on the classifier to learn significant features and overcome the unbalance of SOZ/nSOZ contacts. Finally, we provided an SOZ prediction and visualization on magnetic resonance imaging (MRI). Ten patients with DRE were selected to verify our method. The experiment performed cross-validation and revealed that MEBM-SC obtains higher sensitivity. Additionally, the spike has better sensitivity while HFOs have better specificity, and the combination of these biomarkers can achieve the best performance. The study confirmed that MEBM-SC can increase the sensitivity and accuracy of SOZ localization and help clinicians to perform a precise and reliable preoperative evaluation based on interictal SEEG.

SEEG-Net: An explainable and deep learning-based cross-subject pathological activity detection method for drug-resistant epilepsy.

OBJECTIVE: Precise preoperative evaluation of drug-resistant epilepsy (DRE) requires accurate analysis of invasive stereoelectroencephalography (SEEG). With the tremendous breakthrough of Artificial intelligence (AI), previous studies can help clinical experts to identify pathological activities automatically. However, they still face limitations when applied in real-world clinical DRE scenarios, such as sample imbalance, cross-subject domain shift, and poor interpretability. Our objective is to propose a model that can address the above problems and realizes high-sensitivity SEEG pathological activity detection based on two real clinical datasets. METHODS: Our proposed innovative and effective SEEG-Net introduces a multiscale convolutional neural network (MSCNN) to increase the receptive field of the model, and to learn SEEG multiple frequency domain features, local and global features. Moreover, we designed a novel focal domain generalization loss (FDG-loss) function to enhance the target sample weight and to learn domain consistency features. Furthermore, to enhance the interpretability and flexibility of SEEG-Net, we explain SEEG-Net from multiple perspectives, such as significantly different features, interpretable models, and model learning process interpretation by Grad-CAM++. RESULTS: The performance of our proposed method is verified on a public benchmark multicenter SEEG dataset and a private clinical SEEG dataset for a robust comparison. The experimental results demonstrate that the SEEG-Net model achieves the highest sensitivity and is state-of-the-art on cross-subject (for different patients) evaluation, and well deal with the known problems. Besides, we provide an SEEG processing and database construction flow, by maintaining consistency with the real-world clinical scenarios. SIGNIFICANCE: According to the results, SEEG-Net is constructed to increase the sensitivity of SEEG pathological activity detection. Simultaneously, we settled certain problems about AI assistance in clinical DRE, built a bridge between AI algorithm application and clinical practice.

Screening of Human Circular RNAs as Biomarkers for Early Onset Detection of Alzheimer's Disease.

Circular RNAs (circRNAs) are a distinctive type of endogenous non-coding RNAs, and their regulatory roles in neurological disorders have received immense attention. CircRNAs significantly contribute to the regulation of gene expression and progression of neurodegenerative disorders including Alzheimer's disease (AD). The current study aimed to identify circRNAs as prognostic and potential biomarkers in AD. The differentially expressed circRNAs among subjective cognitive decline, amnestic mild cognitive impairment, and age-matched normal donors were determined through Arraystar Human circRNA Array V2 analysis. The annotations of circRNAs-microRNA interactions were predicted by employing Arraystar's homemade microRNAs (miRNA) target prediction tool. Bioinformatics analyses comprising gene ontology enrichment, KEGG pathway, and network analysis were conducted. Microarray analysis revealed the 33 upregulated and 11 downregulated differentially expressed circRNAs (FC ≥ 1.5 and p-values ≤ 0.05). The top 10 differentially expressed upregulated and downregulated circRNAs have been chosen for further expression validation through quantitative real-time PCR and subsequently, hsa-circRNA_001481 and hsa_circRNA_000479 were confirmed experimentally. Bioinformatics analyses determined the circRNA-miRNA-mRNA interactions and microRNA response elements to inhibit the expression of miRNAs and mRNA targets. Gene ontology enrichment and KEGG pathways analysis revealed the functional clustering of target mRNAs suggesting the functional verification of these two promising circRNAs. It is concluded that human circRNA_001481 and circRNA_000479 could be utilized as potential biomarkers for the early onset detection of AD and the development of effective therapeutics.

Aberrant topological organization and age-related differences in the human connectome in subjective cognitive decline by using regional morphology from magnetic resonance imaging.

Subjective cognitive decline (SCD) is characterized by self-experienced deficits in cognitive capacity with normal performance in objective cognitive tests. Previous structural covariance studies showed specific insights into understanding the structural alterations of the brain in neurodegenerative diseases. Moreover, in subjects with neurodegenerative diseases, accelerated brain degeneration with aging was shown. However, the age-related variations in coordinated topological patterns of morphological networks in individuals with SCD remain poorly understood. In this study, 77 individual morphological networks were constructed, including 42 normal controls (NCs) and 35 SCD individuals, from structural magnetic resonance imaging (sMRI). A stepwise linear regression model and partial correlation analysis were constructed to evaluate the differences in age-related alterations of the network properties in individuals with SCD compared with NCs. Compared with NC, the properties of integration and segregation in individuals with SCD were lower, and the aberrant metrics were negatively correlated with age in SCD. The rich-club connections persevered, but the paralimbic system connections were disrupted in individuals with SCD compared with NCs. In addition, age-related differences in nodal global efficiency are distributed mainly in prefrontal cortex regions. In conclusion, the age-related disruption of topological organizations in individuals with SCD may indicate that the degeneration of brain efficiency with aging was accelerated in individuals with SCD.

Exploring brain glucose metabolic patterns in cognitively normal adults at risk of Alzheimer's disease: A cross-validation study with Chinese and ADNI cohorts.

OBJECTIVE: Disease-related metabolic brain patterns have been verified for a variety of neurodegenerative diseases including Alzheimer's disease (AD). This study aimed to explore and validate the pattern derived from cognitively normal controls (NCs) in the Alzheimer's continuum. METHODS: This study was based on two cohorts; one from the Alzheimer's Disease Neuroimaging Initiative (ADNI) and the other from the Sino Longitudinal Study on Cognitive Decline (SILCODE). Each subject underwent [18F]fluoro-2-deoxyglucose positron emission tomography (PET) and [18F]florbetapir-PET imaging. Participants were binary-grouped based on ß-amyloid (Aß) status, and the positivity was defined as Aß+. Voxel-based scaled subprofile model/principal component analysis (SSM/PCA) was used to generate the "at-risk AD-related metabolic pattern (ARADRP)" for NCs. The pattern expression score was obtained and compared between the groups, and receiver operating characteristic curves were drawn. Notably, we conducted cross-validation to verify the robustness and correlation analyses to explore the relationships between the score and AD-related pathological biomarkers. RESULTS: Forty-eight Aß+ NCs and 48 Aß- NCs were included in the ADNI cohort, and 25 Aß+ NCs and 30 Aß- NCs were included in the SILCODE cohort. The ARADRPs were identified from the combined cohorts and the two separate cohorts, characterized by relatively lower regional loadings in the posterior parts of the precuneus, posterior cingulate, and regions of the temporal gyrus, as well as relatively higher values in the superior/middle frontal gyrus and other areas. Patterns identified from the two separate cohorts showed some regional differences, including the temporal gyrus, basal ganglia regions, anterior parts of the precuneus, and middle cingulate. Cross-validation suggested that the pattern expression score was significantly higher in the Aß+ group of both cohorts (p < 0.01), and contributed to the diagnosis of Aß+ NCs (with area under the curve values of 0.696-0.815). The correlation analysis revealed that the score was related to tau pathology measured in cerebrospinal fluid (p-tau: p < 0.02; t-tau: p < 0.03), but not Aß pathology assessed with [18F]florbetapir-PET (p > 0.23). CONCLUSIONS: ARADRP exists for NCs, and the acquired pattern expression score shows a certain ability to discriminate Aß+ NCs from Aß- NCs. The SSM/PCA method is expected to be helpful in the ultra-early diagnosis of AD in clinical practice.

Role of High-Frequency Oscillation in Locating an Epileptogenic Zone for Radiofrequency Thermocoagulation.

Radiofrequency thermocoagulation (RFTC) has been proposed as a first-line surgical treatment option for patients with drug-resistant focal epilepsy (DRE) that is associated with gray matter nodular heterotopia (GMNH). Excellent results on seizures have been reported following unilateral RFTC performed on ictal high-frequency-discharge, fast-rhythm, and low-voltage initiation areas. Complex cases (GMNH plus other malformations of cortical development) do not have good outcomes with RFTC. Yet, there is little research studying the effect of high-frequency oscillation in locating epileptogenic zones for thermocoagulation on unilateral, DRE with bilateral GMNH. We present a case of DRE with bilateral GMNH, treated using RFTC on unilateral GMNH and the overlying cortex, guided by stereotactic electroencephalogram (SEGG), and followed up for 69 months. Twenty-four-hour EGG recordings, seizure frequency, post-RFTC MRI, and neuropsychological tests were performed once yearly. To date, this patient is seizure-free, the electroencephalogram is normal, neuropsychological problems have not been found, and the trace of RFTC has been clearly identified on MRI. His dosage of antiepileptic medication has, furthermore, been significantly reduced. It is concluded that RFTC on unilateral DRE with bilateral GMNH may achieve good long-term effects, lasting up to, and perhaps longer than, 69 months. Ictal high-frequency oscillation (fast ripple) inside the heterotopia and the overlying cortex may be the key to this successful effect.

Divergent Connectivity Changes in Gray Matter Structural Covariance Networks in Subjective Cognitive Decline, Amnestic Mild Cognitive Impairment, and Alzheimer's Disease.

Alzheimer's disease (AD) has a long preclinical stage that can last for decades prior to progressing toward amnestic mild cognitive impairment (aMCI) and/or dementia. Subjective cognitive decline (SCD) is characterized by self-experienced memory decline without any evidence of objective cognitive decline and is regarded as the later stage of preclinical AD. It has been reported that the changes in structural covariance patterns are affected by AD pathology in the patients with AD and aMCI within the specific large-scale brain networks. However, the changes in structural covariance patterns including normal control (NC), SCD, aMCI, and AD are still poorly understood. In this study, we recruited 42 NCs, 35 individuals with SCD, 43 patients with aMCI, and 41 patients with AD. Gray matter (GM) volumes were extracted from 10 readily identifiable regions of interest involved in high-order cognitive function and AD-related dysfunctional structures. The volume values were used to predict the regional densities in the whole brain by using voxel-based statistical and multiple linear regression models. Decreased structural covariance and weakened connectivity strength were observed in individuals with SCD compared with NCs. Structural covariance networks (SCNs) seeding from the default mode network (DMN), salience network, subfields of the hippocampus, and cholinergic basal forebrain showed increased structural covariance at the early stage of AD (referring to aMCI) and decreased structural covariance at the dementia stage (referring to AD). Moreover, the SCN seeding from the executive control network (ECN) showed a linearly increased extent of the structural covariance during the early and dementia stages. The results suggest that changes in structural covariance patterns as the order of NC-SCD-aMCI-AD are divergent and dynamic, and support the structural disconnection hypothesis in individuals with SCD.

Combined Support Vector Machine Classifier and Brain Structural Network Features for the Individual Classification of Amnestic Mild Cognitive Impairment and Subjective Cognitive Decline Patients.

OBJECTIVE: Individuals with subjective cognitive decline (SCD) or amnestic mild cognitive impairment (aMCI) represent important targets for the early detection and intervention of Alzheimer's disease (AD). In this study, we employed a multi-kernel support vector machine (SVM) to examine whether white matter (WM) structural networks can be used for screening SCD and aMCI. METHODS: A total of 138 right-handed participants [51 normal controls (NC), 36 SCD, 51 aMCI] underwent MRI brain scans. For each participant, three types of WM networks with different edge weights were constructed with diffusion MRI data: fiber number-weighted networks, mean fractional anisotropy-weighted networks, and mean diffusivity (MD)-weighted networks. By employing a multiple-kernel SVM, we seek to integrate information from three weighted networks to improve classification performance. The accuracy of classification between each pair of groups was evaluated via leave-one-out cross-validation. RESULTS: For the discrimination between SCD and NC, an area under the curve (AUC) value of 0.89 was obtained, with an accuracy of 83.9%. Further analysis revealed that the methods using three types of WM networks outperformed other methods using single WM network. Moreover, we found that most of discriminative features were from MD-weighted networks, which distributed among frontal lobes. Similar classification performance was also reported in the differentiation between subjects with aMCI and NCs (accuracy = 83.3%). Between SCD and aMCI, an AUC value of 0.72 was obtained, with an accuracy of 72.4%, sensitivity of 74.5% and specificity of 69.4%. The highest accuracy was achieved with features only selected from MD-weighted networks. CONCLUSION: White matter structural network features help machine learning algorithms accurately identify individuals with SCD and aMCI from NCs. Our findings have significant implications for the development of potential brain imaging markers for the early detection of AD.

Computer-Aided Intracranial EEG Signal Identification Method Based on a Multi-Branch Deep Learning Fusion Model and Clinical Validation.

Surgical intervention or the control of drug-refractory epilepsy requires accurate analysis of invasive inspection intracranial EEG (iEEG) data. A multi-branch deep learning fusion model is proposed to identify epileptogenic signals from the epileptogenic area of the brain. The classical approach extracts multi-domain signal wave features to construct a time-series feature sequence and then abstracts it through the bi-directional long short-term memory attention machine (Bi-LSTM-AM) classifier. The deep learning approach uses raw time-series signals to build a one-dimensional convolutional neural network (1D-CNN) to achieve end-to-end deep feature extraction and signal detection. These two branches are integrated to obtain deep fusion features and results. Resampling is employed to split the imbalanced epileptogenic and non-epileptogenic samples into balanced subsets for clinical validation. The model is validated over two publicly available benchmark iEEG databases to verify its effectiveness on a private, large-scale, clinical stereo EEG database. The model achieves high sensitivity (97.78%), accuracy (97.60%), and specificity (97.42%) on the Bern-Barcelona database, surpassing the performance of existing state-of-the-art techniques. It is then demonstrated on a clinical dataset with an average intra-subject accuracy of 92.53% and cross-subject accuracy of 88.03%. The results suggest that the proposed method is a valuable and extremely robust approach to help researchers and clinicians develop an automated method to identify the source of iEEG signals.

Automatic Detection of High-Frequency Oscillations Based on an End-to-End Bi-Branch Neural Network and Clinical Cross-Validation.

Accurate identification of high-frequency oscillation (HFO) is an important prerequisite for precise localization of epileptic foci and good prognosis of drug-refractory epilepsy. Exploring a high-performance automatic detection method for HFOs can effectively help clinicians reduce the error rate and reduce manpower. Due to the limited analysis perspective and simple model design, it is difficult to meet the requirements of clinical application by the existing methods. Therefore, an end-to-end bi-branch fusion model is proposed to automatically detect HFOs. With the filtered band-pass signal (signal branch) and time-frequency image (TFpic branch) as the input of the model, two backbone networks for deep feature extraction are established, respectively. Specifically, a hybrid model based on ResNet1d and long short-term memory (LSTM) is designed for signal branch, which can focus on both the features in time and space dimension, while a ResNet2d with a Convolutional Block Attention Module (CBAM) is constructed for TFpic branch, by which more attention is paid to useful information of TF images. Then the outputs of two branches are fused to realize end-to-end automatic identification of HFOs. Our method is verified on 5 patients with intractable epilepsy. In intravalidation, the proposed method obtained high sensitivity of 94.62%, specificity of 92.7%, and F1-score of 93.33%, and in cross-validation, our method achieved high sensitivity of 92.00%, specificity of 88.26%, and F1-score of 89.11% on average. The results show that the proposed method outperforms the existing detection paradigms of either single signal or single time-frequency diagram strategy. In addition, the average kappa coefficient of visual analysis and automatic detection results is 0.795. The method shows strong generalization ability and high degree of consistency with the gold standard meanwhile. Therefore, it has great potential to be a clinical assistant tool.

User Association and Power Control for Energy Efficiency Maximization in M2M-Enabled Uplink Heterogeneous Networks with NOMA.

To support a vast number of devices with less energy consumption, we propose a new user association and power control scheme for machine to machine enabled heterogeneous networks with non-orthogonal multiple access (NOMA), where a mobile user (MU) acting as a machine-type communication gateway can decode and forward both the information of machine-type communication devices and its own data to the base station (BS) directly. MU association and power control are jointly considered in the formulated as optimization problem for energy efficiency (EE) maximization under the constraints of minimum data rate requirements of MUs. A many-to-one MU association matching algorithm is firstly proposed based on the theory of matching game. By taking swap matching operations among MUs, BSs, and sub-channels, the original problem can be solved by dealing with the EE maximization for each sub-channel. Then, two power control algorithms are proposed, where the tools of sequential optimization, fractional programming, and exhaustive search have been employed. Simulation results are provided to demonstrate the optimality properties of our algorithms under different parameter settings.

An Opportunistic Cooperative Packet Transmission Scheme in Wireless Multi-Hop Networks.

Cooperative routing, combining cooperative communication in the physical layer and routing technology in the network layer, is one of the most widely used technologies for improving end-to-end transmission reliability and delay in the wireless multi-hop networks. However, the existing cooperative routing schemes are designed based on an optimal fixed-path routing so that the end-to-end performance is greatly restricted by the low spatial efficiency. To address this problem, in this paper an opportunistic cooperative packet transmission (OCPT) scheme is explored by combining cooperative communication and opportunistic routing. The proposed scheme divides the multi-hop route into multiple virtual multiple-input-multiple-output (MIMO) transmissions. Before each transmission, based on the idea of opportunistic routing, a cluster head (CH) is introduced to determine the multiple transmitters and multiple receivers to form a cluster. Then, the single-hop transmission distance is defined as the metric of forward progress to the destination. Each intra-cluster cooperative packet transmission is formulated as a transmit beamforming optimization problem, and an iterative optimal beamforming policy is proposed to solve the problem and maximize the single-hop transmission distance. CH organizes multiple transmitters to cooperatively transmit packets to multiple receivers with the optimized transmit beamforming vector. Finally, according to the transmission results, the cluster is updated and the new cooperative transmission is started. Iteratively, the transmission lasts until the destination has successfully received the packet. We comprehensively evaluate the OCPT scheme by comparing it with conventional routing schemes. The simulation results demonstrate that the proposed OCPT scheme is effective on shortening the end-to-end transmission delay, increasing the number of successful packet transmissions and improving the packet arrival ratio and transmission efficiency.

Attention-Guided Convolutional Neural Network for Detecting Pneumonia on Chest X-Rays.

Pneumonia is a common infectious disease in the world. Its main diagnostic method is chest X-ray (CXR) examination. However, the high visual similarity between a large number of pathologies in CXR makes the interpretation and differentiation of pneumonia a challenge. In this paper, we propose an improved convolutional neural network (CNN) model for pneumonia detection. In order to guide the CNN to focus on disease-specific attended region, the pneumonia area of image is erased and marked as a non-pneumonia sample. In addition, transfer learning is used to segment the interest region of lungs to suppress background interference. The experimental results show that the proposed method is superior to the state-of-the-art object detection model in terms of accuracy and false positive rate.

3D multi-view convolutional neural networks for lung nodule classification.

The 3D convolutional neural network (CNN) is able to make full use of the spatial 3D context information of lung nodules, and the multi-view strategy has been shown to be useful for improving the performance of 2D CNN in classifying lung nodules. In this paper, we explore the classification of lung nodules using the 3D multi-view convolutional neural networks (MV-CNN) with both chain architecture and directed acyclic graph architecture, including 3D Inception and 3D Inception-ResNet. All networks employ the multi-view-one-network strategy. We conduct a binary classification (benign and malignant) and a ternary classification (benign, primary malignant and metastatic malignant) on Computed Tomography (CT) images from Lung Image Database Consortium and Image Database Resource Initiative database (LIDC-IDRI). All results are obtained via 10-fold cross validation. As regards the MV-CNN with chain architecture, results show that the performance of 3D MV-CNN surpasses that of 2D MV-CNN by a significant margin. Finally, a 3D Inception network achieved an error rate of 4.59% for the binary classification and 7.70% for the ternary classification, both of which represent superior results for the corresponding task. We compare the multi-view-one-network strategy with the one-view-one-network strategy. The results reveal that the multi-view-one-network strategy can achieve a lower error rate than the one-view-one-network strategy.