Activation Mapping
Recently Published Documents





Ge Su ◽  
Bo Lin ◽  
Wei Luo ◽  
Jianwei Yin ◽  
Shuiguang Deng ◽  

Parkinson’s disease is the second most common neurodegenerative disorder, commonly affecting elderly people over the age of 65. As the cardinal manifestation, hypomimia, referred to as impairments in normal facial expressions, stays covert. Even some experienced doctors may miss these subtle changes, especially in a mild stage of this disease. The existing methods for hypomimia recognition are mainly dominated by statistical variable-based methods with the help of traditional machine learning algorithms. Despite the success of recognizing hypomimia, they show a limited accuracy and lack the capability of performing semantic analysis. Therefore, developing a computer-aided diagnostic method for semantically recognizing hypomimia is appealing. In this article, we propose a Semantic Feature based Hypomimia Recognition network , named SFHR-NET , to recognize hypomimia based on facial videos. First, a Semantic Feature Classifier (SF-C) is proposed to adaptively adjust feature maps salient to hypomimia, which leads the encoder and classifier to focus more on areas of hypomimia-interest. In SF-C, the progressive confidence strategy (PCS) ensures more reliable semantic features. Then, a two-stream framework is introduced to fuse the spatial data stream and temporal optical stream, which allows the encoder to semantically and progressively characterize the rigid process of hypomimia. Finally, to improve the interpretability of the model, Gradient-weighted Class Activation Mapping (Grad-CAM) is integrated to generate attention maps that cast our engineered features into hypomimia-interest regions. These highlighted regions provide visual explanations for decisions of our network. Experimental results based on real-world data demonstrate the effectiveness of our method in detecting hypomimia.

Mohammad Shorfuzzaman ◽  
M. Shamim Hossain ◽  
Abdulmotaleb El Saddik

Diabetic retinopathy (DR) is one of the most common causes of vision loss in people who have diabetes for a prolonged period. Convolutional neural networks (CNNs) have become increasingly popular for computer-aided DR diagnosis using retinal fundus images. While these CNNs are highly reliable, their lack of sufficient explainability prevents them from being widely used in medical practice. In this article, we propose a novel explainable deep learning ensemble model where weights from different models are fused into a single model to extract salient features from various retinal lesions found on fundus images. The extracted features are then fed to a custom classifier for the final diagnosis of DR severity level. The model is trained on an APTOS dataset containing retinal fundus images of various DR grades using a cyclical learning rates strategy with an automatic learning rate finder for decaying the learning rate to improve model accuracy. We develop an explainability approach by leveraging gradient-weighted class activation mapping and shapely adaptive explanations to highlight the areas of fundus images that are most indicative of different DR stages. This allows ophthalmologists to view our model's decision in a way that they can understand. Evaluation results using three different datasets (APTOS, MESSIDOR, IDRiD) show the effectiveness of our model, achieving superior classification rates with a high degree of precision (0.970), sensitivity (0.980), and AUC (0.978). We believe that the proposed model, which jointly offers state-of-the-art diagnosis performance and explainability, will address the black-box nature of deep CNN models in robust detection of DR grading.

2021 ◽  
Vol 13 (20) ◽  
pp. 4139
Zhenpeng Feng ◽  
Hongbing Ji ◽  
Ljubiša Stanković ◽  
Jingyuan Fan ◽  
Mingzhe Zhu

Convolutional neural networks (CNNs) have successfully achieved high accuracy in synthetic aperture radar (SAR) target recognition; however, the intransparency of CNNs is still a limiting or even disqualifying factor. Therefore, visually interpreting CNNs with SAR images has recently drawn increasing attention. Various class activation mapping (CAM) methods are adopted to discern the relationship between CNN’s decision and image regions. Unfortunately, most existing CAM methods are based on optical images; thus, they usually lead to a limiting visualization effect for SAR images. Although a recently proposed Self-Matching CAM can obtain a satisfactory effect for SAR images, it is quite time-consuming, due to there being hundreds of self-matching operations per image. G-SM-CAM reduces the time of such operation dramatically, but at the cost of visualization effect. Based on the limitations of the above methods, we propose an efficient method, Spectral-Clustering Self-Matching CAM (SC-SM CAM). Spectral clustering is first adopted to divide feature maps into groups for efficient computation. In each group, similar feature maps are merged into an enhanced feature map with more concentrated energy in a specific region; thus, the saliency heatmaps may more accurately tally with the target. Experimental results demonstrate that SC-SM CAM outperforms other SOTA CAM methods in both effect and efficiency.

Diagnostics ◽  
2021 ◽  
Vol 11 (10) ◽  
pp. 1887
Wentao Zhao ◽  
Wei Jiang ◽  
Xinguo Qiu

As the COVID-19 pandemic continues to ravage the world, the use of chest X-ray (CXR) images as a complementary screening strategy to reverse transcription-polymerase chain reaction (RT-PCR) testing continues to grow owing to its routine clinical application to respiratory diseases. We performed extensive convolutional neural network (CNN) fine-tuning experiments and identified that models pretrained on larger out-of-domain datasets show an improved performance. This suggests that a priori knowledge of models from out-of-field training should also apply to X-ray images. With appropriate hyperparameters selection, we found that higher resolution images carry more clinical information, and the use of mixup in training improved the performance of the model. The experimental showed that our proposed transfer learning present state-of-the-art results. Furthermore, we evaluated the performance of our model with a small amount of downstream training data and found that the model still performed well in COVID-19 identification. We also explored the mechanism of model detection using a gradient-weighted class activation mapping (Grad-CAM) method for CXR imaging to interpret the detection of radiology images. The results helped us understand how the model detects COVID-19, which can be used to discover new visual features and assist radiologists in screening.

2021 ◽  
Vol 8 ◽  
Zhi Jiang ◽  
Qifang Liu ◽  
Ye Tian ◽  
Yidong Zhao ◽  
Wei Liu ◽  

Background: The origin distribution in right ventricular outflow tract (RVOT) ventricular arrhythmias (VAs), as well as the initial ablation effectiveness of reversed U-curve method and antegrade method, remains unclear.Objectives: To investigate the origin distribution of RVOT-type VAs and compare the initial ablation effectiveness of the two methods.Method: Consecutive patients who had idiopathic RVOT-type VAs were prospectively enrolled. After activation mapping, patients were randomly assigned to supravalvular strategy using the reversed U-curve or subvalvular strategy using the antegrade method. The primary outcome was initial ablation (IA) success, defined as the successful ablation within the first three attempts.Results: Sixty-one patients were enrolled from November 2018 to June 2020. Activation mapping revealed that 34/61 (55.7%) of the earliest ventricular activating (EVA) sites were above the pulmonary valves (PVs). The IA success rate was 25/33 (75.8%) in the patients assigned to supravalvular strategy as compared with 16/28 (57.1%) in those assigned to subvalvular strategy (p = 0.172). Multivariate analysis revealed a substantial and qualitative interaction between the EVA sites and IA strategies (pinteraction < 0.001). Either strategy had a remarkably higher IA success rate in treating its ipsilateral EVA sites than contralateral ones (p < 0.0083).Conclusion: Of the idiopathic RVOT-type VA origins, half were located above the PV. The supravalvular and subvalvular strategies did not differ in IA success rates. However, they were complementary to reveal the EVA sites and facilitate ipsilateral ablation, which produces a significantly higher IA success rate.Clinical Trial Registration: Chinese Clinical Trial Registry number,, ChiCTR2000029331.

2021 ◽  
Vol 2021 ◽  
pp. 1-5
Moritz Nies ◽  
Ruben Schleberger ◽  
Leon Dinshaw ◽  
Andreas Rillig ◽  
Andreas Metzner ◽  

We report the case of an 80-year-old female presenting with polymorphic premature ventricular contractions, nonischemic cardiomyopathy, and severe, secondary mitral regurgitation. Despite a low intraprocedural PVC burden, activation mapping and successful ablation of different morphologies were achieved using a novel mapping tool, which facilitates simultaneous mapping of different PVC morphologies.

2021 ◽  
Vol 11 ◽  
Yuzhang Tao ◽  
Xiao Huang ◽  
Yiwen Tan ◽  
Hongwei Wang ◽  
Weiqian Jiang ◽  

BackgroundHistopathological diagnosis of bone tumors is challenging for pathologists. We aim to classify bone tumors histopathologically in terms of aggressiveness using deep learning (DL) and compare performance with pathologists.MethodsA total of 427 pathological slides of bone tumors were produced and scanned as whole slide imaging (WSI). Tumor area of WSI was annotated by pathologists and cropped into 716,838 image patches of 256 × 256 pixels for training. After six DL models were trained and validated in patch level, performance was evaluated on testing dataset for binary classification (benign vs. non-benign) and ternary classification (benign vs. intermediate vs. malignant) in patch-level and slide-level prediction. The performance of four pathologists with different experiences was compared to the best-performing models. The gradient-weighted class activation mapping was used to visualize patch’s important area.ResultsVGG-16 and Inception V3 performed better than other models in patch-level binary and ternary classification. For slide-level prediction, VGG-16 and Inception V3 had area under curve of 0.962 and 0.971 for binary classification and Cohen’s kappa score (CKS) of 0.731 and 0.802 for ternary classification. The senior pathologist had CKS of 0.685 comparable to both models (p = 0.688 and p = 0.287) while attending and junior pathologists showed lower CKS than the best model (each p < 0.05). Visualization showed that the DL model depended on pathological features to make predictions.ConclusionDL can effectively classify bone tumors histopathologically in terms of aggressiveness with performance similar to senior pathologists. Our results are promising and would help expedite the future application of DL-assisted histopathological diagnosis for bone tumors.

Shijie Zhou ◽  
Amir AbdelWahab ◽  
John L. Sapp ◽  
Eric Sung ◽  
Konstantinos N. Aronis ◽  

Background We have previously developed an intraprocedural automatic arrhythmia‐origin localization (AAOL) system to identify idiopathic ventricular arrhythmia origins in real time using a 3‐lead ECG. The objective was to assess the localization accuracy of ventricular tachycardia (VT) exit and premature ventricular contraction (PVC) origin sites in patients with structural heart disease using the AAOL system. Methods and Results In retrospective and prospective case series studies, a total of 42 patients who underwent VT/PVC ablation in the setting of structural heart disease were recruited at 2 different centers. The AAOL system combines 120‐ms QRS integrals of 3 leads (III, V2, V6) with pace mapping to predict VT exit/PVC origin site and projects that site onto the patient‐specific electroanatomic mapping surface. VT exit/PVC origin sites were clinically identified by activation mapping and/or pace mapping. The localization error of the VT exit/PVC origin site was assessed by the distance between the clinically identified site and the estimated site. In the retrospective study of 19 patients with structural heart disease, the AAOL system achieved a mean localization accuracy of 6.5±2.6 mm for 25 induced VTs. In the prospective study with 23 patients, mean localization accuracy was 5.9±2.6 mm for 26 VT exit and PVC origin sites. There was no difference in mean localization error in epicardial sites compared with endocardial sites using the AAOL system (6.0 versus 5.8 mm, P =0.895). Conclusions The AAOL system achieved accurate localization of VT exit/PVC origin sites in patients with structural heart disease; its performance is superior to current systems, and thus, it promises to have potential clinical utility.

2021 ◽  
Vol 13 (19) ◽  
pp. 3985
Minsoo Park ◽  
Dai Quoc Tran ◽  
Seungsoo Lee ◽  
Seunghee Park

Given the explosive growth of information technology and the development of computer vision with convolutional neural networks, wildfire field data information systems are adopting automation and intelligence. However, some limitations remain in acquiring insights from data, such as the risk of overfitting caused by insufficient datasets. Moreover, most previous studies have only focused on detecting fires or smoke, whereas detecting persons and other objects of interest is equally crucial for wildfire response strategies. Therefore, this study developed a multilabel classification (MLC) model, which applies transfer learning and data augmentation and outputs multiple pieces of information on the same object or image. VGG-16, ResNet-50, and DenseNet-121 were used as pretrained models for transfer learning. The models were trained using the dataset constructed in this study and were compared based on various performance metrics. Moreover, the use of control variable methods revealed that transfer learning and data augmentation can perform better when used in the proposed MLC model. The resulting visualization is a heatmap processed from gradient-weighted class activation mapping that shows the reliability of predictions and the position of each class. The MLC model can address the limitations of existing forest fire identification algorithms, which mostly focuses on binary classification. This study can guide future research on implementing deep learning-based field image analysis and decision support systems in wildfire response work.

2021 ◽  
Eric S. Ho ◽  
Zhaoyi Ding

AbstractBackground and purposesStroke is the second leading cause of death globally after ischemic heart disease, also a risk factor of cardioembolic stroke. Thus, we postulate that heartbeats encapsulate vital signals related to stroke. With the rapid advancement of deep neural networks (DNNs), it emerges as a powerful tool to decipher intriguing heartbeat patterns associated with post-stroke patients. In this study, we propose the use of a one-dimensional convolutional network (1D-CNN) architecture to build a binary classifier that distinguishes electrocardiogram s (ECGs) between the post-stroke and the stroke-free.MethodsWe have built two 1D-CNNs that were used to identify distinct patterns from an openly accessible ECG dataset collected from elderly post-stroke patients. In addition to prediction accuracy, which is the primary focus of existing ECG deep neural network methods, we have utilized Gradient-weighted Class Activation Mapping (GRAD-CAM) to ease model interpretation by uncovering ECG patterns captured by our model.ResultsOur stroke model has achieved ∼90% accuracy and 0.95 area under the Receiver Operating Characteristic curve. Findings suggest that the core PQRST complex alone is important but not sufficient to differentiate the post-stroke and the stroke-free.ConclusionsWe have developed an accurate stroke model using the latest DNN method. Importantly, our work has illustrated an approach to enhance model interpretation, overcoming the black-box issue facing DNN, fostering higher user confidence and adoption of DNN in medicine.

Sign in / Sign up

Export Citation Format

Share Document