Objective To automatically segment diabetic retinal exudation features from deep learning color fundus images. Methods An applied study. The method of this study is based on the U-shaped network model of the Indian Diabetic Retinopathy Image Dataset (IDRID) dataset, introduces deep residual convolution into the encoding and decoding stages, which can effectively extract seepage depth features, solve overfitting and feature interference problems, and improve the model's feature expression ability and lightweight performance. In addition, by introducing an improved context extraction module, the model can capture a wider range of feature information, enhance the perception ability of retinal lesions, and perform excellently in capturing small details and blurred edges. Finally, the introduction of convolutional triple attention mechanism allows the model to automatically learn feature weights, focus on important features, and extract useful information from multiple scales. Accuracy, recall, Dice coefficient, accuracy and sensitivity were used to evaluate the ability of the model to detect and segment the automatic retinal exudation features of diabetic patients in color fundus images. Results After applying this method, the accuracy, recall, dice coefficient, accuracy and sensitivity of the improved model on the IDRID dataset reached 81.56%, 99.54%, 69.32%, 65.36% and 78.33%, respectively. Compared with the original model, the accuracy and Dice index of the improved model are increased by 2.35% , 3.35% respectively. Conclusion The segmentation method based on U-shaped network can automatically detect and segment the retinal exudation features of fundus images of diabetic patients, which is of great significance for assisting doctors to diagnose diseases more accurately.
Recent studies have introduced attention models for medical visual question answering (MVQA). In medical research, not only is the modeling of “visual attention” crucial, but the modeling of “question attention” is equally significant. To facilitate bidirectional reasoning in the attention processes involving medical images and questions, a new MVQA architecture, named MCAN, has been proposed. This architecture incorporated a cross-modal co-attention network, FCAF, which identifies key words in questions and principal parts in images. Through a meta-learning channel attention module (MLCA), weights were adaptively assigned to each word and region, reflecting the model’s focus on specific words and regions during reasoning. Additionally, this study specially designed and developed a medical domain-specific word embedding model, Med-GloVe, to further enhance the model’s accuracy and practical value. Experimental results indicated that MCAN proposed in this study improved the accuracy by 7.7% on free-form questions in the Path-VQA dataset, and by 4.4% on closed-form questions in the VQA-RAD dataset, which effectively improves the accuracy of the medical vision question answer.
Electrocardiogram (ECG) can visually reflect the physiological electrical activity of human heart, which is important in the field of arrhythmia detection and classification. To address the negative effect of label imbalance in ECG data on arrhythmia classification, this paper proposes a nested long short-term memory network (NLSTM) model for unbalanced ECG signal classification. The NLSTM is built to learn and memorize the temporal characteristics in complex signals, and the focal loss function is used to reduce the weights of easily identifiable samples. Then the residual attention mechanism is used to modify the assigned weights according to the importance of sample characteristic to solve the sample imbalance problem. Then the synthetic minority over-sampling technique is used to perform a simple manual oversampling process on the Massachusetts institute of technology and Beth Israel hospital arrhythmia (MIT-BIH-AR) database to further increase the classification accuracy of the model. Finally, the MIT-BIH arrhythmia database is applied to experimentally verify the above algorithms. The experimental results show that the proposed method can effectively solve the issues of imbalanced samples and unremarkable features in ECG signals, and the overall accuracy of the model reaches 98.34%. It also significantly improves the recognition and classification of minority samples and has provided a new feasible method for ECG-assisted diagnosis, which has practical application significance.
To assist grassroots sonographers in accurately and rapidly detecting intussusception lesions from children's abdominal ultrasound images, this paper proposes an improved YOLOv8n children's intussusception detection algorithm, called EMC-YOLOv8n. Firstly, the EfficientViT network with a cascaded group attention module was used as the backbone network to enhance the speed of target detection. Secondly, the improved C2fMBC module was used to replace the C2f module in the neck network to reduce network complexity, and the coordinate attention (CA) module was introduced after each C2fMBC module to enhance attention to positional information. Finally, experiments were conducted on the self-built dataset of intussusception in children. The results showed that the recall rate, average detection accuracy (mAP@0.5) and precision of the EMC-YOLOv8n algorithm improved by 3.9%, 2.1% and 0.9%, respectively, compared to the baseline algorithm. Despite slightly increased network parameters and computational load, significant improvements in detection accuracy enable efficient completion of detection tasks, demonstrating substantial economic and social value.
In order to realize the quantitative assessment of muscle strength in hand function rehabilitation and then formulate scientific and effective rehabilitation training strategies, this paper constructs a multi-scale convolutional neural network (MSCNN) - convolutional block attention module (CBAM) - bidirectional long short-term memory network (BiLSTM) muscle strength prediction model to fully explore the spatial and temporal features of the data and simultaneously suppress useless features, and finally achieve the improvement of the accuracy of the muscle strength prediction model. To verify the effectiveness of the model proposed in this paper, the model in this paper is compared with traditional models such as support vector machine (SVM), random forest (RF), convolutional neural network (CNN), CNN - squeeze excitation network (SENet), MSCNN-CBAM and MSCNN-BiLSTM, and the effect of muscle strength prediction by each model is investigated when the hand force application changes from 40% of the maximum voluntary contraction force (MVC) to 60% of the MVC. The research results show that as the hand force application increases, the effect of the muscle strength prediction model becomes worse. Then the ablation experiment is used to analyze the influence degree of each module on the muscle strength prediction result, and it is found that the CBAM module plays a key role in the model. Therefore, by using the model in this article, the accuracy of muscle strength prediction can be effectively improved, and the characteristics and laws of hand muscle activities can be deeply understood, providing assistance for further exploring the mechanism of hand functions.
The aim of this study is to evaluate the effect of laparoscopic simulation training with different attention. Attention was appraised using the sample entropy and θ/β value, which were calculated according to electroencephalograph (EEG) signal collected with BrainLink. The effect of laparoscopic simulation training was evaluated using the completion time, error number and fixation number, which were calculated according to eye movement signal collected with Tobii eye tracker. Twenty volunteers were recruited in this study. Those with the sample entropy lower than 0.77 were classified into group A and those higher than 0.77 into group B. The results showed that the sample entropy of group A was lower than that of group B, and fluctuations of A were more steady. However, the sample entropy of group B showed steady fluctuations in the first five trainings, and then demonstrated relatively dramatic fluctuates in the later five trainings. Compared with that of group B, the θ/β value of group A was smaller and shows steady fluctuations. Group A has a shorter completion time, less errors and faster decrease of fixation number. Therefore, this study reached the following conclusion that the attention of the trainees would affect the training effect. Members in group A, who had a higher attention were more efficient and faster training. For those in group B, although their training skills have been improved, they needed a longer time to reach a plateau.
Skin cancer is a significant public health issue, and computer-aided diagnosis technology can effectively alleviate this burden. Accurate identification of skin lesion types is crucial when employing computer-aided diagnosis. This study proposes a multi-level attention cascaded fusion model based on Swin-T and ConvNeXt. It employed hierarchical Swin-T and ConvNeXt to extract global and local features, respectively, and introduced residual channel attention and spatial attention modules for further feature extraction. Multi-level attention mechanisms were utilized to process multi-scale global and local features. To address the problem of shallow features being lost due to their distance from the classifier, a hierarchical inverted residual fusion module was proposed to dynamically adjust the extracted feature information. Balanced sampling strategies and focal loss were employed to tackle the issue of imbalanced categories of skin lesions. Experimental testing on the ISIC2018 and ISIC2019 datasets yielded accuracy, precision, recall, and F1-Score of 96.01%, 93.67%, 92.65%, and 93.11%, respectively, and 92.79%, 91.52%, 88.90%, and 90.15%, respectively. Compared to Swin-T, the proposed method achieved an accuracy improvement of 3.60% and 1.66%, and compared to ConvNeXt, it achieved an accuracy improvement of 2.87% and 3.45%. The experiments demonstrate that the proposed method accurately classifies skin lesion images, providing a new solution for skin cancer diagnosis.
In clinical, manually scoring by technician is the major method for sleep arousal detection. This method is time-consuming and subjective. This study aimed to achieve an end-to-end sleep-arousal events detection by constructing a convolutional neural network based on multi-scale convolutional layers and self-attention mechanism, and using 1 min single-channel electroencephalogram (EEG) signals as its input. Compared with the performance of the baseline model, the results of the proposed method showed that the mean area under the precision-recall curve and area under the receiver operating characteristic were both improved by 7%. Furthermore, we also compared the effects of single modality and multi-modality on the performance of the proposed model. The results revealed the power of single-channel EEG signals in automatic sleep arousal detection. However, the simple combination of multi-modality signals may be counterproductive to the improvement of model performance. Finally, we also explored the scalability of the proposed model and transferred the model into the automated sleep staging task in the same dataset. The average accuracy of 73% also suggested the power of the proposed method in task transferring. This study provides a potential solution for the development of portable sleep monitoring and paves a way for the automatic sleep data analysis using the transfer learning method.
This study aims to explore the differences of event related potential (ERP) between attention deficit hyperactivity disorder (ADHD) and normal children, so that these differences provide scientific basis for the diagnosis of ADHD. Eight children were identified to be ADHD group by the diagnostic criteria of DSM IV (diagnostic and statistical manual of mental disorders IV), and the control group also consisted of 8 normal children. Modified visual continuous performance test (CPT) was used as the experiment paradigm. The experiment included two major conditions, i.e. Go and NoGo. All the 16 subjects participated in the study. A high density EEG acquisition instrument was used to record the EEG signal and processed these EEG data by means of ERP and spectrum analysis. P2 N2 peak peak value and spectral peak around 11 Hz were analyzed between ADHD subjects and those in the control group, and then statistical tests were applied to these two groups. Results showed that: ① Under the condition of Go, ADHD group had a significant lower P2 N2 peak peak value than the values in the control group ( P< 0.05); but under the condition of NoGo there was no significant difference in between. ② Compared with the control group, the ADHD group had significant lower spectral amplitude around 11 Hz under the condition of NoGo ( P< 0.05). However, under the condition of Go the difference was insignificant. In conclusion, there is certain cognitive dysfunction in ADHD children. P2-N2 peak-peak value and spectral peak around 11 Hz could be considered as clinical evaluation indexes of ADHD children′s cognitive function. These two objective indexes provide an early diagnosis and effective treatment of ADHD .
Objective To recognize the different phases of Korotkoff sounds through deep learning technology, so as to improve the accuracy of blood pressure measurement in different populations. Methods A classification model of the Korotkoff sounds phases was designed, which fused attention mechanism (Attention), residual network (ResNet) and bidirectional long short-term memory (BiLSTM). First, a single Korotkoff sound signal was extracted from the whole Korotkoff sounds signals beat by beat, and each Korotkoff sound signal was converted into a Mel spectrogram. Then, the local feature extraction of Mel spectrogram was processed by using the Attention mechanism and ResNet network, and BiLSTM network was used to deal with the temporal relations between features, and full-connection layer network was applied in reducing the dimension of features. Finally, the classification was completed by SoftMax function. The dataset used in this study was collected from 44 volunteers (24 females, 20 males with an average age of 36 years), and the model performance was verified using 10-fold cross-validation. Results The classification accuracy of the established model for the 5 types of Korotkoff sounds phases was 93.4%, which was higher than that of other models. Conclusion This study proves that the deep learning method can accurately classify Korotkoff sounds phases, which lays a strong technical foundation for the subsequent design of automatic blood pressure measurement methods based on the classification of the Korotkoff sounds phases.