The NerveStitcher demonstrated that a set of in vivo confocal microscopy (IVCM) images can be merged under a framework of graph convolutional neural network. However, the high similarity of the nerval structure on IVCM image results in mis-stitching when the internal images in sequence happen to drop. Particularly on large gap between a pair of adjacent images, NerveStitcher sometimes cannot detected. In this paper, we advance the concept of global optical flow of IVCM image pair and intergrate it to the existing framework. The large improvements in algorithm robustness are caused by that we make a trade of global displacement of the images sequence according to optical flow, and the local corresponding of key points. We firstly analyze the global shifts in pixel intensity and position between consecutive images, which allows us to determine the motion of each pixel. The displacement vector in the optical flow is then used to calculate the displacement distance of the image. After obtaining the displacement distance of each image using the optical flow, we then use NerveStitcher to stitch the same image sequence. By using a feature point matching algorithm, we can calculate the displacement distance of matching feature points in the two images. We then subtract the displacement distances obtained by the two methods and locate and modify the incorrect stitching results based on the difference value range. Experimental results showed that the improved algorithm, named NerveStitcher2.0, decreases the estimation error by more than 25%. The implemented code is available at https://github.com/better-77/nervestitcher2.0.
Lung cancer is a significant cause of cancer-related deaths globally. X-ray image has been widely used for first-stage screening as it is affordable and widely available. Recently, with the development of the gas sensor IC chip, low-cost enose sensing exhaled breath from patients can potentially be used for the first-stage screening in the near future. We propose a share-embedding model combining x-ray images and enose sensory signals to diagnose lung cancer. Our model contains two branches: the image branch and the enose branch. Since the lack of the enose data, we try to use the pretrained image model to guide the enose branch to align toward the embedding space that the image model learned. Our share-embedding model is designed to be robust to domain shifts across devices and environments. To further improve performance, we use semi-supervised learning with instance weighting to transfer the model to the unlabeled target domain. To train and evaluate the performance, we collect the first paired X-ray images and enose data across multiple devices and clinical environments. In the experiments, our method outperforms each individual branch and a feature concatenation fusion method. In the cross-device setting, our method leveraging semi-supervised learning achieves the best performance.
Gibbs-ringing artifact is a common artifact in medical image processing. It is usually caused by image enhancement procedures, like image deblurring, or physically caused in magnetic resonance imaging (MRI), where the k-space is a finite sampling of underlying signals. Under-sampling in Fourier space is often employed to reduce imaging time for real-time applications. In addition, ringing effect is regularly accompanied by noise boosting due to a low-grade sensor, resulting in lowering image quality to a large extent. In this work we propose hybrid approach for suppressing the stated artifacts, which utilizes recently proposed deep learning Fourier Neural Operator over the classical solvers: Perona-Malik diffusion and Kellner algorithm.
Driver distraction has been one of the primary causes of traffic accidents. Electroencephalography (EEG), a record of the electric potential from the scalp, is considered as a reliable indicator of brain activities. It has been widely used to detect driver distraction. Previous studies have analyzed driver distraction based on time and frequency domain features of EEG. However, challenges still exist in manifesting the distraction information of EEG which contains a large amount of complex information about driver distraction in realistic driving scenarios from the perspective of complexity. In this paper, we propose a driver distraction detection framework using Random Forest (RF) based on the complexity feature fusion of EEG in real driving environment. Five entropy-based features of EEG are firstly extracted with a sliding window. Then, an RF classifier is trained with the extracted features to detect driver distraction. Our results show that differential entropy (DE) with an accuracy of 72.9% achieves the best result while single type feature is applied to detect distraction. The classifier's accuracy is further increased by about 7% using fused multiple features compared with the highest accuracy obtained by single type feature. In terms of feature contribution, we found that the feature with the best distraction detection result by using single type features may not contribute the most when using fused multiple features.
Vision transformer models began gaining recognition alongside NLP. However, their performance compared to Convolutional Neural Network (CNN) models in this domain still requires more significant investigations. Hence, this article comprehensively analyzes their impact and effectiveness against CNN models in medical imaging. We conducted experiments using ViT, MobileViT, and Swin transformers against a pure CNN ConvNeXt trained on Magnetic Resonance Image (MRI) scans. While our findings show promising advancements in imaging with transformers, we observed challenges in their scalability and deployment due to their cost and complexity. We also noticed that they require more medical data, specifically higher-quality MRI scans, when considering better reliability. Nonetheless, comparing their performance in terms of accuracy, even with such limitations, these visual transformer models have shown better detection and diagnosis of brain tumors in MRI scans compared to pure CNN models selected in this study.
Depression is one of the major mental disorders. Depression screening is a crucial part of treatment. EEG signals can detect depression using machine learning and overcome the limitations of traditional screening techniques. The EEG signal is nonstationary and can easily get corrupted by artifacts. To improve the signal quality and get proper analysis the artifact must be removed from EEG signals. In this study, the effect of artifact removal in Machine Learning Based Depression Screening using EEG. Here the data of 32 university students EEG data was used where 19 were depressed and 13 had no depression. The data was processed using a novel artifact removal algorithm. Then the Hjorth parameters were extracted and the SVM classifier was applied. From the analysis, it was observed that the classification accuracy improved by 8.13% after applying the artifact removal and the highest classification performance from the left frontal channels.
The high prevalence of late stage colorectal cancer underscores the need for robust detection systems capable of mitigating its progression during its early stages. While routine colonoscopies have been the industry-standard for identifying signs of early colorectal cancer, it is crucial to uphold several key quality benchmarks to ensure their effectiveness and precision. These quality indices include factors like the scope withdrawal rate and bowel preparation, among others. Our approach leverages on image processing and deep learning to establish a supportive system that highlights areas requiring improvement during scope procedures for clinical practitioners. We demonstrate this via a fine-tuned ResNet-50 architecture to assess bowel preparation yielding 98.5% average accuracy, and a curvature-tracking based approach for colonic anatomical localization for precise monitoring of the withdrawal speed and bowel preparation. We show a pilot iteration of this integrated system on pre-recorded colonoscopy videos, and propose steps for further clinical testing.
This research focuses on the development of an innovative system for detecting instances of patients falling out of bed using advanced image processing and artificial intelligence techniques. By analyzing real-time images from a webcam placed in the patient's bed area, the system identifies specific key points on the patient's body to assess fall risks. The study involved testing twelve sleep patterns representing safe sleeping positions and six patterns simulating scenarios with a risk of falling. The system demonstrated remarkable accuracy in detecting fall risks, triggering warnings and alarms when key points associated with falling were not detected. Importantly, the system avoided false alarms caused by incorrect detections. These findings contribute to the improvement of bed exit alarm systems and have significant implications for patient safety.
Spondyloarthritis and rheumatoid arthritis are considered as two different arthritis autoimmune diseases because spondyloarthritis does not show the presence of rheumatoid factor and anti-CCP antibodies. Although the pathogenesis, clinical picture and genetic factors of these diseases are different, they have similar clinical manifestations and can be overlap. A 48-year-old woman came to the rheumatology outpatient clinic at Dr. Soetomo Hospital Surabaya with the main complaint of pain, stiffness and limited joint movement in almost all parts of the body. The patient met the criteria of Inflammatory Back Pain and Radiographic axial spondyloarthritis according to Assessment of Spondyloarthritis International Society (ASAS). Unexpected of laboratory result shows that HLA-B27 is negative but reactive rheumatoid factor and high titer of Anti-CCP was found . This case report is interesting because MRI assessment is needed to confirm the pathological process of arthritis
Body constitution of traditional Chinese medicine(TCM) can be used to guide the prevention, diagnosis, treatment, rehabilitation and health preservation of diseases. At present, the identification of damp-heat constitution and balanced constitution is mostly determined by questionnaire, which leads to the great influence of subjective factors. Therefore, we propose an effective classification model DenseNet-CBAM for identifying damp-heat constitution and balanced constitution, which can help doctors objectively identify TCM constitution. By adding the Convolutional Attention Block Module (CBAM) to the DenseNet network, the feature extraction ability of the network can be effectively improved. We preprocess 700 voice data of 34 subjects to obtain the corresponding Mel chromatograms, and use ImageNet and AudioSet pre-trained DenseNet-CBAM model to classify them. The accuracy of our method is 82.69 %, which higher than AlexNet, ResNet and DenseNet, respectively. It can be seen that our method can improve the efficiency of constitution identification and promote the objectification of constitution identification.
In liver vascular intervention surgery, the absence or narrowing vascular images may impact the understanding of patients’ vascular structure and morphology. Repairing and analyzing vascular images is a crucial task in computer-aided diagnosis and surgery for minimally invasive vascular diseases. Previous research has addressed the problem of repairing small areas with regular shapes or minor curvature changes in the holes, often leading to modifications to the existing points and encounter noise and geometric loss. Unlike existing point cloud completion algorithms, this paper aims to address the issue of missing vessels in liver vascular images that have not been segmented. It employs axis-aligned bounding boxes to extract missing vascular images. Based on the geometric characteristics of the liver vascular centerline, two methods of cross-section and spherical wave propagation are proposed for repair process. These methods take localized missing vascular images as input and preserve the complete topological structure of the vessels. The repair results are evaluated using the Chamfer distance and Dice similarity coefficient. Experimental results demonstrate that the proposed approach effectively repairs large-scale, non-closed missing vascular images in liver vascular images and generates complete vessel geometric models with accurate liver vascular structure and morphology. The validity of the vessel models is confirmed through hemodynamic analysis, providing robust support for medical image analysis and diagnosis.
Automated classification of Whole Slide Images (WSIs) is of great significance for early diagnosis of cancer. Existing approaches are trained on a specific level which affects the analysis performance due to weak supervision of patches and variants. Additionally, it is difficult to distinguish cancer subtype patches accurately from different magnification levels of WSIs. However, this can be improved by employing artificial intelligence models to address these problems, we propose a novel clustering-based cancer diagnosis (CBCD) method for WSI classification. The CBCD constructs three modules: first, we extracted patches from each magnification level of WSIs with respective cancer sub-types. Second, we employed two features (global and local) to learn discriminative and salient information of each patch. Then we find the meaningful cluster regions based on these features to quantify (select) the best patches of salient cancer subtypes by only relying on the collective characteristics of patches from different magnification levels. The clustering techniques used are k-means, gaussian mixture model, and agglomerative clustering. The quality of each clustering technique was determined using adjusted rand, and calinski harabasz scores. Later we used five state-of-the-art (SOTA) deep learning models to learn and classify cancer subtype regions of WSIs based on two types of features of patches. We also showed the results with no clustering techniques in an end-to-end supervised way by directly extracting patches from WSIs. Our method is evaluated on the public WSI dataset (KBSMC) for cancer sub-types classification and achieves better performance and great interpretability compared with the SOTA methods.
In view of the common problems in image saliency detection, such as inaccurate positioning of saliency objects and easy loss of detail information in complex scenario. This paper proposes a saliency detection based on feature fusion and weighted hypergraph. Firstly, SLIC superpixel segmentation is performed to extract the color features, spatial features and deep features of the image, and a weighted hypergraph model is constructed. Then the saliency score of each superpixel block is obtained by using the random walk algorithm, the primary saliency map is generated according to the order of each superpixel score. Compared with seven advanced algorithms on three challenging datasets (ECSSD, HKU-IS, PASCAL-S), the proposed model significantly improves the performance of saliency map details and salient object localization under complex scenario.
The rapid increase in the skin cancer incidence has become a major concern worldwide. The early diagnosis and effective treatment play a vital role in the efficacy of skin cancer treatment. Therefore, an automated diagnosis system would assist the experts in timely diagnosing the skin cancer. This research proposes a hybrid CNN and transformer based method to distinguish between dermoscopic images of Melanoma and Nonmelanoma (Nevus). The proposed SkinCaNet method incorporates feed forward network (FFN) blocks, multi-head self attention (MHSA) blocks and multi-layer perceptron (MLP) head to efficiently capture both the local and global features in order to enhance the accuracy of Melanoma and Nonmelanoma classification. The proposed SkinCaNet model is trained from scratch on HAM10000 dataset and achieved accuracy of 92.38% whereas it achieved precision of 95.62%, recall of 82.92% and specificity of 97.82%. The proposed method demonstrated superiority over SOTA models and achieved better accuracy in comparison to other works in the literature for classifying Melanoma and Nonmelanoma.
Anxiety is a mental disorder that leads to palpitation, chest pain, behavioral abnormalities, etc. There are 301 million sufferers of anxiety disorders worldwide. Anxiety must be treated properly thus anxiety screening should be done more efficiently. EEG can detect abnormalities in the brain caused due to anxiety disorder. So, it can be used for screening anxiety. It may be able to overcome the drawbacks of conventional screening techniques. A machine learning-based method for screening anxiety among university-going students using wireless EEG is proposed in this study. EEG was recorded using a wireless 14-channel EEG headset from 40 students aged between 18-25 years. After using GAD-7 for screening the participants the EEG data of the participants was divided into anxiety group and anxiety control group. The data was filtered into 6 frequency bands. After extracting some nonlinear features, the SVM classifier with 10-fold cross-validation was used. Among all the bands the beta band (12-30Hz) had the highest accuracy of 94.88% with the Precision of 94.4%, NPV of 95.6%, Sensitivity of 97.2%, Specificity of 94.1%, and 0.96 F1 score.
Current research in digital pathology uses evermore complex computational and imaging tools to improve diagnostic efficiency and solve a variety of scientific tasks. Also, digital pathology tools are actively used in medical education and can improve the efficiency of training students and residents. Despite the active development of software tools in the field of digital pathology, the existing solutions still require a high entry threshold, are not truly universal and have complex and outdated interfaces. In this paper we present PathScribe, which is a new cloud-based software for working with whole slide histological images that simplifies the interaction with gigapixel slides and is designed to be used for educational and research purposes. This new software allows to comfortably access and work with large collections of histological images from almost any device using internet connection. In this work we describe the architecture and working principles of PathScribe software and provide the real world examples of its usage by pathologists both for education and research purposes.
Deep learning-based methods have made significant progress in image tampering detection. However, they often overlook the characteristics of small-size tampered regions, leading to missed or false detections. To address these issues, this paper proposes a novel approach for image splicing tampering detection using Faster R-CNN, which incorporates multi-layer feature refinement fusion. This approach aims to enhance the expression of features by employing a multi-layer feature refinement fusion strategy, thereby improving the accuracy of detecting tampered regions in small areas. The proposed method utilizes the RGB image as the input for the color image channel and employs the SRM filter processed image for the steganographic analysis channel. These two channels are fused in a refinement fusion network, combining deep refinement features with shallow refinement features. This fusion effectively utilizes refined feature information from different levels. The extracted features from the two channels are further fused using a bilinear pooling layer. Additionally, the RGB channels adaptively adjust the anchor frame shapes to enhance the accuracy of detecting splicing tampering in small areas. Experimental results demonstrate that the proposed method achieves im-proved detection accuracy compared to the original approach on publicly available datasets. It effectively detects regions affected by image splicing tampering. The findings highlight the potential of the proposed method in addressing the challenges associated with small-size tampered regions, contributing to the advancement of image tampering detection.
This paper proposes a novel structure based on Convolutional Neural Network (CNN) to identify magnetic resonance images (MRI) of Alzheimer's disease. The model has a strong generalization ability and can help doctors to identify the patient's state. The advantages of this model are that it has a simple structure and high accuracy, which can quickly make a high-confidence classification judgment on the input image, and can be applied to hardware-level design. The paper also compares the differences between this method and traditional image processing and classification methods, showing the advantages of deep methods.
The global COVID-19 pandemic has caused a health crisis globally. Automated diagnostic methods can control the spread of the pandemic, as well as assists physicians to tackle high workload conditions through the quick treatment of affected patients. Owing to the scarcity of medical images and from different resources, the present image heterogeneity has raised challenges for achieving effective approaches to network training and effectively learning robust features. We propose a multi-joint unit network for the diagnosis of COVID-19 using the joint unit module, which leverages the receptive fields from multiple resolutions for learning rich representations. Existing approaches usually employ a large number of layers to learn the features, which consequently requires more computational power and increases the network complexity. To compensate, our joint unit module extracts low-, same-, and high-resolution feature maps simultaneously using different phases. Later, these learned feature maps are fused and utilized for classification layers. We observed that our model helps to learn sufficient information for classification without a performance loss and with faster convergence. We used three public benchmark datasets to demonstrate the performance of our network. Our proposed network consistently outperforms existing state-of-the-art approaches by demonstrating better accuracy, sensitivity, and specificity and F1-score across all datasets.
In this work, the auxiliary bladder filling degree perception function is established for patients with impaired bladder sensory function to help patients improve their quality of life. This study intends to build a 3D finite element model of the lower abdomen of the human body by modeling CT images. The simulation experiments based on the three-dimensional finite element model aim to determine the placement of the double electrodes and to explore the effect of different signal frequencies on the sensitivity of bladder filling degree. The results of the simulation experiments were then verified by recording the changes in electrical impedance during the natural bladder filling process of one healthy volunteer after 550ml of water was drunk. The electrical impedance measured in the human experiments has a negative correlation with the bladder filling degree in time accumulation, which is consistent with the simulation experiments results. The results of this study verify the possibility of finding the best electrode placement point in the lower abdomen of the human body through simulation experiments and human experiments. And The sensitivity of measuring bladder filling degree is better at the frequency of 1-10Khz.
