Background: Ultra-wide-field (UWF) fundus photography represents an emerging retinal imaging technique offering a broader field of view, thus enhancing its utility in screening and diagnosing various eye diseases, notably diabetic retinopathy (DR). However, the application of computer-aided diagnosis for DR using UWF images confronts two major challenges. The first challenge arises from the limited availability of labeled UWF data, making it daunting to train diagnostic models due to the high cost associated with manual annotation of medical images. Secondly, existing models' performance requires enhancement due to the absence of prior knowledge to guide the learning process. Purpose: By leveraging extensively annotated datasets within the field, which encompass large-scale, high-quality color fundus image datasets annotated at either image-level or pixel-level, our objective is to transfer knowledge from these datasets to our target domain through unsupervised domain adaptation. Methods: Our approach presents a robust model for assessing the severity of diabetic retinopathy (DR) by leveraging unsupervised lesion-aware domain adaptation in ultra-wide-field (UWF) images. Furthermore, to harness the wealth of detailed annotations in publicly available color fundus image datasets, we integrate an adversarial lesion map generator. This generator supplements the grading model by incorporating auxiliary lesion information, drawing inspiration from the clinical methodology of evaluating DR severity by identifying and quantifying associated lesions. Results: We conducted both quantitative and qualitative evaluations of our proposed method. In particular, among the six representative DR grading methods, our approach achieved an accuracy (ACC) of 68.18% and a precision (pre) of 67.43%. Additionally, we conducted extensive experiments in ablation studies to validate the effectiveness of each component of our proposed method. Conclusion: In conclusion, our method not only improves the accuracy of DR grading, but also enhances the interpretability of the results, providing clinicians with a reliable DR grading scheme.
Segmentation of corneal layer interfaces in optical coherence tomography (OCT) images is important for diagnostic and surgical purposes, while manual segmentation is a time-consuming and tedious process. This paper presents a novel technique for the automatic segmentation of corneal layer interfaces using customized initial layer estimation and a gradient-based segmentation method. The proposed method was also extended to three-dimensional OCT images. Validation was performed on two corneal datasets, one with 37 B-scan images of healthy human eyes and the other with a 3D volume scan of a porcine eye. The approach showed robustness in extracting different layer boundaries in the low-SNR region with lower computational cost but higher accuracy compared to existing techniques. It achieved segmentation errors below 2.1 pixels for both the anterior and posterior layer boundaries in terms of mean unsigned surface positioning error for the first dataset and 2.6 pixels (5.2 µm) for segmenting all three layers that can be resolved in the second dataset. On average, it takes 0.7 and 0.4 seconds to process a cross-sectional B-scan image for datasets one and two, respectively. Our comparative study also showed that it outperforms state-of-the-art methods for quantifying layer interfaces in terms of accuracy and time efficiency.
Purpose: Fast and automated reconstruction of retinal hyperreflective foci (HRF) is of great importance for many eye-related disease understanding. In this paper, we introduced a new automated framework, driven by recent advances in deep learning to automatically extract 12 three-dimensional parameters from the segmented hyperreflective foci in optical coherence tomography (OCT). Methods: Unlike traditional convolutional neural networks, which struggle with long-range feature correlations, we introduce a spatial and channel attention module within the bottleneck layer, integrated into the nnU-Net architecture. Spatial Attention Block aggregates features across spatial locations to capture related features, while Channel Attention Block heightens channel feature contrasts. The proposed model was trained and tested on 162 retinal OCT volumes of patients with diabetic macular edema (DME), yielding robust segmentation outcomes. We further investigate HRF's potential as a biomarker of DME. Results: Results unveil notable discrepancies in the amount and volume of HRF subtypes. In the whole retinal layer (WR), the mean distance from HRF to the retinal pigmented epithelium was significantly reduced after treatment. In WR, the improvement in central macular thickness resulting from intravitreal injection treatment was positively correlated with the mean distance from HRF subtypes to the fovea. Conclusion: Our study demonstrates the applicability of OCT for automated quantification of retinal HRF in DME patients, offering an objective, quantitative approach for clinical and research applications.
Trigeminal neuralgia caused by paroxysmal and severe pain in the distribution of the trigeminal nerve is a rare chronic pain disorder. It is generally accepted that compression of the trigeminal root entry zone by vascular structures is the major cause of primary trigeminal neuralgia, and vascular decompression is the prior choice in neurosurgical treatment. Therefore, accurate preoperative modeling/segmentation/visualization of trigeminal nerve and its surrounding cerebrovascular is important to surgical planning. In this paper, we propose an automated method to segment trigeminal nerve and its surrounding cerebrovascular in the root entry zone, and to further reconstruct and visual these anatomical structures in three-dimensional (3D) Magnetic Resonance Angiography (MRA). The proposed method contains a two-stage neural network. Firstly, a preliminary confidence map of different anatomical structures is produced by a coarse segmentation stage. Secondly, a refinement segmentation stage is proposed to refine and optimize the coarse segmentation map. To model the spatial and morphological relationship between trigeminal nerve and cerebrovascular structures, the proposed network detects the trigeminal nerve, cerebrovasculature, and brainstem simultaneously. The method has been evaluated on a dataset including 50 MRA volumes, and the experimental results show the state-of-the-art performance of the proposed method with an average Dice similarity coefficient, Hausdorff distance, and average surface distance error of 0.8645, 0.2414, and 0.4296 on multi-tissue segmentation, respectively.
A graph-based groupwise shape registration algorithm for building statistical shape model (SSM) is proposed, which has been successfully applied to shape prediction of foot scans. Establishing unbiased and effective shape correspondences of large-scale data sets is extremely challenging, for the inappropriate selection of initial mean shape and non-rigid registration of shape with large-scale deformation. To address these issues, first, we use a simplified graph to model the shape distribution in metric space and an edge-guided graph shrinkage to deform the shapes. Then, the groupwise registration is performed by iteratively performing the graph shrinkage until the shape converges. And, the correspondences of training shapes are obtained by propagating the converged shape to the original data along each shrinkage path. Compared with traditional forward and backward models of groupwise registration, the proposed method is data-driven without initial mean shape as input. Moreover, under the constraint of the established graph, the non-rigid registration can perform more accurately by restricting shape register to its neighbors. Based on the shape correspondence, the SSM of foot shapes is constructed and applied to shape prediction by taking the collected anthropometric information as predictor. Experiments demonstrate that the proposed method can obtain robust shape correspondences and SSM capability with respect to model generalization, specificity, and compactness. The application of shape prediction model shows an average prediction error lower than 1% for general foot size. Graphical abstract The graphical abstract of unbiased groupwise registration for foot prediction.
Algorithms , Foot/anatomy & histology , Image Processing, Computer-Assisted , Imaging, Three-Dimensional , Adolescent , Adult , Aged , Female , Humans , Male , Middle Aged , Models, Statistical , Reproducibility of Results , Young Adult
OBJECTIVE: This paper proposes an unobtrusive blood pressure (BP) measurement system design with a motion artifact (MA) compensation strategy as a potential surrogate to the traditional cuff-based sphygmomanometer for self-monitoring in a less restricted environment. APPROACH: A dual-channel photoplethysmographic signal acquisition system is designed and implemented for cuff-less BP measurement based on the peripheral pulse transit time (PPTT) acquired from the forearm and wrist. Comprising a motion decision, singular spectrum analysis, PPTT calculation and BP measurement, a novel approach is proposed to realize BP measurements and suppress MA interference. MAIN RESULTS: Compared with the reference BP recorded by a cuff-based sphygmomanometer, our preliminary examinations on 30 subjects found that the correlation coefficients for systolic BP estimation and diastolic BP estimation were 0.75 and 0.78, and the mean absolute differences were 7.61 mmHg and 6.82 mmHg, respectively. Meanwhile, the proposed approach was compared with the other most widely used pulse transit time (PTT) measuring methods and BP-PTT models. All the results indicate that our work was highly effective in realizing the BP measurement. SIGNIFICANCE: The proposed system and approach have resulted in remarkable progress in cuff-less BP measurements with MA removal, and have great potential value in wearable applications without environmental restrictions.
Blood Pressure Determination/methods , Photoplethysmography , Pulse Wave Analysis , Signal Processing, Computer-Assisted , Adult , Algorithms , Female , Healthy Volunteers , Humans , Male
This paper quantifies the registration and fusion display errors of augmented reality-based nasal endoscopic surgery (ARNES). We comparatively investigated the spatial calibration process for front-end endoscopy and redefined the accuracy level of a calibrated endoscope by using a calibration tool with improved structural reliability. We also studied how registration accuracy was combined with the number and distribution of the deployed fiducial points (FPs) for positioning and the measured registration time. A physically integrated ARNES prototype was customarily configured for performance evaluation in skull base tumor resection surgery with an innovative approach of dynamic endoscopic vision expansion. As advised by surgical experts in otolaryngology, we proposed a hierarchical rendering scheme to properly adapt the fused images with the required visual sensation. By constraining the rendered sight in a known depth and radius, the visual focus of the surgeon can be induced only on the anticipated critical anatomies and vessel structures to avoid misguidance. Furthermore, error analysis was conducted to examine the feasibility of hybrid optical tracking based on point cloud, which was proposed in our previous work as an in-surgery registration solution. Measured results indicated that the error of target registration for ARNES can be reduced to 0.77 ± 0.07â¯mm. For initial registration, our results suggest that a trade-off for a new minimal time of registration can be reached when the distribution of five FPs is considered. For in-surgery registration, our findings reveal that the intrinsic registration error is a major cause of performance loss. Rigid model and cadaver experiments confirmed that the scenic integration and display fluency of ARNES are smooth, as demonstrated by three clinical trials that surpassed practicality.
Endoscopy/methods , Nose Diseases/diagnostic imaging , Nose Diseases/surgery , Surgery, Computer-Assisted/methods , Tomography, X-Ray Computed , User-Computer Interface , Algorithms , Calibration , Humans , Imaging, Three-Dimensional
BACKGROUND: Plantar pressure measurement has become increasingly useful in the evaluation of plantar health conditions thanks to the recent progression in sensing technology. Due to the large volume and high energy consumption of monitoring devices, traditional systems for plantar pressure measurement are only focused on static or short-term dynamic monitoring. It makes them inappropriate for early detections of plantar symptoms usually presented in long-term activities. METHODS: A prototype of monitoring system based on body sensor network (BSN) is proposed for quantitative assessment of plantar conditions. To further assess the severity of plantar symptoms which can be reflected from the pressure distribution in motion status, an approach to conjoint analysis of pressure distribution and exercise load quantification based on the strike frequency (SF) and heart rate (HR) is also proposed. RESULTS: An examination was tested on 30 subjects to verify the capabilities of the proposed system. The estimated correlation rate with reference devices ([Formula: see text]) and error rate on the average ([Formula: see text]) of HR and SF indicated equal measuring capabilities as the existing commercial products . Comprised of the conjoint analysis based on HR and SF, the proposed method of exercise load quantification was examined on all subjects' recordings. CONCLUSIONS: A prototype of an innovative BSN-based bio-physiological measurement system has been implemented for the long-term monitoring and early evaluation of plantar condition. The experimental results indicated that the proposed system has a great potential value in the applications of long-term plantar health monitoring and evaluation.
Exercise/physiology , Foot , Monitoring, Physiologic/instrumentation , Adult , Equipment Design , Female , Health , Heart Rate , Humans , Male , Pressure , Young Adult
Focal cortical dysplasia (FCD) is the main cause of epilepsy and can be automatically detected via magnetic resonance (MR) images. However, visual detection of lesions is time consuming and highly dependent on the doctor's personal knowledge and experience. In this paper, we propose a new framework for positive unanimous voting (PUV) to detect FCD lesions. Maps of gray matter thickness, gradient, relative intensity, and gray/white matter width are computed in the proposed framework to enhance the differences between lesional and non-lesional regions. Feature maps are further compared with the feature distributions of healthy controls to obtain feature difference maps. PUV driven by feature and feature difference maps is then applied to classify image voxels into lesion and non-lesion. The connected region analysis then refines the classification results by removing the tiny fragment regions consisting of falsely classified positive voxels. The proposed method correctly identified 8/10 patients with FCD lesions and 30/31 healthy people. Experimental results on the small FCD samples demonstrated that the proposed method can effectively reduce the number of false positives and guarantee correct detection of lesion regions compared with four single classifiers and two recent methods.
The heart sound signal is a reflection of heart and vascular system motion. Long-term continuous electrocardiogram (ECG) contains important information which can be helpful to prevent heart failure. A single piece of a long-term ECG recording usually consists of more than one hundred thousand data points in length, making it difficult to derive hidden features that may be reflected through dynamic ECG monitoring, which is also very time-consuming to analyze. In this paper, a Dynamic Time Warping based on MapReduce (MRDTW) is proposed to make prognoses of possible lesions in patients. Through comparison of a real-time ECG of a patient with the reference sets of normal and problematic cardiac waveforms, the experimental results reveal that our approach not only retains high accuracy, but also greatly improves the efficiency of the similarity measure in dynamic ECG series.
Algorithms , Arrhythmias, Cardiac/diagnosis , Diagnosis, Computer-Assisted/methods , Electrocardiography/methods , Heart Rate , Pattern Recognition, Automated/methods , Arrhythmias, Cardiac/physiopathology , Humans , Machine Learning , Reproducibility of Results , Sensitivity and Specificity , Signal Processing, Computer-Assisted
BACKGROUND: The widespread access to portable medical devices or new personal devices is boosting the amount of biomedical data. These devices provide a growing massive data that far exceeds the analytical ability of a professional doctor. The computer-assisted analysis of biomedical data has become an essential tool in medicine diagnosis. OBJECTIVE: Due to the advantages of discrete, noise elimination and dimensionality reduction, symbolic representation of biomedical data has attracted great interest. The symbolization results provide efficiently performing at data mining, such as pattern discovery, anomaly detection and association rules mining, so we want to use the method to improving the biomedical data classification. METHODS: In this paper, we introduce a novel symbolic representation method, called Trend Feature Symbolic Approximation (TFSA). RESULTS: The TFSA focuses on retaining most of the original series' trend features, and it also very suitable for subsequent mining work, such as association rules mining. CONCLUSION: The TFSA provides the lower bounding guarantee and the experimental results show that comparing with some existing methods, its classification accuracy is improved.
Electrocardiography/methods , Information Storage and Retrieval/methods , Signal Processing, Computer-Assisted , Algorithms , Humans
The 3rd International Conference on Biomedical Engineering and Biotechnology (iCBEB 2014), held in Beijing from the 25th to the 28th of September 2014, is an annual conference that intends to provide an opportunity for researchers and practitioners around the world to present the most recent advances and future challenges in the fields of biomedical engineering, biomaterials, bioinformatics and computational biology, biomedical imaging and signal processing, biomechanical engineering and biotechnology, amongst others. The papers published in this issue are selected from this conference, which witnesses the advances in biomedical engineering and biotechnology during 2013-2014.
Biomedical Engineering/trends , Biotechnology/trends , Diagnostic Imaging/trends , Forecasting , Nanocapsules , Nanomedicine/trends
A non-linear control method, known as Variable Structure Control (VSC), is employed to reduce the duration of ultrasonic (US) transducer transients. A physically realizable system using a simplified form of the VSC algorithm is proposed for standard piezoelectric transducers and simulated. Results indicate a VSC-controlled transmitter reduces the transient duration to less than a carrier wave cycle. Applications include high capacity ultrasound communication and localization systems.
Algorithms , Image Enhancement/instrumentation , Micro-Electrical-Mechanical Systems/instrumentation , Transducers , Ultrasonography/instrumentation , Computer Simulation , Computer-Aided Design , Equipment Design , Equipment Failure Analysis , Feedback , Models, Theoretical
High-brightness, edge-emitting diode laser arrays integrated with a phase shifter have been designed and fabricated at a wavelength of about 910 nm. Stable out-of-phase mode is generated through coupling evanescently and converted to be nearly in-phase by the phase modulation from the phase shifter. With a very simple manufacture process, stable single-lobe far-field pattern is achieved in the slow axis when the continuous wave output power exceeds 460 mW/facet, and the divergence angle is only 2.7 times the diffraction-limited value. Such device shows a promising future for high-brightness application with low cost and easy fabrication.
In this Letter, a III-V/silicon hybrid single-mode laser operating at C band for photonic integration circuit is presented. The InGaAlAs gain structure is bonded onto a patterned silicon-on insulator through wafer to wafer directly. The mode selected mechanism based on a hybrid III-V/silicon straight cavity with periodic microstructures is applied, which only need low cost i-line projection photolithography in the whole technological process. At room temperature, we obtain 0.62 mW output power in continuous-wave. The side mode suppression ratio of larger than 20 dB is obtained from experiments. [corrected].
In this paper, a III-V/Silicon hybrid single mode laser operating at a long wavelength for photonic integration circuit is presented. The InGaAlAs gain structure is bonded onto a patterned silicon-on insulator wafer directly. The novel mode selected mechanism based on a slotted silicon waveguide is applied, which only need standard photolithography in the whole technological process. The side mode suppression ratio of larger than 20dB is obtained from experiments.
