This paper presents a new deformable model using both population-based and patient-specific shape statistics to segment lung fields from serial chest radiographs. There are two novelties in the proposed model. First, modified scale invariant feature transform (SIFT) local descriptor, which is more distinctive than general intensity gradient features, used characterize image features vicinity of each pixel. Second, contour constrained by statistics, it yields robust accurate segmentation for...

Lung field segmentation in the posterior-anterior (PA) chest radiograph is important for pulmonary disease diagnosis and hemodialysis treatment. Due to high shape variation boundary ambiguity, accurate lung from still a challenging task. To tackle these challenges, we propose joint appearance sparse learning method robust segmentation. The main contributions of this paper are: 1) initialization designed achieve an initial that close under segmentation; 2) set local composition models are...

Intravenous thrombolysis is the most commonly used drug therapy for patients with acute ischemic stroke, which often accompanied by complications of intracerebral hemorrhage transformation (HT). This study proposed to build a reliable model pretreatment prediction HT. Specifically, 5400 radiomics features were extracted from 20 regions interest (ROIs) multiparametric MRI images 71 patients. Furthermore, minimal set all-relevant selected LASSO all ROIs and through random forest (RF). To...

The accelerated MRI reconstruction poses a challenging ill-posed inverse problem due to the significant undersampling in k-space. Deep neural networks, such as CNNs and ViT, have shown substantial performance improvements for this task while encountering dilemma between global receptive fields efficient computation. To end, paper pioneers exploring Mamba, new paradigm long-range dependency modeling with linear complexity, effective reconstruction. However, directly applying Mamba faces three...

The accelerated MRI reconstruction process presents a challenging ill-posed inverse problem due to the extensive under-sampling in k-space. Recently, Vision Transformers (ViTs) have become mainstream for this task, demonstrating substantial performance improvements. However, there are still three significant issues remain unaddressed: (1) ViTs struggle capture high-frequency components of images, limiting their ability detect local textures and edge information, thereby impeding restoration;...

Abstract Patients with frontal lobe gliomas often experience neurocognitive dysfunctions before surgery, which affects the default mode network (DMN) to different degrees. This study quantitatively analyzed this effect from perspective of cerebral hemispheric functional connectivity (FC). We collected resting-state fMRI data 20 glioma patients treatment and healthy controls. All controls were right-handed. After pre-processing images, FC maps built seed defined in left or right posterior...

Purpose Deformable multimodal image registration, which can benefit radiotherapy and guided surgery by providing complementary information, remains a challenging task in the medical analysis field due to difficulty of defining proper similarity measure. This article presents novel, robust fast binary descriptor, discriminative local derivative pattern ( dLDP ), is able encode images different modalities into similar representations. Methods calculates string for each voxel according...

Computer-assisted intervention often depends on multimodal deformable registration to provide complementary information. However, remains a challenging task. This paper introduces novel robust and fast modality-independent 3D binary descriptor, called miLBP, which integrates the principle of local self-similarity with form pattern can robustly extract similar geometry features from volumes across different modalities. miLBP is bit string that be computed by simply thresholding voxel...

Segmentation of the magnetic resonance (MR) images is fundamentally important in medical image analysis. Intensity inhomogeneity due to unknown noise and weak boundary makes it a difficult problem. The paper presents novel level set geodesic model which integrates local global intensity information signed pressure force (SPF) function suppress implement segmentation. First, new region based SPF proposed extract order ensure flexible initialization object contours. Second, adaptively balanced...

Accurate identification and localization of the vertebrae in CT scans is a critical standard pre-processing step for clinical spinal diagnosis treatment. Existing methods are mainly based on integration multiple neural networks, most them use heatmaps to locate vertebrae's centroid. However, process obtaining centroid coordinates using non-differentiable, so it impossible train network label directly. Therefore, end-to-end differential training scans, robust accurate automatic vertebral...

Accurate segmentation of anatomical brain structures is crucial for many neuroimaging applications, e.g., early development studies and the study imaging biomarkers neurodegenerative diseases. Although multi-atlas (MAS) has achieved successes in medical area, this approach encounters limitations segmenting associated with poor image contrast. To address issue, we propose a new MAS method that uses hypergraph learning framework to model complex subject-within subject-to-atlas voxel...

This paper presented a novel complex network with one-way ANOVA F-test feature selection to diagnose early-stage Parkinson’s disease (PD) on quantitative susceptibility mapping (QSM). Experimental results QSM images of 30 PD patients and 27 healthy controls (HC) proved that the scheme was effective achieved good classification results. The accuracy, AUC, sensitivity specificity our method were 0.96, 0.97, 0.99 0.95, respectively, which improved by 15%, 4%, 29% 2%, respectively comparison...

Purpose: In adaptive radiation therapy of prostate cancer, fast and accurate registration between the planning image treatment images patient is essential importance. With authors' recently developed deformable surface model, boundaries in each can be rapidly segmented their correspondences (or relative deformations) to are also established automatically. However, dense on nonboundary regions, which important especially for transforming plan designed space space, remained unresolved. This...

Plaque assaying, measurement of the number, diameter, and area plaques in a Petri dish image, is standard procedure gauging concentration phage biology. This paper presented novel effective method for implementing automatic plaqu

Accurate identification and localization of the vertebrae in CT scans is a critical standard preprocessing step for clinical spinal diagnosis treatment. Existing methods are mainly based on integration multiple neural networks, most them use Gaussian heat map to locate vertebrae's centroid. However, process obtaining centroid coordinates using maps non-differentiable, so it impossible train network label directly. Therefore, end-to-end differential training vertebra scans, robust accurate...

Sarcopenia is generally diagnosed by the total area of skeletal muscle in CT axial slice located third lumbar (L3) vertebra. However, patients with severe liver cirrhosis cannot accurately obtain corresponding because their abdominal muscles are squeezed, which affects diagnosis sarcopenia. This study proposes a novel network to automatically segment multi-regional from images, and explores relationship between cirrhotic sarcopenia each region. utilizes characteristics different spatial...

Image-guided computer-aided navigation system is an indispensable part of computer assisted orthopaedic surgery. However, the location and number fiducial markers, time required to localise markers in existing systems affect their effectiveness.The study proposed that spatial surface registration between point cloud on fusion model based preoperative knee MRI CT images cartilage captured by intraoperative laser scanner could solve above limitations.The experimental results show error method...