Objective: This study aims to overcome challenges in lumbar spine imaging, particularly lumbar spinal stenosis, by developing an automated segmentation model using advanced techniques. Traditional manual measurement and lesion detection methods are limited by subjectivity and inefficiency. The objective is to create an accurate and automated segmentation model that identifies anatomical structures in lumbar spine magnetic resonance imaging scans. Methods: Leveraging a dataset of 539 lumbar spinal stenosis patients, the study utilizes the residual U-Net for semantic segmentation in sagittal and axial lumbar spine magnetic resonance images. The model, trained to recognize specific tissue categories, employs a geometry algorithm for anatomical structure quantification. Validation metrics, like Intersection over Union (IOU) and Dice coefficients, validate the residual U-Net’s segmentation accuracy. A novel rotation matrix approach is introduced for detecting bulging discs, assessing dural sac compression, and measuring yellow ligament thickness. Results: The residual U-Net achieves high precision in segmenting lumbar spine structures, with mean IOU values ranging from 0.82 to 0.93 across various tissue categories and views. The automated quantification system provides measurements for intervertebral disc dimensions, dural sac diameter, yellow ligament thickness, and disc hydration. Consistency between training and testing datasets assures the robustness of automated measurements. Conclusion Automated lumbar spine segmentation with residual U-Net and deep learning exhibits high precision in identifying anatomical structures, facilitating efficient quantification in lumbar spinal stenosis cases. The introduction of a rotation matrix enhances lesion detection, promising improved diagnostic accuracy, and supporting treatment decisions for lumbar spinal stenosis patients.

Objective: This study aims to overcome challenges in lumbar spine imaging, particularly lumbar spinal stenosis, by developing an automated segmentation model using advanced techniques. Traditional manual measurement and lesion detection methods are limited by subjectivity and inefficiency. The objective is to create an accurate and automated segmentation model that identifies anatomical structures in lumbar spine magnetic resonance imaging scans. Methods: Leveraging a dataset of 539 lumbar spinal stenosis patients, the study utilizes the residual U-Net for semantic segmentation in sagittal and axial lumbar spine magnetic resonance images. The model, trained to recognize specific tissue categories, employs a geometry algorithm for anatomical structure quantification. Validation metrics, like Intersection over Union (IOU) and Dice coefficients, validate the residual U-Net’s segmentation accuracy. A novel rotation matrix approach is introduced for detecting bulging discs, assessing dural sac compression, and measuring yellow ligament thickness. Results: The residual U-Net achieves high precision in segmenting lumbar spine structures, with mean IOU values ranging from 0.82 to 0.93 across various tissue categories and views. The automated quantification system provides measurements for intervertebral disc dimensions, dural sac diameter, yellow ligament thickness, and disc hydration. Consistency between training and testing datasets assures the robustness of automated measurements. Conclusion Automated lumbar spine segmentation with residual U-Net and deep learning exhibits high precision in identifying anatomical structures, facilitating efficient quantification in lumbar spinal stenosis cases. The introduction of a rotation matrix enhances lesion detection, promising improved diagnostic accuracy, and supporting treatment decisions for lumbar spinal stenosis patients.

Objective: This study aims to overcome challenges in lumbar spine imaging, particularly lumbar spinal stenosis, by developing an automated segmentation model using advanced techniques. Traditional manual measurement and lesion detection methods are limited by subjectivity and inefficiency. The objective is to create an accurate and automated segmentation model that identifies anatomical structures in lumbar spine magnetic resonance imaging scans. Methods: Leveraging a dataset of 539 lumbar spinal stenosis patients, the study utilizes the residual U-Net for semantic segmentation in sagittal and axial lumbar spine magnetic resonance images. The model, trained to recognize specific tissue categories, employs a geometry algorithm for anatomical structure quantification. Validation metrics, like Intersection over Union (IOU) and Dice coefficients, validate the residual U-Net’s segmentation accuracy. A novel rotation matrix approach is introduced for detecting bulging discs, assessing dural sac compression, and measuring yellow ligament thickness. Results: The residual U-Net achieves high precision in segmenting lumbar spine structures, with mean IOU values ranging from 0.82 to 0.93 across various tissue categories and views. The automated quantification system provides measurements for intervertebral disc dimensions, dural sac diameter, yellow ligament thickness, and disc hydration. Consistency between training and testing datasets assures the robustness of automated measurements. Conclusion Automated lumbar spine segmentation with residual U-Net and deep learning exhibits high precision in identifying anatomical structures, facilitating efficient quantification in lumbar spinal stenosis cases. The introduction of a rotation matrix enhances lesion detection, promising improved diagnostic accuracy, and supporting treatment decisions for lumbar spinal stenosis patients.

Objective: This study aims to overcome challenges in lumbar spine imaging, particularly lumbar spinal stenosis, by developing an automated segmentation model using advanced techniques. Traditional manual measurement and lesion detection methods are limited by subjectivity and inefficiency. The objective is to create an accurate and automated segmentation model that identifies anatomical structures in lumbar spine magnetic resonance imaging scans. Methods: Leveraging a dataset of 539 lumbar spinal stenosis patients, the study utilizes the residual U-Net for semantic segmentation in sagittal and axial lumbar spine magnetic resonance images. The model, trained to recognize specific tissue categories, employs a geometry algorithm for anatomical structure quantification. Validation metrics, like Intersection over Union (IOU) and Dice coefficients, validate the residual U-Net’s segmentation accuracy. A novel rotation matrix approach is introduced for detecting bulging discs, assessing dural sac compression, and measuring yellow ligament thickness. Results: The residual U-Net achieves high precision in segmenting lumbar spine structures, with mean IOU values ranging from 0.82 to 0.93 across various tissue categories and views. The automated quantification system provides measurements for intervertebral disc dimensions, dural sac diameter, yellow ligament thickness, and disc hydration. Consistency between training and testing datasets assures the robustness of automated measurements. Conclusion Automated lumbar spine segmentation with residual U-Net and deep learning exhibits high precision in identifying anatomical structures, facilitating efficient quantification in lumbar spinal stenosis cases. The introduction of a rotation matrix enhances lesion detection, promising improved diagnostic accuracy, and supporting treatment decisions for lumbar spinal stenosis patients.

Objective: This study aims to overcome challenges in lumbar spine imaging, particularly lumbar spinal stenosis, by developing an automated segmentation model using advanced techniques. Traditional manual measurement and lesion detection methods are limited by subjectivity and inefficiency. The objective is to create an accurate and automated segmentation model that identifies anatomical structures in lumbar spine magnetic resonance imaging scans. Methods: Leveraging a dataset of 539 lumbar spinal stenosis patients, the study utilizes the residual U-Net for semantic segmentation in sagittal and axial lumbar spine magnetic resonance images. The model, trained to recognize specific tissue categories, employs a geometry algorithm for anatomical structure quantification. Validation metrics, like Intersection over Union (IOU) and Dice coefficients, validate the residual U-Net’s segmentation accuracy. A novel rotation matrix approach is introduced for detecting bulging discs, assessing dural sac compression, and measuring yellow ligament thickness. Results: The residual U-Net achieves high precision in segmenting lumbar spine structures, with mean IOU values ranging from 0.82 to 0.93 across various tissue categories and views. The automated quantification system provides measurements for intervertebral disc dimensions, dural sac diameter, yellow ligament thickness, and disc hydration. Consistency between training and testing datasets assures the robustness of automated measurements. Conclusion Automated lumbar spine segmentation with residual U-Net and deep learning exhibits high precision in identifying anatomical structures, facilitating efficient quantification in lumbar spinal stenosis cases. The introduction of a rotation matrix enhances lesion detection, promising improved diagnostic accuracy, and supporting treatment decisions for lumbar spinal stenosis patients.

Background/Aims: Ineffective esophageal motility (IEM) is common in patients with gastroesophageal reflux disease (GERD) and can be associated with poor esophageal contraction reserve on multiple rapid swallows. Alterations in the esophageal microbiome have been reported in GERD, but the relationship to presence or absence of contraction reserve in IEM patients has not been evaluated. We aim to investigate whether contraction reserve influences esophageal microbiome alterations in patients with GERD and IEM. Methods: We prospectively enrolled GERD patients with normal endoscopy and evaluated esophageal motility and contraction reserve with multiple rapid swallows during high-resolution manometry. The esophageal mucosa was biopsied for DNA extraction and 16S ribosomal RNA gene V3-V4 (Illumina)/full-length (Pacbio) amplicon sequencing analysis. Results: Among the 56 recruited patients, 20 had normal motility (NM), 19 had IEM with contraction reserve (IEM-R), and 17 had IEM without contraction reserve (IEM-NR). Esophageal microbiome analysis showed a significant decrease in microbial richness in patients with IEM-NR when compared to NM. The beta diversity revealed different microbiome profiles between patients with NM or IEM-R and IEM-NR (P = 0.037). Several esophageal bacterial taxa were characteristic in patients with IEM-NR, including reduced Prevotella spp.and Veillonella dispar, and enriched Fusobacterium nucleatum. In a microbiome-based random forest model for predicting IEM-NR, an area under the receiver operating characteristic curve of 0.81 was yielded. Conclusions In symptomatic GERD patients with normal endoscopic findings, the esophageal microbiome differs based on contraction reserve among IEM. Absent contraction reserve appears to alter the physiology and microbiota of the esophagus.

Background: Shared decision making using patient decision aids (PtDAs) was established over a decade ago, but few studies have evaluated its efficacy in Asian countries. We therefore evaluated the application of PtDAs in a decision conflict between two muscle relaxant reversal agents, neostigmine and sugammadex, and sequentially analyzed the regional differences and operating room turnover rates. Methods: This multicenter, outcome-assessor-blind, randomized controlled trial included 3,132 surgical patients from two medical centers admitted between March 2020 and August 2020. The patients were randomly divided into the classical and PtDA groups for pre-anesthesia consultations. Their clinicodemographic characteristics were analyzed to identify variables influencing the choice of reversal agent. On the day of the pre-anesthesia consultation, the patients completed the four SURE scale (sure of myself, understand information, risk-benefit ratio, encouragement) screening items. The operating turnover rates were also evaluated using anesthesia records. Results: Compared with the classical group, the PtDA group felt more confident about receiving sufficient medical information (P < 0.001), felt better informed about the advantages and disadvantages of the medications (P < 0.001), exhibited a superior understanding of the benefits and risks of their options (P < 0.001), and felt surer about their choice (P < 0.001). Moreover, the PtDA group had a significantly greater tendency to choose sugammadex over neostigmine (P < 0.001). Conclusions PtDA interventions in pre-anesthesia consultations provided surgical patients with clear knowledge and better support. PtDAs should be made available in other medical fields to enhance shared clinical decision-making.

Background/Aims: Direct‐acting antivirals (DAAs) have been approved for hepatitis C virus (HCV) treatment in patients with end-stage renal disease (ESRD) on hemodialysis. Nevertheless, the complicated comedications and their potential drug-drug interactions (DDIs) with DAAs might limit clinical practice in this special population. Methods: The number, class, and characteristics of comedications and their potential DDIs with five DAA regimens were analyzed among HCV-viremic patients from 23 hemodialysis centers in Taiwan. Results: Of 2,015 hemodialysis patients screened in 2019, 169 patients seropositive for HCV RNA were enrolled (mean age, 65.6 years; median duration of hemodialysis, 5.8 years). All patients received at least one comedication (median number, 6; mean class number, 3.4). The most common comedication classes were ESRD-associated medications (94.1%), cardiovascular drugs (69.8%) and antidiabetic drugs (43.2%). ESRD-associated medications were excluded from DDI analysis. Sofosbuvir/velpatasvir/voxilaprevir had the highest frequency of potential contraindicated DDIs (red, 5.6%), followed by glecaprevir/pibrentasvir (4.0%), sofosbuvir/ledipasvir (1.3%), sofosbuvir/velpatasvir (1.3%), and elbasvir/grazoprevir (0.3%). For potentially significant DDIs (orange, requiring close monitoring or dose adjustments), sofosbuvir/velpatasvir/voxilaprevir had the highest frequency (19.9%), followed by sofosbuvir/ledipasvir (18.2%), glecaprevir/pibrentasvir (12.6%), sofosbuvir/velpatasvir (12.6%), and elbasvir/grazoprevir (7.3%). Overall, lipid-lowering agents were the most common comedication class with red-category DDIs to all DAA regimens (n=62), followed by cardiovascular agents (n=15), and central nervous system agents (n=10). Conclusions HCV-viremic patients on hemodialysis had a very high prevalence of comedications with a broad spectrum, which had varied DDIs with currently available DAA regimens. Elbasvir/grazoprevir had the fewest potential DDIs, and sofosbuvir/velpatasvir/voxilaprevir had the most potential DDIs.

Background/Aims: Obstacles exist in facilitating hepatitis C virus (HCV) care cascade. To increase timely and accurate diagnosis, disease awareness and accessibility, in-hospital HCV reflex testing followed by automatic appointments and a late call-back strategy (R.N.A. model) was applied. We aimed to compare the HCV treatment rate of patients treated with this strategy compared to those without. Methods: One hundred and twenty-five anti-HCV seropositive patients who adopted the R.N.A. model in 2020 and another 1,396 controls treated in 2019 were enrolled to compare the gaps in accurate HCV RNA diagnosis to final treatment allocation. Results: The HCV RNA testing rate was significantly higher in patients who received reflex testing than in those without reflex testing (100% vs. 84.8%, P<0.001). When patients were stratified according to the referring outpatient department, a significant improvement in the HCV RNA testing rate was particularly noted in patients from non-hepatology departments (100% vs. 23.3%, P<0.001). The treatment rate in HCV RNA seropositive patients was 83% (83/100) after the adoption of the R.N.A. model, among whom 96.1% and 73.9% of patients were from the hepatology and non-hepatology departments, respectively. Compared to subjects without R.N.A. model application, a significant improvement in the treatment rate was observed for patients from non-hepatology departments (73.9% vs. 27.8%, P=0.001). The application of the R.N.A. model significantly increased the in-hospital HCV treatment uptake from 6.4% to 73.9% for patients from non-hepatology departments (P<0.001). Conclusions The care cascade increased the treatment uptake and set up a model for enhancing in-hospital HCV elimination.