Severity detection of COVID-19 infection with machine learning of clinical records and CT images.

BACKGROUND: Coronavirus disease 2019 (COVID-19) is a deadly viral infection spreading rapidly around the world since its outbreak in 2019. In the worst case a patient's organ may fail leading to death. Therefore, early diagnosis is crucial to provide patients with adequate and effective treatment. OBJECTIVE: This paper aims to build machine learning prediction models to automatically diagnose COVID-19 severity with clinical and computed tomography (CT) radiomics features. METHOD: P-V-Net was used to segment the lung parenchyma and then radiomics was used to extract CT radiomics features from the segmented lung parenchyma regions. Over-sampling, under-sampling, and a combination of over- and under-sampling methods were used to solve the data imbalance problem. RandomForest was used to screen out the optimal number of features. Eight different machine learning classification algorithms were used to analyze the data. RESULTS: The experimental results showed that the COVID-19 mild-severe prediction model trained with clinical and CT radiomics features had the best prediction results. The accuracy of the GBDT classifier was 0.931, the ROUAUC 0.942, and the AUCPRC 0.694, which indicated it was better than other classifiers. CONCLUSION: This study can help clinicians identify patients at risk of severe COVID-19 deterioration early on and provide some treatment for these patients as soon as possible. It can also assist physicians in prognostic efficacy assessment and decision making.

A Deep Learning-based Method to Extract Lumen and Media-Adventitia in Intravascular Ultrasound Images.

Intravascular ultrasound (IVUS) imaging allows direct visualization of the coronary vessel wall and is suitable for assessing atherosclerosis and the degree of stenosis. Accurate segmentation and lumen and median-adventitia (MA) measurements from IVUS are essential for such a successful clinical evaluation. However, current automated segmentation by commercial software relies on manual corrections, which is time-consuming and user-dependent. We aim to develop a deep learning-based method using an encoder-decoder deep architecture to automatically and accurately extract both lumen and MA border. Inspired by the dual-path design of the state-of-the-art model IVUS-Net, our method named IVUS-U-Net++ achieved an extension of the U-Net++ model. More specifically, a feature pyramid network was added to the U-Net++ model, enabling the utilization of feature maps at different scales. Following the segmentation, the Pearson correlation and Bland-Altman analyses were performed to evaluate the correlations of 12 clinical parameters measured from our segmentation results and the ground truth. A dataset with 1746 IVUS images from 18 patients was used for training and testing. Our segmentation model at the patient level achieved a Jaccard measure (JM) of 0.9080 ± 0.0321 and a Hausdorff distance (HD) of 0.1484 ± 0.1584 mm for the lumen border; it achieved a JM of 0.9199 ± 0.0370 and an HD of 0.1781 ± 0.1906 mm for the MA border. The 12 clinical parameters measured from our segmentation results agreed well with those from the ground truth (all p-values are smaller than .01). Our proposed method shows great promise for its clinical use in IVUS segmentation.

Continuous-light versus pulsed-light accelerated corneal crosslinking with ultraviolet-A and riboflavin.

PURPOSE: To determine whether the pulsed-light ultraviolet-A (UVA) accelerated corneal crosslinking (CXL) procedure is more efficacious and selective than its continuous-light counterpart in rabbits. SETTING: School of Ophthalmology and Optometry, Wenzhou Medical University, Wenzhou, Zhejiang, China. DESIGN: Experimental study. METHODS: Fifty-four rabbits were divided into 2 groups. Group 1 had continuous-light accelerated CXL using 9 mW/cm2 UVA for 10 minutes (5.4 J/cm2). Group 2 had pulsed-light accelerated CXL by exposing them to 9 mW/cm2 UVA for 20 minutes (1 second on/1 second off). Corneal stromal demarcation line depth, in vivo confocal microscopic analysis, biomechanical stiffness, endothelial cell density, and keratocyte apoptosis were measured after performing these CXL procedures. RESULTS: The mean stromal demarcation line depth was 254.7 µm ± 47.4 (SD) in Group 1 and 341.1 ± 36.1 µm in Group 2 (P < .01). One day after CXL, confocal analysis and histological staining identified keratocyte apoptotic fragments in the anterior stroma in the Group 2 corneas whereas all cells were obliterated in Group1. Seven days after treatment, the thicknesses in Group 1 were significantly greater than those in Group 2 (P < .05). Endothelial cell losses were reversible; however, in Group 1, some losses were still evident on day 7. Increases in both the stress-strain relationship and tangent modulus in Group 2 were greater than those in Group 1. CONCLUSION: The pulsed-light accelerated CXL protocol was less injurious and more efficacious at inducing CXL than the continuous-light accelerated CXL protocol in rabbit corneas.

Hyperosmolarity-induced AQP5 upregulation promotes inflammation and cell death via JNK1/2 Activation in human corneal epithelial cells.

Tear film hyperosmolarity and anterior ocular inflammation are two clinical signs that may be indicative of dry eye disease (DED). This condition can cause pathological and functional changes to the anterior ocular surface tissues. A contributing factor may be dysfunctional aquaporin 5 (AQP5) water channels as they are the AQP subtype that expressed in the corneal epithelium and contribute to fluid efflux needed for corneal function. We determined if described hyperosmolarity-induced increases in proinflammatory cytokine expression and cell death are mediated through AQP5 upregulation and JNK1/2 MAPK signaling activation in both primary human corneal epithelial cells (HCECs), and in a HCEC line. Real time RT-PCR identified rises in IL-1ß, IL-6, IL-8, TNF-α, caspase-1, and AQP5 mRNA levels upon step increases in osmolarity up to 550 mOsm. Western blot analysis and the TUNEL assay identified corresponding rises in AQP5 and p-JNK1/2 protein expression and cell death respectively. JNK1/2 inhibition with SP600125, or siRNA AQP5 gene silencing reduced hypertonic-induced rises in proinflammatory cytokine expression and cell death. Taken together, hypertonicity-induced AQP5 upregulation leads to increases in proinflammatory cytokine expression and cell death through JNK1/2 MAPK activation. These results suggest that drug targeting AQP5 upregulation may be a therapeutic option in DED management.

High-intensity corneal collagen crosslinking with riboflavin and UVA in rat cornea.

Corneal collagen cross-linking (CXL) halts human corneal ectasias progression by increasing stromal mechanical stiffness. Although some reports describe that this procedure is effective in dealing with some infectious and immunologic corneal thinning diseases, there is a need for more animal models whose corneal thickness more closely resemble those occurring in these patients. To meet this need, we describe here high-intensity protocols that are safe and effective for obtaining CXL in rat corneas. Initially, a range of potentially effective UVA doses were evaluated based on their effectiveness in increasing tissue enzymatic resistance to dissolution. At UVA doses higher than a threshold level of 0.54 J/cm2, resistance to enzymatic digestion increased relative to that in non-irradiated corneas. Based on the theoretical threshold CXL dose, a CXL regimen was established in which the UVA tissue irradiance was 9 mW/cm2, which was delivered at doses of either 2.16, 2.7 or 3.24 J/cm2. Their dose dependent effects were evaluated on ocular surface morphological integrity, keratocyte apoptotic frequency, tissue thickness and endothelial cell layer density. Doses of 2.16 and 2.7 J/cm2 transiently decreased normal corneal transparency and increased thickness. These effects were fully reversed after 14 days. In contrast, 3.24 J/cm2 had more irreversible side effects. Three days after treatment, apoptotic frequency in the CXL-2.16 group was lower than that at higher doses. Endothelial cell losses remained evident only in the CXL-3.24 group at 42 days posttreatment. Stromal fiber thickening was evident in all the CXL-treated groups. We determined both the threshold UVA dose using the high-intensity CXL procedure and identified an effective dose range that provides optimal CXL with minimal transient side effects in the rat cornea. These results may help to provide insight into how to improve the CXL outcome in patients afflicted with a severe corneal thinning disease.