CauDR: A causality-inspired domain generalization framework for fundus-based diabetic retinopathy grading.

Diabetic retinopathy (DR) is the most common diabetic complication, which usually leads to retinal damage, vision loss, and even blindness. A computer-aided DR grading system has a significant impact on helping ophthalmologists with rapid screening and diagnosis. Recent advances in fundus photography have precipitated the development of novel retinal imaging cameras and their subsequent implementation in clinical practice. However, most deep learning-based algorithms for DR grading demonstrate limited generalization across domains. This inferior performance stems from variance in imaging protocols and devices inducing domain shifts. We posit that declining model performance between domains arises from learning spurious correlations in the data. Incorporating do-operations from causality analysis into model architectures may mitigate this issue and improve generalizability. Specifically, a novel universal structural causal model (SCM) was proposed to analyze spurious correlations in fundus imaging. Building on this, a causality-inspired diabetic retinopathy grading framework named CauDR was developed to eliminate spurious correlations and achieve more generalizable DR diagnostics. Furthermore, existing datasets were reorganized into 4DR benchmark for DG scenario. Results demonstrate the effectiveness and the state-of-the-art (SOTA) performance of CauDR. Diabetic retinopathy (DR) is the most common diabetic complication, which usually leads to retinal damage, vision loss, and even blindness. A computer-aided DR grading system has a significant impact on helping ophthalmologists with rapid screening and diagnosis. Recent advances in fundus photography have precipitated the development of novel retinal imaging cameras and their subsequent implementation in clinical practice. However, most deep learning-based algorithms for DR grading demonstrate limited generalization across domains. This inferior performance stems from variance in imaging protocols and devices inducing domain shifts. We posit that declining model performance between domains arises from learning spurious correlations in the data. Incorporating do-operations from causality analysis into model architectures may mitigate this issue and improve generalizability. Specifically, a novel universal structural causal model (SCM) was proposed to analyze spurious correlations in fundus imaging. Building on this, a causality-inspired diabetic retinopathy grading framework named CauDR was developed to eliminate spurious correlations and achieve more generalizable DR diagnostics. Furthermore, existing datasets were reorganized into 4DR benchmark for DG scenario. Results demonstrate the effectiveness and the state-of-the-art (SOTA) performance of CauDR.

Challenging Aromaticity: Revealing a Thioesterase Domain in a Fungal Nonreducing Polyketide Synthase Governing the Production of 3-Methylene Isochromanone.

Thioesterase (TE) domain exerts a great influence over the structure of the final product and TE-released nonreduced polyketides (nrPKs) retain aromaticity. 3-Methylene isochromanones are lactones with a unique olefin at C3 that disrupts the aromaticity, whose biosynthetic details are speculative. Our study unveils the complete biosynthesis of ascochin, in which the construction of the 3-methylene isochromanone backbone is achieved by a nonreducing polyketide synthase (nrPKS) alone and two subsequent oxidations are involved. Intriguingly, the TEAscD serves as a gatekeeper to direct the product release toward formation of nonaromatic 3-methylene isochromanone, rather than the typical aromatic product.

THE DIAGNOSTIC VALUE OF MULTISPECTRAL FUNDUS IMAGING IN HYPERTENSIVE RETINOPATHY.

PURPOSE: To observe the diagnostic value of multispectral fundus imaging (MSI) in hypertensive retinopathy (HR). METHODS: A total of 100 patients with HR were enrolled in this cross-sectional study, and all participants received fundus photography and MSI. Participants with severe HR received fundus fluorescein angiography (FFA). The diagnostic consistency between fundus photography and MSI in the diagnosis of HR was calculated. The sensitivity of MSI in the diagnosis of severe HR was calculated by comparison with FFA. Choroidal vascular index was calculated in patients with HR using MSI at 780 nm. RESULTS: MSI and fundus photography were highly concordant in the diagnosis of HR with a Kappa value = 0.883. MSI had a sensitivity of 96% in diagnosing retinal hemorrhage, a sensitivity of 89.47% in diagnosing retinal exudation, a sensitivity of 100% in diagnosing vascular compression indentation, and a sensitivity of 96.15% in diagnosing retinal arteriosclerosis. The choroidal vascular index of the patients in the HR group was significantly lower than that of the control group, whereas there was no significant difference between the affected and fellow eyes. CONCLUSION: As a noninvasive modality of observation, MSI may be a new tool for the diagnosis and assessment of HR.

Candidalysin amplifies the immune inflammatory response in Candida albicans keratitis through the TREM-1/DAP12 pathway.

Candidalysin is a fungal peptide toxin secreted by Candida albicans hyphae during invasion into epithelial cells. In Candida albicans-infected mucosa, candidalysin causes epithelial cell damage and activates downstream inflammatory responses, especially the release of inflammatory cytokines. However, the role of candidalysin in Candida albicans corneal keratitis remains unexplored. Moreover, it remains unclear whether candidalysin regulates the inflammatory response through the TREM-1/DAP12 pathway in Candida albicans corneal keratitis. In this study, we determined the expression pattern of TREM-1 in a mouse model of Candida albicans corneal keratitis and investigated the molecular mechanism underlying the inflammatory response regulation by candidalysin. The corneal keratitis model was established in C57BL/6 mice. In the GF9 group, mice were pretreated and then treated with the TREM-1 inhibitor GF9; in the candidalysin group, mice were treated with peptide candidalysin; and in the PD98059 group, mice were pretreated with the ERK inhibitor PD98059. Slit-lamp photography, clinical scoring, PCR, western blotting and immunofluorescence assay were performed to observe disease response and GF9 therapeutic efficacy. Pretreatment with candidalysin or PD98059 was performed before Candida albicans infection. GF9 treatment reduced the expression of TREM-1 and cytokines in the infected mouse cornea, whereas candidalysin treatment increased the expression of TREM-1, p-ERK, and cytokines, and this increase was inhibited by GF9. The candidalysin-induced increment of TREM-1, p-ERK, and cytokines was inhibited by PD98059 pretreatment. These data suggest that candidalysin can initiate inflammatory response in Candida albicans corneal keratitis through the TREM-1/DAP12 pathway and can regulate cytokine expression by enhancing ERK phosphorylation.