Automated Foveal Avascular Zone Segmentation in Optical Coherence Tomography Angiography Across Multiple Eye Diseases Using Knowledge Distillation DOI Creative Commons

Peter Racioppo,

Aya Alhasany,

Nhu‐An Pham

et al.

Bioengineering, Journal Year: 2025, Volume and Issue: 12(4), P. 334 - 334

Published: March 23, 2025

Optical coherence tomography angiography (OCTA) is a noninvasive imaging technique used to visualize retinal blood flow and identify changes in vascular density enlargement or distortion of the foveal avascular zone (FAZ), which are indicators various eye diseases. Although several automated FAZ detection segmentation algorithms have been developed for use with OCTA, their performance can vary significantly due differences data accessibility OCTA different pathologies, image quality subjects and/or devices. For example, from direct macular damage, such as age-related degeneration (AMD), more readily available clinics, while on damage systemic diseases like Alzheimer’s disease often less accessible; healthy may better than ophthalmic pathologies. Typically, make convolutional neural networks and, recently, vision transformers, both long-range context fine-grained detail. However, transformers known be data-hungry, overfit small datasets, those common there limited access clinical practice. To improve model generalization low-data imbalanced settings, we propose multi-condition transformer-based architecture that uses four teacher encoders distill knowledge into shared base model, enabling transfer learned features across multiple datasets. These include intra-modality distillation using datasets ocular conditions: aging eyes, disease, AMD, diabetic retinopathy; inter-modality incorporating color fundus photographs undergoing laser photocoagulation therapy. Our achieved mean Dice Index 83.8% pretraining, outperforming single-condition models (mean 83.1%) all conditions. Pretraining images improved average by margin conditions except AMD (1.1% models, 0.1% models). demonstrates potential broader applications detecting analyzing diverse settings.

Language: Английский

A blood supply pathophysiological microcirculatory mechanism for the Long COVID DOI Open Access
Aristotle G. Koutsiaris

Published: July 30, 2024

Background: The term “Long COVID” is commonly used to describe persisting symptoms after acute COVID‑19. Until now, proposed mechanisms for the explanation of Long COVID have not related quantitative measurements basic laws. In this work, a common framework pathophysiological mechanism presented, based on blood supply deprivation and flow diffusion equation. Methods: Case-control studies with statistically significant differences between cases (post-COVID patients) controls, from multiple tissues geographical areas, were gathered tabulated. Microvascular loss (ML) was quantified by vessel density reduction (VDR), foveal avascular zone enlargement (FAZE), capillary (CDR), percentage perfused vessels (PPVR). Both ML hemodynamic decrease (HD), incorporated in tissue (SR) estimation. Results: data found 763 post-COVID patients an average VDR, FAZE, CDR, PPVR 16%, 31%, 14%, 21%, respectively. HD 72 37%. estimated SR 634 reached sizeable 47%. This large creates conditions lower mass rates, hypoxia, undernutrition, which at multi-tissue level, long time, can explain wide variety symptoms. Conclusions: Disruption peripheral contribution both here be principal cause leading

Language: Английский

Citations

4
Overview of Inflammatory and Coagulation Markers in Elderly Patients with COVID-19: Retrospective Analysis of Laboratory Results DOI Creative Commons
Corina Popazu,

Aurelia Romila,

M. Petrea

et al.

Life, Journal Year: 2025, Volume and Issue: 15(3), P. 370 - 370

Published: Feb. 26, 2025

Background: Elderly patients with COVID-19 often exhibit a complex interplay between hypercoagulability and coagulopathy, key factors in determining the risk of severe complications mortality. This study aimed to analyze coagulation inflammatory markers identify critical predictors adverse outcomes this vulnerable population. Material Methods: The retrospective was conducted on sample 1429 elderly (≥60 years) diagnosed COVID-19, hospitalized “Sf. Ap. Andrei” St. Apostle Andrew’s County Emergency Hospital various wards March 2020 August 2022. Data were collected from medical records included (C-reactive protein, procalcitonin, ESR) (prothrombin time, INR, fibrinogen, D-dimer). SPSS 2.0 statistical software used conduct study. Results:Coagulation markers: Prothrombin activity averaged 74.22%, below normal levels, indicating heightened bleeding risk, while fibrinogen levels significantly elevated (mean: 531.69 mg/dL), reflecting hypercoagulability. Prolonged prothrombin time 17.28 s) INR (International normalized ratio) 1.51) associated increased mortality, emphasizing their role stratification. Elevated D-dimer 2.75 mg/L) further highlighted thromboembolic risks. Inflammatory C-reactive protein (CRP) erythrocyte sedimentation rate (ESR) showed marked elevations (mean CRP: 92.09 mg/L, mean ESR: 58.47 mm/h), correlating systemic inflammation poor outcomes. Bacterial infections: procalcitonin 1.98 ng/mL) suggested secondary bacterial infections, particularly mechanically ventilated patients, worsening prognosis. Conclusions: duality coagulopathy underscores importance consistently monitoring such as D-dimer, fibrinogen. Simultaneously, infections require prompt therapeutic interventions. highlights need for personalized management strategies mitigate reduce mortality high-risk

Language: Английский

Citations

0
A Coupled Model of the Cardiovascular and Immune Systems to Analyze the Effects of COVID-19 Infection DOI Creative Commons
Camila Greff Passos,

Alexandre Altamir Moreira,

Ruy Freitas Reis

et al.

BioTech, Journal Year: 2025, Volume and Issue: 14(1), P. 19 - 19

Published: March 12, 2025

The COVID-19 pandemic has underscored the importance of understanding interplay between cardiovascular and immune systems during viral infections. SARS-CoV-2 enters human cells via ACE-2 enzyme, initiating a cascade responses. This study presents coupled mathematical model that integrates system (CVS) (IS), capturing their complex interactions infection. CVS model, based on ordinary differential equations, describes heart dynamics pulmonary systemic circulation, while IS simulates responses to SARS-CoV-2, including cell cytokine production. A coupling strategy transfers information from at specific intervals, enabling exploration immune-driven effects. Numerical simulations examined how these influence infection severity recovery. accurately replicated evolution cardiac function in survivors non-survivors COVID-19. Survivors exhibited left ventricular ejection fraction (LVEF) reduction up 25% remaining within normal limits, whereas showed severe 4-fold decline, indicative myocardial dysfunction. Similarly, right (RV EF) decreased by approximately 50% but underwent drastic 5-fold non-survivors. These findings highlight model's capacity distinguish dysfunction across clinical outcomes its potential enhance our pathophysiology.

Language: Английский

Citations

0
Automated Foveal Avascular Zone Segmentation in Optical Coherence Tomography Angiography Across Multiple Eye Diseases Using Knowledge Distillation DOI Creative Commons

Peter Racioppo,

Aya Alhasany,

Nhu‐An Pham

et al.

Bioengineering, Journal Year: 2025, Volume and Issue: 12(4), P. 334 - 334

Published: March 23, 2025

Optical coherence tomography angiography (OCTA) is a noninvasive imaging technique used to visualize retinal blood flow and identify changes in vascular density enlargement or distortion of the foveal avascular zone (FAZ), which are indicators various eye diseases. Although several automated FAZ detection segmentation algorithms have been developed for use with OCTA, their performance can vary significantly due differences data accessibility OCTA different pathologies, image quality subjects and/or devices. For example, from direct macular damage, such as age-related degeneration (AMD), more readily available clinics, while on damage systemic diseases like Alzheimer’s disease often less accessible; healthy may better than ophthalmic pathologies. Typically, make convolutional neural networks and, recently, vision transformers, both long-range context fine-grained detail. However, transformers known be data-hungry, overfit small datasets, those common there limited access clinical practice. To improve model generalization low-data imbalanced settings, we propose multi-condition transformer-based architecture that uses four teacher encoders distill knowledge into shared base model, enabling transfer learned features across multiple datasets. These include intra-modality distillation using datasets ocular conditions: aging eyes, disease, AMD, diabetic retinopathy; inter-modality incorporating color fundus photographs undergoing laser photocoagulation therapy. Our achieved mean Dice Index 83.8% pretraining, outperforming single-condition models (mean 83.1%) all conditions. Pretraining images improved average by margin conditions except AMD (1.1% models, 0.1% models). demonstrates potential broader applications detecting analyzing diverse settings.

Language: Английский

Citations

0