Ophthalmology
VisionFM: A Generalist AI Surpasses Single-Modality Models in Ophthalmic Diagnostics
25 Dec, 2024 | 13:41h | UTCBackground: Ophthalmic AI models typically address single diseases or modalities. Their limited generalizability restricts broad clinical application. This study introduces VisionFM, a novel foundation model trained on 3.4 million images from over 500,000 individuals. It covers eight distinct ophthalmic imaging modalities (e.g., fundus photography, OCT, slit-lamp, ultrasound, MRI) and encompasses multiple diseases. Compared with prior single-task or single-modality approaches, VisionFM’s architecture and large-scale pretraining enable diverse tasks such as disease screening, lesion segmentation, prognosis, and prediction of systemic markers.
Objective: To develop and validate a generalist ophthalmic AI framework that can handle multiple imaging modalities, recognize multiple diseases, and adapt to new clinical tasks through efficient fine-tuning, potentially easing the global burden of vision impairment.
Methods: VisionFM employs individual Vision Transformer–based encoders for each of the eight imaging modalities, pretrained with self-supervised learning (iBOT) focused on masked image modeling. After pretraining, various task-specific decoders were fine-tuned for classification, segmentation, and prediction tasks. The model was evaluated on 53 public and 12 private datasets, covering eight disease categories (e.g., diabetic retinopathy, glaucoma, cataract), five imaging modalities (fundus photographs, OCT, etc.), plus additional tasks (e.g., MRI-based orbital tumor segmentation). Performance metrics included AUROCs, Dice similarity coefficients, F1 scores, and comparisons with ophthalmologists of varying clinical experience.
Results: VisionFM achieved an average AUROC of 0.950 (95% CI, 0.941–0.959) across eight disease categories in internal validation. External validation showed AUROCs of 0.945 (95% CI, 0.934–0.956) for diabetic retinopathy and 0.974 (95% CI, 0.966–0.983) for AMD, surpassing baseline deep learning approaches. In a 12-disease classification test involving 38 ophthalmologists, VisionFM’s accuracy matched intermediate-level specialists. It successfully handled modality shifts (e.g., grading diabetic retinopathy on previously unseen OCTA), with an AUROC of 0.935 (95% CI, 0.902–0.964). VisionFM also predicted glaucoma progression (F1, 72.3%; 95% CI, 55.0–86.3) and flagged possible intracranial tumors (AUROC, 0.986; 95% CI, 0.960–1.00) from fundus images.
Conclusions: VisionFM offers a versatile, scalable platform for comprehensive ophthalmic tasks. Through self-supervised learning and efficient fine-tuning, it extends specialist-level performance to multiple clinical scenarios and imaging modalities. The study demonstrates that large-scale, multimodal pretraining can enable robust generalization to unseen data, potentially reducing data annotation burdens and accelerating AI adoption worldwide.
Implications for Practice: VisionFM may help address global shortages of qualified ophthalmologists and expand care in low-resource settings, though clinical decision-making still requires appropriate human oversight. Further multicenter studies are needed before widespread implementation, especially for higher-risk use cases such as tumor detection.
Study Strengths and Limitations: Strengths include its unique multimodal design, large-scale pretraining, and extensive external validation. Limitations involve demographic bias toward Chinese datasets, the need for larger cohorts in certain applications (e.g., intracranial tumor detection), and the challenges of matching real-world clinical complexity when only image-based data are used.
Future Research: Further validation in diverse populations, integration of new imaging modalities (e.g., widefield imaging, ultrasound variants), and expansion to additional diseases are planned. Hybridization with large language models could facilitate automatic generation of clinical reports.
Reference: Qiu J, Wu J, Wei H, et al. Development and Validation of a Multimodal Multitask Vision Foundation Model for Generalist Ophthalmic Artificial Intelligence. NEJM AI 2024;1(12). DOI: http://doi.org/10.1056/AIoa2300221
Summary: Perioperative Management of Patients Taking Direct Oral Anticoagulants
19 Sep, 2024 | 21:12h | UTCDirect oral anticoagulants (DOACs)—including apixaban, rivaroxaban, edoxaban, and dabigatran—are increasingly used for stroke prevention in atrial fibrillation and for treating venous thromboembolism. Effective perioperative management of DOACs is essential to minimize bleeding and thromboembolic risks during surgical and nonsurgical procedures. Below are practical recommendations focused on the perioperative management of patients taking DOACs, based on a recent JAMA review article.
Elective Surgical or Nonsurgical Procedures
Classify Bleeding Risk of Procedures:
- Minimal Risk:
- Minor dental procedures (e.g., cleaning, extractions)
- Minor dermatologic procedures (e.g., skin lesion removal)
- Cataract surgery
- Low to Moderate Risk:
- Endoscopic procedures without high-risk interventions
- Cholecystectomy
- Inguinal hernia repair
- High Risk:
- Major surgery (e.g., cancer surgery, joint replacement)
- Procedures involving neuraxial anesthesia
- Endoscopic procedures with high-risk interventions (e.g., large polyp removal)
DOAC Management Strategies:
- Minimal Bleeding Risk Procedures:
- Option 1: Continue DOACs without interruption.
- Option 2: For added safety, withhold the morning dose on the day of the procedure (especially for twice-daily DOACs like apixaban and dabigatran).
- Low to Moderate Bleeding Risk Procedures:
- Preoperative:
- Discontinue DOACs 1 day before the procedure.
- This allows approximately 2 half-lives for drug clearance.
- Postoperative:
- Resume DOACs 1 day after the procedure, ensuring adequate hemostasis.
- Preoperative:
- High Bleeding Risk Procedures:
- Preoperative:
- Discontinue DOACs 2 days before the procedure.
- This allows approximately 4-5 half-lives for drug clearance.
- Postoperative:
- Resume DOACs 2-3 days after the procedure, based on bleeding risk and hemostasis.
- Preoperative:
Evidence Supporting These Strategies:
- The PAUSE study demonstrated that standardized interruption protocols without heparin bridging result in low rates of:
- Thromboembolism: 0.2%–0.4%
- Major Bleeding: 1%–2%
Postoperative DOAC Resumption:
- Assess surgical-site hemostasis before resuming DOACs.
- Delay resumption if there is ongoing bleeding or concerns about hemostasis.
- For high bleeding risk procedures, consider a longer delay (2–3 days).
Perioperative Heparin Bridging:
- Not recommended for patients on DOACs.
- Bridging increases bleeding risk without reducing thromboembolism.
- DOACs have rapid offset and onset, making bridging unnecessary.
Special Considerations
Patients with Impaired Renal Function:
- For CrCl 30–50 mL/min:
- Dabigatran: Extend preoperative discontinuation by an additional day.
- For CrCl <30 mL/min:
- Dabigatran is contraindicated.
- For other DOACs, consider extending discontinuation to 3–4 days before surgery.
Patients Undergoing Neuraxial Anesthesia:
- Discontinue DOACs for 3 days (apixaban, edoxaban, rivaroxaban) or 4 days (dabigatran) before the procedure.
- Minimizes risk of spinal or epidural hematoma.
Dental Procedures:
- Generally safe to continue DOACs.
- For added safety:
- Omit or delay the dose on the day of the procedure.
- Employ local hemostatic measures (e.g., tranexamic acid mouthwash).
Endoscopic Procedures:
- Low-risk procedures (e.g., diagnostic endoscopy without biopsy):
- Follow standard DOAC interruption for low to moderate bleeding risk.
- High-risk procedures (e.g., polypectomy of large polyps):
- Extend DOAC discontinuation by an additional day pre- and post-procedure.
Patients Unable to Resume Oral Medications Postoperatively:
- Use prophylactic low-molecular-weight heparin (LMWH) until oral intake is possible.
- Avoid therapeutic-dose LMWH due to bleeding risk.
Emergent, Urgent, or Semiurgent Procedures
Risks:
- Higher bleeding risk: Up to 23%
- Thromboembolism risk: Up to 11%
Management Strategies:
- Assess Time Since Last DOAC Dose:
- If within 48 hours, consider that significant anticoagulant effect may persist.
- Laboratory Testing (if available):
- DOAC Level Testing:
- ≥50 ng/mL: Consider using reversal agents.
- <50 ng/mL: May proceed without reversal agents.
- DOAC Level Testing:
- Use of Reversal Agents:
- For Dabigatran:
- Idarucizumab (5 g IV)
- For Factor Xa Inhibitors (apixaban, rivaroxaban, edoxaban):
- Andexanet alfa (dosing based on last dose timing and amount)
- Prothrombin Complex Concentrates (PCCs): If andexanet alfa is unavailable or contraindicated.
- For Dabigatran:
- Proceeding Without Testing:
- If testing is unavailable and last DOAC dose was within 48 hours, consider reversal agents.
- If >48 hours since last dose, may proceed without reversal.
Considerations:
- Reversal agents are expensive and may carry thrombotic risks.
- Use should be judicious, weighing risks and benefits.
- Consult hematology or thrombosis experts when possible.
Key Takeaways
- Elective Procedures:
- Utilize standardized protocols based on procedural bleeding risk.
- Routine preoperative DOAC level testing is unnecessary.
- Avoid heparin bridging.
- Emergent/Urgent Procedures:
- Reversal agents may be appropriate when significant DOAC levels are present.
- Decision to use reversal agents should consider bleeding risk, time since last dose, and availability of DOAC level testing.
- Patient Communication:
- Ensure patients understand the plan for DOAC interruption and resumption.
- Provide clear instructions regarding timing and dosing.
- Interdisciplinary Coordination:
- Collaborate with surgical teams, anesthesiologists, and pharmacists.
- Use electronic medical records and clinical decision support tools to enhance communication.
Conclusion
By applying standardized perioperative management protocols, clinicians can effectively balance the risks of bleeding and thromboembolism in patients taking DOACs who require surgical or nonsurgical procedures. These strategies simplify decision-making, avoid unnecessary interventions like heparin bridging, and promote patient safety.
Cross-Sectional Study: AI Model Accurately Detects Myopia, Strabismus, and Ptosis in Children Using Smartphone Photos – JAMA Netw Open
10 Aug, 2024 | 21:21h | UTCStudy Design and Population: This cross-sectional study was conducted at Shanghai Ninth People’s Hospital from October 2022 to September 2023, including 476 children diagnosed with myopia, strabismus, or ptosis. A total of 1,419 images were used to develop an AI model to detect these conditions based on mobile phone photographs.
Main Findings: The AI model demonstrated strong performance with a sensitivity of 0.84 for myopia, 0.73 for strabismus, and 0.85 for ptosis. The model achieved overall accuracies exceeding 0.80 for myopia and strabismus and 0.92 for ptosis, indicating its effectiveness in early detection of these pediatric eye conditions using only smartphone images.
Implications for Practice: The findings suggest that AI-based screening tools could enable early detection of common pediatric eye diseases at home, reducing the reliance on hospital-based screenings. This approach could facilitate timely intervention, improving visual outcomes and reducing the burden on healthcare systems.
Phase 1-2 Study: Safety and efficacy of EDIT-101 CRISPR-Cas9 treatment for CEP290-associated retinal degeneration – N Engl J Med
25 May, 2024 | 19:42h | UTCThis study evaluates the safety and effectiveness of EDIT-101, a CRISPR-Cas9 gene-editing therapy, in treating inherited retinal degeneration caused by CEP290 IVS26 variants. Conducted as a phase 1-2, open-label, single-ascending-dose trial, it involved 14 participants (12 adults aged 17-63 and 2 children aged 9 and 14) who received subretinal injections of EDIT-101. Treatment doses varied, with two participants at a low dose, seven at an intermediate dose, and five at a high dose. The primary safety assessment showed no serious adverse events or dose-limiting toxic effects. Notably, 64% of participants exhibited significant improvements in visual acuity, retinal sensitivity to red light, or mobility. Additionally, improvements in vision-related quality of life were documented in six participants. These promising results suggest that EDIT-101 is safe and potentially effective, warranting further investigation into CRISPR-Cas9 gene therapy for similar genetic retinal conditions.
Commentary on X:
Original Article: Gene Editing for CEP290-Associated Retinal Degeneration (BRILLIANCE) https://t.co/LgSs2RGBZv #ARVO2024 @ARVOinfo pic.twitter.com/WztX2OvzXM
— NEJM (@NEJM) May 7, 2024
Reference (link to abstract – $ for full-text):
RCT: Early patching proves more effective than extended optical treatment in pediatric amblyopia management – The Lancet
5 May, 2024 | 15:00h | UTCStudy Design and Population:
The EuPatch study was a multicenter, randomized controlled trial conducted across 30 hospitals in multiple European countries, including the UK, Greece, Austria, Germany, and Switzerland. It targeted children aged 3–8 years diagnosed with amblyopia due to anisometropia, strabismus, or both, with an interocular difference ≥0.30 logMAR in best corrected visual acuity (BCVA). Participants were divided into two groups: one underwent 18 weeks of glasses use before patching (Extended Optical Treatment, EOT), and the other just 3 weeks (early patching), each supplemented with an intensive patching regimen.
Main Findings:
Out of the 334 initially randomized participants, 317 were analyzed for the primary outcome. The early patching group demonstrated a significantly higher success rate, achieving ≤0.20 logMAR interocular difference in BCVA in 67% of cases compared to 54% in the EOT group after 12 weeks of patching (p=0.019). The median follow-up times were 42 weeks for the EOT group and 27 weeks for the early patching group, with dropout rates of 14% and 6%, respectively.
Implications for Practice:
The findings suggest that early patching is superior to EOT in the management of amblyopia in children, presenting a viable option for enhancing treatment effectiveness. This study supports the potential personalization of amblyopia treatments based on the quicker onset of patching. These results could influence future guidelines and clinical practices in pediatric ophthalmology.
Reference (link to free full-text):
Presented at ASRS Meeting | Studies link GLP-1 agonists to progression of diabetic retinopathy
11 Aug, 2023 | 15:38h | UTCStudies link GLP-1 agonists to progression of diabetic retinopathy – MDedge
RCT | No significant impact of low-dose 0.01% atropine on myopia progression in children
25 Jul, 2023 | 13:40h | UTCLow-Dose 0.01% Atropine Eye Drops vs Placebo for Myopia Control: A Randomized Clinical Trial – JAMA Ophthalmology (link to abstract – $ for full-text)
See also: Visual Abstract
Author Interview: Low-Dose 0.01% Atropine Eye Drops vs Placebo for Myopia Control – JAMA
Commentary: Atropine 0.01 Percent Eye Drops Do Not Slow Myopia Progression – HealthDay
Related:
RCT | 0.01% atropine effective in reducing progression of myopia in children over 3 years
RCT | Effect of low-concentration atropine eyedrops vs. placebo on myopia incidence in children
Crossover RCT | Spectacle lenses with highly aspherical lenslets for slowing myopia.
RCT | Effect of text messaging parents of school-aged children on outdoor time to control myopia.
RCT | Myopia control effect of repeated low-level red-light therapy in children.
Guideline | Prevention and management of childhood progressive myopia
18 Jul, 2023 | 13:35h | UTC
RCT | Intravitreal therapy for uveitic macular edema—ranibizumab versus methotrexate versus the dexamethasone implant
29 Jun, 2023 | 13:57h | UTCIntravitreal Therapy for Uveitic Macular Edema—Ranibizumab versus Methotrexate versus the Dexamethasone Implant – Ophthalmology (link to abstract – $ for full-text)
SR | Anti‐vascular endothelial growth factors in combination with vitrectomy for complications of proliferative diabetic retinopathy
29 Jun, 2023 | 13:49h | UTC
Editorial | How to become a good surgeon
22 Jun, 2023 | 15:12h | UTCHow to become a good surgeon – Advances in Ophthalmology Practice and Research
RCT | 0.01% atropine effective in reducing progression of myopia in children over 3 years
14 Jun, 2023 | 14:41h | UTCNews Release: Eye drops slow nearsightedness progression in kids – Ohio State University
See also: Visual Abstract
Related:
RCT | Effect of low-concentration atropine eyedrops vs. placebo on myopia incidence in children
Crossover RCT | Spectacle lenses with highly aspherical lenslets for slowing myopia.
RCT | Effect of text messaging parents of school-aged children on outdoor time to control myopia.
RCT | Myopia control effect of repeated low-level red-light therapy in children.
RCT | Immediate vs. delayed sequential bilateral cataract surgery: non-inferior in safety, higher cost-effectiveness
6 Jun, 2023 | 14:41h | UTCSafety, effectiveness, and cost-effectiveness of immediate versus delayed sequential bilateral cataract surgery in the Netherlands (BICAT-NL study): a multicentre, non-inferiority, randomised controlled trial – The Lancet (link to abstract – $ for full-text)
News Release: Cataract surgery: two-sided treatment better than one-by-one approach – Maastricht University
Review | Integrating eye care in low-income and middle-income settings
6 Jun, 2023 | 14:21h | UTCIntegrating eye care in low-income and middle-income settings: a scoping review – BMJ Open
Eye disease and international travel: a critical literature review and practical recommendations
2 Jun, 2023 | 12:21h | UTC
Study finds potential glaucoma risk in patients using calcium channel blockers
1 Jun, 2023 | 11:51h | UTCCommentary: Calcium Channel Blocker Use Associated with Higher Prevalence of Glaucoma – HCP Live
RCT | Effect of repeated low-level red light on myopia prevention among children in China with premyopia
10 May, 2023 | 15:52h | UTCCommentary: Repeated Low-Level Red-Light Intervention Prevents Myopia in Children – HealthDay
Related:
RCT | Effect of low-concentration atropine eyedrops vs. placebo on myopia incidence in children
Crossover RCT | Spectacle lenses with highly aspherical lenslets for slowing myopia.
RCT | Effect of text messaging parents of school-aged children on outdoor time to control myopia.
RCT | Myopia control effect of repeated low-level red-light therapy in children.
Commentary on Twitter
RCT: Repeated low-level red-light (RLRL) is a novel and effective intervention for myopia prevention in children with premyopia, up to 54% reduction in incident myopia within 12 months, with good user acceptability and safety. https://t.co/geMajH5ms8
— JAMA Network Open (@JAMANetworkOpen) April 26, 2023
SR | Blood pressure control for diabetic retinopathy
17 Apr, 2023 | 12:45h | UTCBlood pressure control for diabetic retinopathy – Cochrane Library
Summary: Blood pressure control for diabetic retinopathy – Cochrane Library
SR | Device‐modified trabeculectomy for glaucoma
17 Apr, 2023 | 12:46h | UTCDevice‐modified trabeculectomy for glaucoma – Cochrane Library
Summary: Device-modified trabeculectomy for glaucoma – Cochrane Library
RCT | Efficacy and safety of a water-free topical cyclosporine, 0.1%, solution for the treatment of dry eye disease
11 Apr, 2023 | 14:10h | UTCCommentary: ESSENCE-2 Trial Confirms Efficacy of Cyclosporine Solution 0.1% for DED – HCP Live
Commentary on Twitter
ESSENCE-2 trial demonstrated treatment with water-free cyclosporine 0.1% (CyclASol) drops effective and tolerable through Day 29 for treating dry eye related keratitis with rapid onset of action. https://t.co/qJkef2fPTg
— JAMA Ophthalmology (@JAMAOphth) April 6, 2023
M-A | Retinopathy of prematurity and neurodevelopmental outcomes in preterm infants
10 Apr, 2023 | 13:36h | UTC
SR | Anti‐vascular endothelial growth factor for neovascular glaucoma
10 Apr, 2023 | 13:33h | UTC
M-A | Efficacy of atropine for myopia control in children
5 Apr, 2023 | 12:45h | UTCRelated:
RCT | Effect of low-concentration atropine eyedrops vs. placebo on myopia incidence in children
Crossover RCT | Spectacle lenses with highly aspherical lenslets for slowing myopia.
RCT | Effect of text messaging parents of school-aged children on outdoor time to control myopia.
RCT | Myopia control effect of repeated low-level red-light therapy in children.
RCT | Pulsed oral azithromycin vs. 6-week oral doxycycline for moderate to severe meibomian gland dysfunction
31 Mar, 2023 | 13:33h | UTCPulsed Oral Azithromycin vs 6-Week Oral Doxycycline for Moderate to Severe Meibomian Gland Dysfunction: A Randomized Clinical Trial – JAMA Ophthalmology (link to abstract – $ for full-text)
Commentary: 3-Week MGD treatment equivalent to 6-week course in efficacy and adverse events – Ophthalmology Times
Commentary on Twitter
A 3-week course of weekly oral azithromycin was equivalent to a 6-week course of oral doxycycline in treating moderate to severe meibomian gland dysfunction. https://t.co/nE312KxGki pic.twitter.com/xWmqYWby7j
— JAMA Ophthalmology (@JAMAOphth) March 23, 2023
SR | Anti‐vascular endothelial growth factor for proliferative diabetic retinopathy
29 Mar, 2023 | 13:15h | UTCAnti‐vascular endothelial growth factor for proliferative diabetic retinopathy – Cochrane Library