🤖 Generated with AI assistance · Does not replace individualized medical advice.
AI Transparency Notice: This content was generated by an AI assistant to summarize and structure medical literature for educational purposes. It does not constitute medical advice, diagnosis, or treatment. Always consult a qualified healthcare provider for personal medical concerns.
AI in Lacrimal Drainage Disorders: A Narrative Review of Diagnostic and Surgical Applications
Key Facts
| Domain | AI Application | Key Finding |
|---|---|---|
| Screening | AS-OCT Tear Meniscus Analysis | Deep learning models can screen for lacrimal duct obstruction using anterior segment optical coherence tomography images. PMID: 42782271 |
| Diagnosis | Lacrimal Scintigraphy Classification | ML/DL models classify anatomic versus functional epiphora, reducing reliance on subjective operator interpretation. PMID: 42782271 |
| Surgical Planning | CBCT Segmentation | nnU-Net v2 architecture enables automatic segmentation of the bony nasolacrimal canal on cone-beam CT. PMID: 42782271 |
| Decision Support | LLM Benchmarking | Large language models (e.g., ChatGPT, DeepSeek) show average but rapidly improving performance on lacrimal knowledge tasks. PMID: 42782271 |
Why It Matters / Context
Lacrimal drainage disorders are a common presentation in oculoplastic practice, yet their diagnostic evaluation remains heavily dependent on tests that are variably subjective, operator-dependent, and sometimes invasive. Traditional methods, such as dacryocystography (DCG) and manual probing, carry inherent risks and variability. The integration of AI offers a pathway toward standardized, noninvasive, and precise diagnostic workflows.
The review places these advancements within the broader context of AI in oculoplastic surgery. By leveraging conventional machine learning (ML), deep learning (DL), radiomics, and large language models (LLMs), clinicians can potentially move from reactive, invasive diagnostics to proactive, image-based screening. This shift is particularly relevant for early detection of dacryocystitis and precise localization of obstructions, which are critical for determining appropriate surgical interventions. The trajectory suggests that while AI in this specific niche is still in its infancy, the underlying technologies are maturing rapidly, with image-based models already approaching clinician-level accuracy for well-defined tasks. PMID: 42782271
Details
The review categorizes AI applications into several major clinical use cases, detailing the specific methodologies and outcomes observed in dedicated studies.
| Clinical Use Case | AI Methodology | Specific Application & Outcome |
|---|---|---|
| Noninvasive Screening | Deep Learning (DL) | Screening for lacrimal duct obstruction using anterior segment optical coherence tomography (AS-OCT) tear meniscus images. PMID: 42782271 |
| Diagnostic Classification | ML & DL | Classifying anatomic versus functional epiphora from lacrimal scintigraphy data. PMID: 42782271 |
| Lesion Localization | DL | Automatically detecting and localizing obstruction on dacryocystography (DCG) images. PMID: 42782271 |
| Predictive Modeling | Conventional ML & Explainable DL | Noninvasively predicting dacryocystitis from ocular surface indicators using explainable deep stacked networks. PMID: 42782271 |
| Surgical Planning | nnU-Net v2 | Segmenting the bony nasolacrimal canal on cone-beam computed tomography (CBCT) to support surgical planning. PMID: 42782271 |
| Knowledge Support | Large Language Models (LLMs) | Benchmarking tools like ChatGPT and DeepSeek on lacrimal knowledge, showing average but improving performance. PMID: 42782271 |
The synthesis of these studies indicates that explainable tabular models make noninvasive screening plausible, while segmentation networks are becoming precise enough to support complex surgical planning. The review emphasizes that these tools are not standalone replacements for clinical judgment but rather decision-support systems that enhance diagnostic accuracy and workflow efficiency. PMID: 42782271
Limitations
Critical Limitation: The current evidence base for AI in lacrimal drainage disorders is early-stage and largely single-center. The review explicitly notes that external validation, prospective evaluation, and standardization of lacrimal imaging datasets are missing. Without multi-center prospective trials, the generalizability of these models to diverse patient populations and clinical settings remains unproven. Furthermore, the integration of these tools into existing clinical workflows has not been rigorously tested, posing a significant barrier to routine adoption. PMID: 42782271
Clinical Takeaway
Clinicians should view current AI tools for lacrimal drainage disorders as promising proof-of-concept technologies that require rigorous external validation and prospective multi-center trials before integration into routine diagnostic or surgical workflows.
Full Reference
Yu X, Wang K, Liang X. Artificial Intelligence for Lacrimal Drainage Disorders: Advances in Diagnosis, Management, and Clinical Translation. International Ophthalmology Clinics. 2026. PMID: 42782271 [DOI: 10.1097/IIO.0000000000000641]