[Paper Review] Automated Process Incorporating Machine Learning Segmentation and Correlation of Oral Diseases with Systemic Health

Gregory Yauney, Aman Rana+4 more|arXiv (Cornell University)|Oct 24, 2018

Oral microbiology and periodontitis research4 citations

TL;DR

This study presents an automated machine learning pipeline that uses intraoral fluorescent imaging and clinical data to detect periodontal disease and correlate it with systemic health conditions. The system achieved an AUC of 0.677 in classifying gingival inflammation and revealed significant associations between periodontal disease and systemic markers such as optic nerve abnormalities (p < 0.0001), swollen joints (p = 0.0422), and family history of eye disease (p = 0.0337).

ABSTRACT

Imaging fluorescent disease biomarkers in tissues and skin is a non-invasive method to screen for health conditions. We report an automated process that combines intraoral fluorescent porphyrin biomarker imaging, clinical examinations and machine learning for correlation of systemic health conditions with periodontal disease. 1215 intraoral fluorescent images, from 284 consenting adults aged 18-90, were analyzed using a machine learning classifier that can segment periodontal inflammation. The classifier achieved an AUC of 0.677 with precision and recall of 0.271 and 0.429, respectively, indicating a learned association between disease signatures in collected images. Periodontal diseases were more prevalent among males (p=0.0012) and older subjects (p=0.0224) in the screened population. Physicians independently examined the collected images, assigning localized modified gingival indices (MGIs). MGIs and periodontal disease were then cross-correlated with responses to a medical history questionnaire, blood pressure and body mass index measurements, and optic nerve, tympanic membrane, neurological, and cardiac rhythm imaging examinations. Gingivitis and early periodontal disease were associated with subjects diagnosed with optic nerve abnormalities (p &lt;0.0001) in their retinal scans. We also report significant co-occurrences of periodontal disease in subjects reporting swollen joints (p=0.0422) and a family history of eye disease (p=0.0337). These results indicate cross-correlation of poor periodontal health with systemic health outcomes and stress the importance of oral health screenings at the primary care level. Our screening process and analysis method, using images and machine learning, can be generalized for automated diagnoses and systemic health screenings for other diseases.

Motivation & Objective

• To develop an automated, non-invasive method for detecting periodontal disease using machine learning and fluorescent imaging.
• To investigate the correlation between oral health markers and systemic health conditions in a primary care screening context.
• To evaluate the generalizability of a machine learning-based segmentation model for periodontal inflammation using clinical and imaging data.
• To identify significant co-occurrences between periodontal disease and systemic health indicators such as blood pressure, BMI, and ocular or neurological abnormalities.
• To demonstrate the feasibility of integrating automated oral screening into routine primary care for early detection of systemic diseases.

Proposed method

• Acquired 1,215 intraoral fluorescent images from 284 adults (18–90 years) using an FDA-approved intraoral camera under blue light illumination (405–450 nm).
• Applied a machine learning classifier to perform pixel-wise segmentation of inflamed gingiva based on porphyrin fluorescence, with ground truth from expert-labeled clinical images.
• Used modified gingival index (MGI) scores from trained dentists as clinical validation for segmentation performance.
• Correlated MGI scores with self-reported medical history, routine health metrics (blood pressure, BMI), and technology-enabled screenings (ECG, retinal imaging, tympanic membrane, neurological exams).
• Performed statistical analysis (p-values) to identify significant co-occurrences between periodontal disease and systemic conditions across demographic subgroups.
• Validated model performance using AUC, precision, and recall metrics on the segmentation task.

Experimental results

Research questions

• RQ1 Can a machine learning classifier accurately segment periodontal inflammation from intraoral fluorescent images?
• RQ2 What is the strength of the correlation between periodontal disease (as measured by MGI) and systemic health conditions such as optic nerve abnormalities or joint swelling?
• RQ3 Are there significant demographic associations between periodontal disease and systemic health markers (e.g., age, gender, BMI, blood pressure)?
• RQ4 Can automated oral screening using fluorescence imaging and machine learning detect systemic disease indicators with statistical significance?
• RQ5 To what extent can this automated system be generalized for broader use in primary care for systemic health screening?

Key findings

• The machine learning classifier achieved an AUC of 0.677 in segmenting periodontal inflammation, with precision of 0.271 and recall of 0.429.
• Periodontal disease was significantly more prevalent among males (p = 0.0012) and older subjects (p = 0.0224).
• Gingivitis and early periodontal disease were strongly associated with optic nerve abnormalities on retinal scans (p < 0.0001).
• Subjects with periodontal disease were significantly more likely to report swollen joints (p = 0.0422) and a family history of eye disease (p = 0.0337).
• Higher MGI scores were significantly correlated with male gender, older age, and abnormal findings in technology-enabled screenings such as retinal and tympanic membrane exams.
• The study demonstrates that automated oral screening using fluorescence imaging and machine learning can detect systemic health correlations with clinical relevance, supporting integration into primary care.