TL;DR
This paper introduces a novel optical system for measuring pH in dental plaque using a single FDA-approved fluorescent dye, fluorescein. By employing a spectral unmixing algorithm, the system overcomes limitations of previous methods and demonstrates potential for early detection of caries susceptibility in hard-to-reach areas of the mouth.
What Problem Does This Paper Address?
Dental caries, or tooth decay, is a widespread issue often initiated by the formation of plaque – a biofilm on tooth surfaces. Within this plaque, acid-producing bacteria metabolize sugars, leading to a drop in pH. When this pH falls below a critical threshold (around 5.5), it causes demineralization of tooth enamel, a precursor to cavities. Early detection of this acid production is crucial for preventative dental care. However, existing methods for measuring plaque pH, such as pH probes or indicator strips, have significant drawbacks. pH probes can be invasive and may not accurately reflect the true acidity of the biofilm, while indicator strips struggle to reach the complex pits and fissures of teeth where plaque often accumulates. This paper aims to address the need for a non-contact, accurate, and accessible method for measuring plaque pH, particularly in these challenging dental locations.
Methods
The core of this work lies in a custom-built optical system and a sophisticated data analysis technique. The system utilizes a pulsed 420-nm LED as an excitation source for fluorescein (FL), an FDA-approved dye known for its fluorescence properties. The emitted fluorescence spectrum is captured by a spectrometer.
The key innovation is the use of a spectral unmixing algorithm. Fluorescein exists in different pH-sensitive molecular forms (anion and dianion) that have overlapping spectral profiles. Traditional intensity-based fluorescence measurements are prone to errors from factors like dye concentration, photobleaching, and excitation intensity variations. This algorithm, however, analyzes the shape of the fluorescence spectrum by treating it as a mixture of the pure anion and dianion spectral "endmembers." By using a least-squares regression, it determines the relative proportion of these species, which directly correlates to the pH. This approach makes the measurement ratiometric and thus less susceptible to the aforementioned confounding factors.
The system also incorporates methods to mitigate noise from tooth autofluorescence (AF) and ambient light through spectral subtraction and filtering. A custom-designed dental probe, shaped like a toothbrush head, allows for non-contact measurements in occlusal pits and fissures.
Datasets
The paper describes experiments using:
- Buffered fluorescein solutions: Used for calibrating the pH measurement system across a pH range of 4.5 to 7.
- In vitro tooth samples: Extracted molars were used to test the system's ability to measure pH on a tooth surface and to validate AF removal.
- In vivo human subjects: A pilot study was conducted on eight human subjects to measure resting plaque pH and the dynamic pH response after a sucrose rinse.
No new benchmarks were introduced; rather, the system's performance was compared against a gold standard pH meter for buffer solutions and in vitro tooth samples.
Key Insights and Ideas
- Spectral unmixing for pH sensing: The realization that the relative spectral contributions of fluorescein's anion and dianion forms can be used as a robust pH indicator, overcoming the limitations of simple intensity measurements.
- Non-contact measurement in challenging areas: The development of a dental probe that can access and measure pH in occlusal pits and fissures, areas notoriously difficult to assess with traditional methods.
- FDA-approved dye for clinical relevance: The use of fluorescein, an FDA-approved dye, significantly lowers the barrier to clinical translation and patient acceptance.
- Immunity to confounding factors: The spectral unmixing approach inherently accounts for variations in dye concentration, excitation intensity, and photobleaching, leading to a more reliable measurement.
Key Results
- Calibration: Linear calibration curves were established for both anion and dianion fractions, showing high R² values (0.99).
- Buffer Sample Verification: The FL system demonstrated good agreement with a gold standard pH meter, with a root mean square error (RMSE) of 0.14 pH for the anion calibration curve and 0.2 pH for the dianion curve.
- In Vitro Tooth Samples: The system showed repeatable results on extracted teeth, with an RMSE of 0.3 pH and a standard deviation of 0.2 pH.
- In Vivo Pilot Study: The system successfully measured resting pH and the dynamic pH drop following a sucrose rinse in human subjects. In six out of eight patients, a pH drop of 0.4 pH or more was observed within 5 minutes of the sucrose rinse. The average resting pH on a control tooth was 6.91, with an average pH drop of 0.4 pH after the sucrose rinse.
Limitations and Future Work
- Ambient Light and Autofluorescence Interference: While mitigated, these factors still influence measurement accuracy, particularly in vivo. The authors suggest using a lock-in amplifier and a dye-in-polymer filter to further reduce interference.
- Illumination Area Variation: The spectral noise removal was susceptible to variations in illumination area. Averaging multiple spots or incorporating an imaging feature to guide the probe could improve this.
- Dye Retention: For interproximal measurements, the FL solution's viscosity was a limitation. Adding glycerol is suggested to improve retention.
- Limited Clinical Correlation: The pilot study had a small sample size and lacked sufficient representation of different caries risk levels, making it difficult to establish a direct correlation between pH measurements and caries susceptibility. A larger clinical study is needed.
- Gold Standard Comparison: The lack of a direct gold standard for intraoral pH measurement in pits and fissures remains a challenge. Future work may involve comparing with scraped plaque pH measurements, acknowledging the limitations of that method.
- Device Simplification: For widespread clinical use, replacing the spectrometer with multiple filtered photodiodes could lead to a more compact and cost-effective device. Machine learning could also be integrated for improved accuracy and heatmap generation.
Summary updated