Back

Multimodal non-ionizing quantitative characterization of enamel hypomineralization

Chin, B.; Thapa, D.; Neshatian, M.; Abrams, S.; Ghaderi Moghadam, H.; Dos Santos, P. H.; Casas, M.; Mandelis, A.; Bozec, L.

2026-01-27 biophysics
10.64898/2026.01.26.701762 bioRxiv
Show abstract

Enamel hypomineralization is clinically graded but difficult to quantify with visual and radiographic assessment. We evaluated whether optical coherence tomography (OCT) and lock-in thermography imaging (LITI) provide non-ionizing, quantitative lesion localization and phenotyping, using clinical mDDE classification as a practical benchmark and micro-CT ({micro}CT) mineral density as an in vitro reference in a subset. Twenty-five extracted first permanent molars were classified as control, hypomineralized (Type 1 white-cream; Type 2 yellow-brown), or other enamel defects. Co-localized OCT A-scan decay slopes and LITI lock-in phase-contrast were measured in predefined ROIs; six teeth underwent {micro}CT with hydroxyapatite calibration. Clinical scoring showed substantial agreement (inter-rater {kappa}=0.61; intra-rater {kappa}=0.70). OCT decay slopes differed across clinical groups (Kruskal-Wallis p=0.00028), separating controls from defect groups, while separation between hypomineralization and other defects was limited. LITI detected thermophotonic "hot spots" at many clinically identified sites (sensitivity 80%) but showed low specificity (20%) at a single modulation frequency, consistent with high anomaly sensitivity but limited etiologic discrimination. In the {micro}CT subset, hypomineralized ROIs showed reduced mineral density relative to adjacent enamel and spatial concordance with OCT and LITI contrast. Together, OCT+LITI support objective lesion mapping and motivate multi-frequency thermophotonics and expanded reference sampling to improve specificity for chairside translation.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.