cs.CVMar 10, 2026

OrthoAI v2: From Single-Agent Segmentation to Dual-Agent Treatment Planning for Clear Aligners

Authors: Lansiaux Edouard, Leman Margaux

Organizations: STaR-AI Research Group, Emergency Department, Lille University Hospital

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

We present OrthoAI v2, the second iteration of our open-source pipeline for AI-assisted orthodontic treatment planning with clear aligners, substantially extending the single-agent framework previously introduced. The first version established a proof-of-concept based on Dynamic Graph Convolutional Neural Networks (\dgcnn{}) for tooth segmentation but was limited to per-tooth centroid extraction, lacked landmark-level precision, and produced a scalar quality score without staging simulation. \vtwo{} addresses all three limitations through three principal contributions: (i)a second agent adopting the Conditioned Heatmap Regression Methodology (\charm{})\cite{rodriguez2025charm} for direct, segmentation-free dental landmark detection, fused with Agent~1 via a confidence-weighted orchestrator in three modes (parallel, sequential, single-agent); (ii)~a composite six-category biomechanical scoring model (biomechanics ×\times 0.30 + staging ×\times 0.20 + attachments ×\times 0.15 + IPR ×\times 0.10 + occlusion ×\times 0.10 + predictability ×\times 0.15) replacing the binary pass/fail check of v1; (iii)~a multi-frame treatment simulator generating F=A×rF = A \times r temporally coherent 6-DoF tooth trajectories via SLERP interpolation and evidence-based staging rules, enabling ClinCheck 4D visualisation. On a synthetic benchmark of 200 crowding scenarios, the parallel ensemble of OrthoAI v2 reaches a planning quality score of 92.8±4.192.8 \pm 4.1 vs.\ 76.4±8.376.4 \pm 8.3 for OrthoAI v1, a +21%+21\% relative gain, while maintaining full CPU deployability (4.2±0.84.2 \pm 0.8~s).

Figures & tables

MovementIncisorCaninePremolarMolar
TxT_{x} MD (mm)4.03.53.52.0
TyT_{y} BL (mm)2.52.53.02.5
TzT_{z} intrusion2.0 mm ( η=0.69\eta=0.69 )
TzT_{z} extrusion1.5 mm ( η=0.42\eta=0.42 , push-only)
RxR_{x} torque ( ∘ )1512108
RyR_{y} tip ( ∘ )1010108
Table 1: Biomechanical movement limits per tooth type. Limits from Glaser [ 13 ] , predictability scores from Kravitz et al. [ 14 ] .
SystemQuality ↑\uparrowσ↓\sigma\downarrowFeas. (%) ↑\uparrowTime (s)
OrthoAI v1 (DGCNN only)76.48.3782.1±0.42.1\pm 0.4
v2: DGCNN only88.3±5.288.3\pm 5.25.2783.4±0.63.4\pm 0.6
v2: CHaRM only89.1±4.889.1\pm 4.84.8810.9±0.10.9\pm 0.1
v2: Sequential91.2±4.391.2\pm 4.34.3862.1±0.42.1\pm 0.4
v2: Parallel (ens.)92.8±4.1\mathbf{92.8\pm 4.1}4.1894.2±0.84.2\pm 0.8
Table 2: OrthoAI v1 vs. v2 on 200 synthetic crowding scenarios. Quality: planning score (0–100). Feas.: fraction of feasible plans. CPU time on Intel i7-11800H.
MethodMEDE (mm) ↓\downarrowMSR (%) ↑\uparrowGPU Time (s)
ALIIOS [ 5 ]2.1748.024.70
TS-MDL [ 4 ]1.8059.03.55
OrthoAI v1 (centroids) [ 1 ]∼\sim 1.80 †–2.1
CHaRNet [ 2 ]1.1268.50.24
Agent 2 v2 (ours)1.38±0.211.38\pm 0.2164.2±3.164.2\pm 3.10.310.31
† Estimated from centroid localisation error reported in [ 1 ] .
Table 3: Landmark detection performance. SOTA from [ 2 ] . Agent 2 on our synthetic dataset.
Limitationv1v2
Centroid-only geometryYesResolved
Scalar quality scoreYesResolved
No staging simulationYesResolved
No SaaS interfaceYesResolved
Synthetic data onlyYesOngoing
Automatic movement gen.NoPlanned
Table 4: Limitation comparison: v1 vs. v2.

Explore similar work

CardsList
Feb 23, 2026

OrthoAI: A Neurosymbolic Framework for Evidence-Grounded Biomechanical Reasoning in Clear Aligner Orthodontics

Automated clinical decision support for clear aligner orthodontics faces a key challenge: bridging geometric perception (3D tooth segmentation) with clinical reasoning (biomechanical feasibility). We address this with OrthOAI, introducing three methodological contributions. First, sparse-supervision segmentation: a landmark-to-point-cloud synthesis protocol enables training from sparse anatomical annotations (6-8 points per tooth) instead of dense labels, combined with a clinically stratified loss mixing label-smoothed cross-entropy and a batch-adaptive Dice term for class imbalance. Second, knowledge-grounded constraint inference: biomechanical feasibility is modeled as a Constraint Satisfaction Problem over a domain ontology of tooth movements, encoding evidence-based per-stage limits as soft and hard constraints. Third, multi-criteria treatment evaluation: treatment quality is scored through a formal Multi-Criteria Decision Analysis framework using a weighted Additive Value Function grounded in clinical priority theory. On landmark-reconstructed point clouds from 3DTeethLand (MICCAI 2024), segmentation reaches 81.4% Tooth Identification Rate with 60,705 parameters. Ablations quantify the impact of each design choice. End-to-end inference runs in under 4 seconds on CPU. We also outline the gap between the current prototype-trained on synthetic ellipsoidal approximations-and clinical deployment, with a roadmap for validation. Code and weights are released.

Edouard Lansiaux, Margaux Leman, Mehdi Ammi

STaR-AI, Emergency Department, Lille University Hospital, 59000, Lille, France · Artificial Intelligence and Data Semantics Laboratory, Paris 8 University, 93200, Saint-Denis, France

3D Point Cloud Segmentation3D Medical Image SegmentationWeakly Supervised Learning
AskPDF
Apr 1, 2026

MATHENA: Mamba-based Architectural Tooth Hierarchical Estimator and Holistic Evaluation Network for Anatomy

Dental diagnosis from Orthopantomograms (OPGs) requires coordination of tooth detection, caries segmentation (CarSeg), anomaly detection (AD), and dental developmental staging (DDS). We propose Mamba-based Architectural Tooth Hierarchical Estimator and Holistic Evaluation Network for Anatomy (MATHENA), a unified framework leveraging Mamba's linear-complexity State Space Models (SSM) to address all four tasks. MATHENA integrates MATHE, a multi-resolution SSM-driven detector with four-directional Vision State Space (VSS) blocks for O(N) global context modeling, generating per-tooth crops. These crops are processed by HENA, a lightweight Mamba-UNet with a triple-head architecture and Global Context State Token (GCST). In the triple-head architecture, CarSeg is first trained as an upstream task to establish shared representations, which are then frozen and reused for downstream AD fine-tuning and DDS classification via linear probing, enabling stable, efficient learning. We also curate PARTHENON, a benchmark comprising 15,062 annotated instances from ten datasets. MATHENA achieves 93.78% mAP@50 in tooth detection, 90.11% Dice for CarSeg, 88.35% for AD, and 72.40% ACC for DDS.

Kyeonghun Kim, Jaehyung Park, Youngung Han +17

OUTTA, Seoul, South Korea · Seoul National University, Seoul, South Korea · Sangmyung University, Seoul, South Korea +4

Dental ImagingMedical ImagingMedical Image SegmentationMedical Image Anomaly DetectionObject DetectionState Space Models
AskPDF
Apr 9, 2026

OralAgent: Integrating Reasoning, Tools, and Knowledge for Interactive Dental Image Analysis

Dental image analysis plays a pivotal role in supporting accurate diagnosis and treatment planning in oral healthcare. Although recent advances have produced dental AI models for specific tasks and individual imaging modalities, their isolated designs limit practical use in real-world clinical workflows. In this paper, we present OralAgent, the first dental-specialized AI agent that unifies multimodal reasoning, tool-based decision-making, and knowledge-grounded retrieval within an end-to-end automated framework. It integrates 22 visual analysis tools and 368 widely-used classical dental textbooks, enabling autonomous reasoning, planning, tool use, knowledge retrieval, and multi-step workflow execution. Furthermore, we introduce OralCorpus, a large-scale, high-quality bilingual textual resource containing 134.8M tokens curated for dental retrieval-augmented generation (RAG). To evaluate models' multidisciplinary dental knowledge, we construct OralQA-ZH, a Chinese multiple-choice question benchmark consisting of 798 items across eleven oral subspecialties. Extensive experiments demonstrate that OralAgent achieves state-of-the-art performance on the MMOral-Uni, MMOral-OPG, and OralQA-ZH benchmarks, highlighting its effectiveness, interpretability, and adaptability in real-world clinical settings. The code and models are publicly available at https://github.com/isjinghao/OralAgent.

Jing Hao, Siyuan Dai, Yongxin Zhang +13

Faculty of Dentistry, the University of Hongkong, Hong Kong SAR, China · Department of Electrical and Computer Engineering, University of Pittsburgh, Pittsburgh, PA, USA · Shenzhen University, China +6

Medical Image AnalysisDental ImagingMultimodal Agents
AskPDF