검색 상세

Accuracy and Reproducibility of Automated 3D CBCT-based Cephalometric Analysis in Children and Adolescents

  • 정혜진 (아주대학교 임상치의학대학원)

초록/요약

Objectives: This study aimed to evaluate the accuracy and reproducibility of an automated landmark identification （ALI） system for three-dimensional cone-beam computed tomography （CBCT）-based cephalometric analysis in children and adolescents. Additionally, the influence of dentition stage on system performance was investigated. Materials and Methods: Sixty patients who underwent CBCT imaging for orthodontic diagnosis were retrospectively analyzed and equally divided into mixed dentition and permanent dentition groups. Thirty cephalometric landmarks were automatically identified using the ALI function of Invivo7 software and compared with manual identification by a single examiner. Landmark accuracy and measurement accuracy were assessed using mean error distance （MED） and mean absolute error （MAE）, respectively. Successful detection rate （SDR） and successful measurement rate （SMR） were calculated based on predefined thresholds of 2 mm and 2°. Results: The ALI system demonstrated high reproducibility and comparable accuracy to manual landmark identification in both dentition groups. Although certain skeletal and dental landmarks showed localization errors greater than 2 mm, there were no significant differences in overall MED or MAE between mixed and permanent dentition stages. However, some measurements, including posterior facial height and U1–SN, exhibited relatively high variability. Conclusion: The ALI system can serve as a reliable adjunctive tool for CBCT-based cephalometric analysis in growing patients. However, careful manual verification is recommended for specific landmarks and measurements that demonstrate high variability. Keywords: Cephalometry, Cone-beam computed tomography, Clinical applicability, Mixed dentition, Permanent dentition

more

초록/요약

본 연구는 소아 및 청소년 환자를 대상으로 자동 랜드마크 식별（automated landmark identification, ALI） 시스템을 이용한 3차원 콘빔전산화단층촬영（CBCT） 기반 두부계측 분석의 정확도와 재현성을 평가하고, 치열 단계에 따른 성능 차이를 분석하고자 하였다. 교정 진단을 위해 CBCT를 촬영한 만 20세 미만 환자 60명（혼합치열기 30명, 영구치열기 30명）을 후향적으로 분석하였으며, Invivo7 소프트웨어를 이용하여 30개의 랜드마크를 자동 식별한 뒤 수동 식별 결과와 비교하였다. 랜드마크 및 측정 정확도는 각각 mean error distance（MED）와 mean absolute error（MAE）로 평가하였고, 2 mm 및 2° 기준에 따른 successful detection rate（SDR）와 successful measurement rate（SMR）를 산출하였다. 연구 결과, ALI 시스템은 두 군 모두에서 높은 재현성과 전반적으로 양호한 정확도를 보였으며, 전체 MED 및 MAE에서 군 간 유의한 차이는 나타나지 않았다. 다만 일부 랜드마크와 posterior facial height（PFH）, U1–SN 등의 측정 항목에서는 상대적으로 높은 변동성이 관찰되었다. 이상의 결과를 통해 ALI 시스템은 성장기 환자의 CBCT 기반 두부계측 분석에서 유용한 보조적 도구로 활용될 수 있으나, 변동성이 큰 항목에 대해서는 수동 검증이 필요할 것으로 생각된다.

more

목차

Ⅰ. INTRODUCTION 1
Ⅱ. MATERIALS AND METHODS 3
A. Study subjects 3
B. Study design 5
1. Radiography 5
2. Coordinate system 6
3. Ground truth 8
4. Measurements 8
C. Statistical methods 10
Ⅲ. RESULTS 11
A. Reliability of landmark identification using manual and ALI methods 11
B. Mean error distance (MED) for landmarks 11
C. Successful detection rate (SDR) for landmarks 15
D. Mean Absolute Error (MAE) for measurements 16
E. Successful measurement rate (SMR) for measurements 19
Ⅳ. DISCUSSION 20
Ⅴ. CONCLUSION 24
REFERENCES 25
국문요약 27

more
반출 Meta View 목록