Title
AI-based predictions of nickel-titanium file separation using thermal imaging: An in vitro study
Authors
Swagat Panda1, Meetkumar Dedania2,*, Ramandeep3, Vipin Arora4, Shaswatee Panda1 & Akash Dupper5
Affiliation
1Department of Conservative Dentistry & Endodontics, Hi Tech Dental College and Hospital, Bhubaneswar, India; 2Department of Conservative Dentistry and Endodontics, K. M. Shah Dental College and Hospital, Sumandeep Vidyapeeth Deemed to be University, Piparia, Vadodara, Gujarat, India; 3Department of Conservative Dentistry & Endodontics, Maharishi Markendshwar College of Dental Sciences & Research, Mullana, Haryana, India; 4Department of Restorative Dental Sciences, Faculty of Dentistry, Taif University, Taif, Kingdom of Saudi Arabia; 5Private practitioner, Akash dental clinic , Yamunangar, India; *Corresponding author
Swagat Panda - E-mail: reachswagat@gmail.com
Meetkumar Dedania - E-mail: meet97247da@gmail.com
Ramandeep - E-mail: renu.malh7@gmail.com
Vipin Arora - E-mail: vipinendodontist@gmail.com
Shaswatee Panda - E-mail: docshaswatee.panda@gmail.com
Akash Dupper - E-mail: akashdentalclinic@yahoo.com
Article Type
Research Article
Date
Received August 1, 2026; Revised August 31, 2026; Accepted August 31, 2026, Published August 31, 2026
Abstract
Nickel-titanium rotary file separation is a sudden procedural complication and clinicians currently lack a non-destructive warning system that can identify fatigue-related failure before fracture occurs. Therefore, it is of interest to evaluate the whether thermal imaging combined with machine learning could predict imminent file separation during simulated curved-canal instrumentation. A total of 120 rotary files were tested in standardized stainless-steel curved canals and thermal signals, time to fracture, rotational cycles and file surface changes were recorded for artificial intelligence model training and validation. Files that separated showed a significantly higher terminal temperature rise than non-separated controls (6.8 ± 1.7°C vs 2.4 ± 0.9°C; p<0.001) and the gradient boosting model predicted separation with 91.7% accuracy, 0.93 sensitivity, 0.90 specificity and an AUC of 0.96. Thus, thermal imaging-based artificial intelligence showed high predictive performance for impending nickel-titanium file separation and may support safer preclinical assessment of rotary endodontic instruments.
Keywords
Nickel-titanium file, thermal imaging, artificial intelligence (AI), cyclic fatigue, file separation, endodontics, machine learning (ML)
Citation
Panda et al. Bioinformation 22(8): 5502-5506 (2026)
Edited by
Hiroj Bagde
ISSN
0973-2063
Publisher
License
This is an Open Access article which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. This is distributed under the terms of the Creative Commons Attribution License.
