Accuracy, Reliability, and Repeatability of a Novel Artificial Intelligence Algorithm Converting Two-Dimensional...

Area of Science:

• Medical imaging
• Artificial intelligence in medicine
• Surgical planning

Background:

• Three-dimensional (3D) imaging is crucial for accurate preoperative surgical planning but incurs additional costs and risks.
• Advanced technologies like robotics necessitate precise 3D visualization for surgical execution.

Purpose of the Study:

• To evaluate the measurement accuracy, reliability, and repeatability of a novel artificial intelligence (AI) algorithm.
• To assess the AI algorithm's capability to convert two-dimensional (2D) radiographs into 3D bone models.

Main Methods:

• An AI algorithm was developed to generate 3D bone reconstructions from 2D radiographs.
• Accuracy was assessed by comparing mean absolute errors between AI-generated models, 3D CT scans, and manual measurements on cadaveric knees.
• Reliability and repeatability were evaluated using inter-observer and intra-observer agreement analyses.

Main Results:

• The AI algorithm demonstrated excellent accuracy, with mean absolute errors under 2mm for 9 out of 12 anatomical parameters when compared to CT scans.
• Inter-observer and intra-observer correlation coefficients exceeded 0.90, indicating high reliability and repeatability.
• The AI algorithm's performance was comparable to CT scans in terms of measurement accuracy.

Conclusions:

• The AI algorithm shows high accuracy, reliability, and repeatability in converting 2D radiographs to 3D bone models, similar to CT scans.
• This AI tool offers potential for efficient, cost-effective, and safer preoperative surgical planning by reducing the need for 3D imaging and associated risks.