Accuracy evaluation of surface registration algorithm using normal distribution transform in stereotactic body...

Area of Science:

• Medical Physics
• Radiotherapy Technology
• Image Registration

Background:

• Surface registration is crucial for accurate patient positioning in stereotactic body radiotherapy (SBRT) and stereotactic radiosurgery (SRS).
• Traditional methods like the Iterative Closest Point (ICP) algorithm have limitations in achieving optimal accuracy.
• Novel algorithms are needed to enhance precision in image-guided radiation therapy.

Purpose of the Study:

• To evaluate the feasibility and performance of the Normal Distribution Transform (NDT) algorithm.
• To compare NDT with the Iterative Closest Point (ICP) method for surface registration accuracy in SBRT/SRS.
• To determine the clinical utility of NDT in optical imaging systems for radiotherapy.

Main Methods:

• Point cloud images were acquired using a depth camera-based optical imaging (OSI) system.
• Surface registration was performed using both NDT and ICP algorithms.
• Registration error and root-mean-square (RMS) values were calculated for surface model, reposition, and target accuracy, with statistical analysis using a paired t-test.

Main Results:

• NDT demonstrated significantly lower registration errors and RMS values for surface model accuracy compared to ICP (p < 0.05).
• NDT also showed superior performance in reposition accuracy, with reduced average error and RMS values (p = 0.005).
• Overall target accuracy improved with NDT, reducing reposition error and RMS by 0.71 mm and 1.32 mm, respectively (p = 0.03).

Conclusions:

• The Normal Distribution Transform (NDT) algorithm provides more reliable accuracy for surface model, reposition, and target accuracies than the Iterative Closest Point (ICP) method.
• NDT is a feasible and accurate surface registration algorithm for optical imaging systems in SBRT/SRS.
• NDT offers a promising advancement for improving precision in image-guided radiotherapy procedures.