Automated vertebrae localization and identification by decision forests and image-based refinement on real-world CT...

Area of Science:

• Medical imaging
• Radiology
• Computer-aided diagnosis

Background:

• Accurate localization of vertebral bodies is crucial for various medical applications, including diagnosis and treatment planning.
• Existing methods for vertebral body identification in computed tomography (CT) scans can be time-consuming and prone to inter-observer variability.

Purpose of the Study:

• To develop a fully automatic algorithm for the precise localization and identification of vertebral bodies in CT images.
• To establish a robust and efficient tool for automated analysis of thoraco-abdominopelvic CT scans.

Main Methods:

• A two-stage automated approach was employed, utilizing decision forests for initial vertebral body prediction.
• Morphological image processing techniques were applied to refine detection by identifying the spinal canal's position.
• The algorithm was trained and validated on a dataset of 232 retrospective thoraco-abdominopelvic CT scans.

Main Results:

• The algorithm demonstrated a mean distance error of 13.7 mm between predicted and actual vertebral body centroids.
• An identification rate of 79.6% was achieved for the thoracic region.
• An identification rate of 74.8% was achieved for the lumbar segment.

Conclusions:

• The developed algorithm offers a novel, automated method for detecting and identifying vertebral bodies in CT scans.
• This automated approach has the potential to enhance the efficiency and accuracy of vertebral analysis in clinical practice.
• The algorithm is applicable to CT scans with arbitrary fields of view, broadening its utility.