Preliminary Study for Automated Recognition of Anatomical Structure from Torso CT images

Area of Science:

• Medical Imaging
• Anatomical Structure Recognition
• Computer-Aided Diagnosis (CAD)

Background:

• Accurate human anatomical structure recognition is crucial for developing effective computer-aided diagnosis (CAD) systems.
• Existing methods may lack automated segmentation capabilities for complex anatomical regions like the human torso.

Purpose of the Study:

• To propose an automated image processing scheme for recognizing the general structure of the human torso from CT images.
• To segment the human torso into seven key components: skin, subcutaneous fat, muscle, bone, diaphragm, thoracic cavity, and abdominal cavity.

Main Methods:

• An image processing scheme was developed to automatically identify and segment the human torso region in CT images.
• Segmentation was achieved by analyzing CT number distribution and the spatial relationships between different organ and tissue regions.
• The scheme was applied to a dataset of 313 patient torso CT images.

Main Results:

• The proposed scheme successfully recognized the general structure of the human torso.
• Automatic segmentation into seven distinct parts (skin, subcutaneous fat, muscle, bone, diaphragm, thoracic cavity, abdominal cavity) was achieved.
• Preliminary experiments confirmed the usefulness of the developed image processing scheme.

Conclusions:

• The developed image processing scheme provides an effective method for automated human torso anatomical structure recognition in CT images.
• This automated segmentation capability is valuable for the development and improvement of computer-aided diagnosis (CAD) systems.
• The preliminary validation on 313 patient cases suggests the scheme's practical utility in medical imaging analysis.