Browse Prior Art Database

AUTOMATED CLASSIFICATION OF COMPUTED TOMOGRAPHY AXIAL SLICES INTO ANATOMICAL REGION LABELS

IP.com Disclosure Number: IPCOM000239810D
Publication Date: 2014-Dec-03
Document File: 6 page(s) / 93K

Publishing Venue

The IP.com Prior Art Database

Abstract

The invention proposes a technique to determine the anatomical region labels for a scanned computed tomography (CT) data. The technique detects and labels different regions present in the scan from complete or partial CT body input data. The technique extracts local context and global material specific features from axial slices which are used as input for a random forest classifier which classifies them into anatomical regions.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 53% of the total text.

 

AUTOMATED CLASSIFICATION OF COMPUTED TOMOGRAPHY AXIAL SLICES INTO ANATOMICAL REGION LABELS

BRIEF ABSTRACT

The invention proposes a technique to determine the anatomical region labels for a scanned computed tomography (CT) data. The technique detects and labels different regions present in the scan from complete or partial CT body input data. The technique extracts local context and global material specific features from axial slices which are used as input for a random forest classifier which classifies them into anatomical regions.

KEYWORDS

Automated Classification, Computed Tomography (CT), anatomic region, label, slice, Algorithm, registration, landmark detection, whole body, partial body, local, global material, axial slices.

DETAILED DESCRIPTION

In general, an image analysis algorithm for segmentation, registration or landmark detection which processes a specific anatomy requires information about the spatial location of the desired anatomy of a whole body scan to start with.

A conventional technique automatically determines which portion of the human body is shown by a computed tomography (CT) volume image. This offers various possibilities such as automatic labeling of images or initializing subsequent image analysis algorithms. The technique takes CT volume as input and outputs the vertical body coordinates of its top and bottom slice in a normalized coordinate system. The origin and unit length of the normalized coordinate system are determined by anatomical landmarks.

However, the above mentioned technique is not fast and is difficult to process in parallel with other algorithm required to be processed for scanned CT data.

 

Therefore, there is a need in the art for a technique to efficiently determine the anatomical region labels in the scanned CT data.

The invention proposes a technique to determine the anatomical region labels for the scanned CT data. The goal of the algorithm is to detect and label all the different regions present in the scan from complete or partial CT body input data. The algorithm works by extracting local context and global material specific features from axial slices. The local context and global material specific features are then used as input for a random forest classifier which classifies them into anatomical regions. The technique serves as a starting point for various applications such as organ localization, organ segmentation andlandmark detection among others.

Axial partitioning of whole body CT image is done for lung or abdomen or pelvic cases. Slice based classification is performed with local context features and global material specific features. The classifier model is built using Random Forest algorithm. The algorithm takes whole body CT or partial body CT data as input and provides labeled regions of lungs or abdomen or pelvic as output...