Anatomical Vector Training for Low-Compute Medical Image Labeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for segmenting and labeling medical patient images require significant computational effort due to the need for registering patient images with anatomical atlases.
Innovation Solution
A method involving a tracked imaging device and an anatomical atlas is used to establish a transformation between the atlas space and the patient space, training a learning algorithm with anatomical vectors and labels to enable efficient segmentation and labeling of patient images without repeated atlas registration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If patient images are registered with anatomical atlases for segmentation and labeling, then segmentation accuracy is improved, but computational effort increases significantly
Solution Approach 1:
The patent pre-calculates and stores anatomical vectors and pre-trained learning algorithms during an offline preparation phase. This preliminary action allows the system to avoid performing computationally intensive atlas registration during actual patient image analysis, thereby reducing real-time computational effort while maintaining segmentation accuracy through the use of pre-computed anatomical information
Solution Approach 2:
The patent creates simplified representations (anatomical vectors) that capture essential anatomical relationships without requiring the full complexity of the original atlas images. These vector representations serve as lightweight copies that can be processed efficiently while preserving the necessary anatomical information for accurate segmentation
2Measurement precision
If atlas registration is performed for each patient image, then anatomical accuracy is improved, but processing time increases
Solution Approach 1:
The system performs atlas-based anatomical vector calculation and learning algorithm training in advance during an offline phase. This preliminary computation stores the anatomical knowledge in a compact form that can be quickly applied to patient images without requiring repeated registration operations, thus reducing processing time while maintaining anatomical accuracy
Solution Approach 2:
The patent extracts essential anatomical relationship information from full atlas images and represents it in compact vector form. This extraction process separates the critical anatomical knowledge from the computationally burdensome image data, enabling fast processing while preserving anatomical accuracy through the use of these extracted anatomical vectors
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
A computer-implemented medical method of training a learning algorithm for determining a relation between a label for indicating a position or type of an anatomical structure in a medical image on the one hand and the position of the anatomical structure in the medical image on the other hand, the method comprising the following steps: a) patient training image data is acquired (S21) which describes digital medical images of an anatomical structure of a plurality of patients; b) atlas data is acquired (S22) which describes a model of an anatomical body part including the anatomical structure; c) viewing direction data is acquired (S23) which describes the viewing direction of an imaging device towards the anatomical structure at the point in time when the imaging device was used to generate the medical image; d) anatomical vector data is determined (S24) based on the viewing direction data and the atlas data, wherein the anatomical vector data describes an anatomical vector which is a result of transforming the viewing direction into a reference system in which positions in the image-based model are defined; e) label data is acquired (S25) which describes a label describing the position or type of the anatomical structure in the model of the anatomical body part; and f) anatomical indicator data is determined (S26) based on the patient image data and the anatomical vector data and the label data, wherein the anatomical indicator data describes model parameters of a learning algorithm for establishing the relation between the position or type of the anatomical structure described by the medical image and the label, wherein the anatomical indicator data is determined by inputting the patient training image data and the label data into a function which establishes the relation.