U-Net Neural Network for 4D Cardiac Image Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current medical image segmentation techniques struggle with accurately segmenting 3-dimensional and 4-dimensional medical images of internal organs, particularly the heart, due to poor resolution, contrast, and the complexity of internal details and dynamic structures.
Innovation Solution
A computer-implemented method using a trained artificial neural network with a U-net architecture for generating 3-dimensional and 4-dimensional medical image segmentations of internal organs. The method involves providing n-dimensional medical images, where n=3 or 4, and using convolutional processing networks to segment structures within these images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional image segmentation techniques are used on 3D/4D medical images, then processing is simplified, but segmentation accuracy deteriorates due to poor resolution and contrast
Solution Approach 1:
The patent transitions from traditional 2D image segmentation to 3D and 4D medical image segmentation, adding spatial and temporal dimensions to capture complex internal organ structures and their dynamics throughout the cardiac cycle, thereby improving segmentation accuracy for three-dimensional and time-dependent structures
Solution Approach 2:
The patent introduces a U-Net architecture with skip connections as an intermediary mechanism that bridges the gap between input medical images and segmentation outputs, enabling accurate segmentation by combining low-level spatial details with high-level semantic information through its unique network topology
2Productivity
If physicians manually segment images, then control over segmentation is maintained, but time consumption increases significantly
Solution Approach 1:
The patent implements an automated segmentation system using trained neural networks that independently perform segmentation tasks without requiring continuous physician intervention, allowing the system to self-process medical images and generate segmentations autonomously, thereby significantly reducing time consumption
Solution Approach 2:
The patent incorporates feedback mechanisms where segmentation results are validated and refined through comparison with ground truth data during training, enabling the model to continuously improve its performance and maintain high accuracy while operating autonomously
3Adaptability or versatility
If 3D/4D segmentation is performed, then dynamic views and overlays are enabled, but computational resources required increase
Solution Approach 1:
The patent performs segmentation computations in advance during the training phase using pre-labeled medical images, creating a reusable model that can quickly generate segmentations for new images without requiring intensive real-time computational resources, thereby enabling dynamic visualization with lower ongoing energy consumption
Data Source
AI summary
A system and a computer-implemented method are provided for generating at least one 4-dimensional medical image segmentation for at least one structure of a human heart. The method includes providing a first 4-dimensional medical image comprising the at least one structure of the human heart, the medical image being based on a computed tomography scan image; generating a segmentation of at least part of the provided first 4-dimensional medical image using at least one first trained artificial neural network, wherein the at least one first trained artificial neural network is configured as a convolutional processing network with U-net architecture; and generating at least one 4-dimensional medical image segmentation for the at least one structure of the human heart based at least on the segmentation generated by the at least one first trained artificial neural network.