Echocardiogram Video Transformer for Precise LVEF Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Echocardiography for measuring left ventricular ejection fraction (LVEF) is subject to intra- and inter-observer variability, and existing deep learning methods lack effective spatiotemporal analysis, leading to inaccuracies and increased computational requirements.
Innovation Solution
The ViViEchoformer model uses a video vision transformer (ViViT) to directly regress LVEF from echocardiogram videos, capturing spatiotemporal tokens and preserving inter-frame relationships without segmentation, thus providing precise, fully automatic EF predictions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If human interpretation is used for echocardiogram analysis, then flexibility and adaptability are maintained, but measurement precision and reliability are compromised due to intra- and inter-observer variability
Solution Approach 1:
The patent replaces human mechanical interpretation with an automated deep learning system (ViViEchoformer) that uses video vision transformer architecture to analyze echocardiogram videos. The model extracts spatiotemporal features and predicts ejection fraction automatically, eliminating human observer variability while maintaining measurement precision and reliability.
2Extent of automation
If existing deep learning methods are used for echocardiogram analysis, then automation is improved, but measurement precision deteriorates due to lack of effective spatiotemporal analysis
Solution Approach 1:
The patent segments the echocardiogram video into temporal frames and extracts spatiotemporal tokens from each frame. The ViViEchoformer model processes these segmented components through transformer architecture, preserving inter-frame relationships while achieving accurate ejection fraction predictions, thus resolving the contradiction between automation and precision.
Solution Approach 2:
The patent transitions from static image analysis to dynamic video analysis by incorporating temporal dimension. The model processes sequences of echocardiogram frames, capturing spatiotemporal patterns that provide more comprehensive information for accurate ejection fraction measurement while maintaining full automation.
3Measurement precision
If traditional echocardiogram analysis methods are used, then computational resources are minimized, but measurement precision and analysis depth are limited
Solution Approach 1:
The patent performs preliminary feature extraction and representation learning during the training phase, where the ViViEchoformer model learns to identify critical spatiotemporal patterns. This preliminary action enables the model to make accurate predictions during inference without requiring excessive computational resources, as the heavy lifting is done during training on standardized datasets.
Data Source
AI summary
A method comprising is described herein comprising receiving first magnetic resonance imaging (MRI) data of a first plurality of subjects, wherein the first MRI image data comprises a first plurality of two-dimensional (2D) images, pre-processing the first MRI data for analysis, converting each two-dimensional image of the first plurality of 2D images into first tokens as input for a transformer encoder, wherein the transformer encoder comprises a time series classification transformer, training the transformer encoder using the first input tokens, receiving second MRI data of a subject, wherein the second MRI data comprises a second plurality of 2D images, converting each image of the second plurality of 2D images into second tokens as input to the trained transformer encoder, and applying the trained transformer encoder to the second tokens to predict a state of disease in a subject.