Hybrid Neuromorphic X-Ray Detector for HDR Motion Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current X-ray imaging technologies are limited by low read-out rates, minimum detectability, and saturation limits, which hinder the ability to achieve low photon flux sensitivity, high saturation, and rapid detection of motion, while requiring complex and costly X-ray sources.
Innovation Solution
A hybrid neuromorphic X-ray detector system that combines integrating pixels with neuromorphic pixels, generating two types of datasets - integrating keyframes and asynchronous event streams - to enable deblurred high dynamic range real-time imaging through neural network processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If conventional integrating digital detectors are used, then X-ray photon detection is achieved, but read-out rate is low and motion detection capability is limited
Solution Approach 1:
The detector array is segmented into two distinct types of pixels: integrating pixels for capturing full-frame images and neuromorphic pixels for detecting temporal changes and motion. This segmentation allows simultaneous operation at different speed regimes, with neuromorphic pixels providing high-speed motion detection while integrating pixels maintain comprehensive imaging capability.
Solution Approach 2:
The system dynamically switches between two operational modes: high-speed mode utilizing neuromorphic pixels for rapid motion detection, and real-time mode utilizing integrating pixels for comprehensive image capture. This dynamic operation allows the system to adapt to different imaging requirements and optimize performance for either speed or completeness based on the specific application needs.
2Reliability
If conventional digital detectors are used, then image detection is achieved, but saturation limit is restricted by maximum charge storage capacity
Solution Approach 1:
A scintillator layer serves as an intermediary between X-ray photons and the detector pixels. This scintillator converts X-ray photons into visible light photons, which are then detected by both integrating and neuromorphic pixels. This intermediary mechanism enables the detector to handle a wider dynamic range of X-ray intensities without saturation, as the scintillator's light output can be detected by pixels with different charge storage capacities.
3Ease of manufacture
If conventional X-ray imaging systems are used, then imaging is achieved, but system complexity and cost increase due to requirements for high-power pulsed X-ray sources
Solution Approach 1:
The system replaces the need for complex high-power pulsed X-ray source control mechanisms with a simpler continuous or low-power pulsed X-ray source combined with neuromorphic detection. Instead of relying on high-power pulses to freeze motion, the neuromorphic pixels detect temporal changes in X-ray intensity, substituting the mechanical/power-intensive pulse generation approach with an intelligent detection approach that works with lower power sources.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The system achieves low photon flux sensitivity, high saturation, and rapid motion detection, reducing system complexity and cost by using neuromorphic techniques to improve current X-ray imaging capabilities.
Implementation Method 1
a scintillator configured to convert the incident X-ray photons into a corresponding number of visible light photons
Implementation Method 2
an array of photodetector pixels, each having an associated readout circuit
Data Source
AI summary
An X-ray imaging system and a method thereof are provided. In one or more embodiments, the imaging system includes an X-ray detector having neuromorphic X-ray pixels that can be combined with integrating pixels in a single hybrid detector package. The X-ray detector may generate two types of datasets (e.g., output data streams), one of integrating keyframes, the other an asynchronous event stream of intensity changes. In one or more embodiments, the method may include detecting X-rays from an imageable event using the X-ray detector, generating the two types of datasets based on the detected X-rays, and reconstructing one or more images of the imageable event based on the two types of generated datasets. By utilizing artificial intelligence/machine learning, these datasets may be combined to generate novel x-ray imaging modes that include deblurred high dynamic range real-time frame sequences or a high-speed, high dynamic range frame sequence for subsequent viewing.