CN-122004909-A - CT dose self-adaptive optimization system and method thereof

CN122004909ACN 122004909 ACN122004909 ACN 122004909ACN-122004909-A

Abstract

The invention discloses a CT dose self-adaptive optimization system and a method thereof, belonging to the technical field of medical images, comprising the following steps: the system comprises a CT scanner, an image data processing workstation, a high risk group analysis device, a dosage parameter self-adaptive optimization device, a dosage parameter automatic adjustment device and an intelligent learning unit, wherein the dosage parameter optimization problem is improved from Euclidean space to Riemann manifold space based on an information geometry theory, a dosage parameter manifold and an image quality manifold are constructed, a Fisher information matrix is used as Riemann measurement, a geodesic line is calculated on the manifold, and an optimal dosage path is found. The system adopts a double-flow collaborative optimization mechanism, establishes a mapping relation between two manifolds, combines a geometric guided genetic algorithm to realize parameter accurate optimization, realizes individual dose control through pre-scanning, high-risk group analysis, information geometric optimization and automatic parameter adjustment, reduces the radiation dose by 30-45% in addition compared with the traditional method, simultaneously maintains or improves the image quality, and obviously improves the safety of CT examination.

Inventors

HU YI
ZHOU XIN
MA ZHIQIANG
MOU SHUAI
LI BINQI
LI XIAOYAN
BAI YIBING

Assignees

中国人民解放军总医院第五医学中心

Dates

Publication Date: 20260512
Application Date: 20251224

Claims (10)

A ct dose adaptive optimization system, comprising: The CT scanner is used for carrying out CT scanning according to the scanning parameters and generating CT image data; the image data processing workstation is connected with the CT scanner and is used for processing the CT image data; The high-risk group analysis device is connected with the image data processing workstation and is used for receiving the image data of the patient to be diagnosed, which is sent by the image data processing workstation, analyzing the high-risk group based on the image data of the patient to be diagnosed, generating the high-risk group layering number, and sending the high-risk group layering number to the image data processing workstation; the dose parameter self-adaptive optimization device is connected with the high-risk group analysis device and the image data processing workstation and is used for receiving the high-risk group layering, receiving pre-scanning image data, constructing an information geometric manifold based on the high-risk group layering and the pre-scanning image data, calculating optimal dose parameters on the information geometric manifold, and outputting the optimal dose parameters, wherein the information geometric manifold comprises a dose parameter manifold and an image quality manifold; The automatic dosage parameter adjusting device is connected with the dosage parameter self-adaptive optimizing device and the CT scanner and is used for receiving the optimal dosage parameter and automatically adjusting the scanning parameter of the CT scanner according to the optimal dosage parameter, and And the intelligent learning unit is connected with the dose parameter self-adaptive optimization device and the image data processing workstation and is used for updating the information geometric manifold according to the historical optimization result.
2. The system of claim 1, wherein the dose parameter adaptive optimization means comprises: the manifold construction module is used for mapping the high-risk group layering number to an n-dimensional parameter space to construct an n-dimensional information geometric manifold, wherein n is the high-risk group layering number; The Fisher information matrix calculation module is used for calculating a Fisher information matrix on the information geometric manifold, and the Fisher information matrix is used as a Riemann metric; the geodesic calculation module is used for calculating an optimal path of the dose parameter on the information geometric manifold; The information entropy evaluation module is used for calculating the information entropy of the image quality, comparing the information entropy with a preset entropy threshold value and judging whether the image quality is qualified or not.
3. The system of claim 2, wherein the dose parameter adaptive optimization means further comprises: The double manifold mapping module is used for establishing a mapping relation between the dose parameter manifold and the image quality manifold; The mapping learning module is used for training the mapping relation based on the historical mapping data; And the covariance structure optimization module is used for calculating a covariance matrix of parameter change and quality change and adjusting an optimization path based on the covariance matrix.
4. The system of claim 1, wherein the dose parameter adaptive optimization means further comprises: a genetic algorithm optimization module for performing geometry-based genetic algorithm optimization on the information geometry manifold; The fitness function construction module is used for constructing a fitness function based on the geodesic distance; the geometric genetic operator module is used for executing geodetic crossover operation, space-cutting mutation operation and geodetic selection operation; and the population dynamics management module is used for dynamically adjusting the population scale and distribution according to the manifold local curvature.
5. The system of claim 1, wherein the high risk group analysis device comprises: The deep learning network is used for analyzing the image data of the patient to be diagnosed; The risk stratification module is used for carrying out risk stratification on patients according to the analysis result of the deep learning network to generate the high-risk group stratification number; the risk database is used for storing risk layering standards and historical risk assessment data; and the pre-scanning parameter generation module is used for generating pre-scanning parameters according to the high-risk group layering number and controlling the CT scanner to generate the pre-scanning image data.
6. The system of claim 5, wherein the deep learning network of the high risk group analysis device is a generated countermeasure network for obtaining sample images of the patient to be diagnosed for high risk group analysis after training.
7. The system of claim 1, wherein the image quality information entropy on the image quality manifold is expressed as: H(y)=-∑p(y)log(p(y)) wherein y is an image quality quantization characterization value, p (y) is the probability of the image quality quantization characterization value y, and the image quality information entropy threshold is 1.1.
8. The system of claim 1, wherein the automatic dose parameter adjustment device comprises: the tube current self-adaptive adjusting module is used for adjusting the tube current of the CT scanner according to the optimal dose parameter; The tube voltage self-adaptive adjusting module is used for adjusting the tube voltage of the CT scanner according to the optimal dose parameter; and the scanning mode self-adaptive selection module is used for selecting one scanning mode of CT flat scanning, low-dose CT scanning or high-definition CT scanning according to the layering number of the high-risk group.
9. The system of claim 1, wherein the intelligent learning unit comprises: the optimization result evaluation module is used for evaluating the dose optimization effect and the image quality; the Fisher information matrix updating module is used for updating the Fisher information matrix according to the latest scanning result; A mapping function optimization module for optimizing a mapping function between the dose parameter manifold and the image quality manifold; and the fitness function adjusting module is used for adjusting the weight coefficient of the fitness function according to the historical optimization result.
A method of ct dose adaptive optimization comprising: receiving image data of a patient to be diagnosed sent by an image data processing workstation; Analyzing the high-risk group based on the image data of the patient to be diagnosed to generate the number of high-risk group layers; generating a pre-scanning parameter according to the layering number of the high-risk group; controlling a CT scanner to perform pre-scanning according to the pre-scanning parameters to generate pre-scanning image data; Constructing an information geometric manifold based on the high risk group layering number and the pre-scanning image data, wherein the information geometric manifold comprises a dose parameter manifold and an image quality manifold; Calculating an image quality information entropy on the information geometric manifold, and judging whether the image quality is qualified or not; calculating optimal dose parameters on the information geometric manifold by applying a geometric-based genetic algorithm; Outputting the optimal dose parameter to a dose parameter automatic adjusting device; automatically adjusting scan parameters of the CT scanner based on the optimal dose parameters, and And updating the information geometric manifold according to the historical optimization result.

Description

CT dose self-adaptive optimization system and method thereof Technical Field The invention relates to the technical field of medical images, in particular to a CT dose self-adaptive optimization system and method based on information geometry and artificial intelligence, which are used for reducing CT scanning radiation dose on the premise of ensuring image quality. Background Computed Tomography (CT) plays an irreplaceable role in clinical applications as an important tool for modern medical diagnostics. However, the X-ray radiation generated during CT scanning presents a potential risk to human health, particularly for patients, children, and radiation-sensitive people who require repeated CT examinations. Therefore, on the premise of ensuring that the image quality meets the diagnosis requirement, how to effectively reduce the CT radiation dose has become an important research topic in the field of medical imaging. The existing CT dose optimization method mainly comprises an automatic tube current adjustment technology based on body type, an iterative reconstruction algorithm, a low-dose scanning protocol and the like. Although the radiation dose is reduced to a certain extent, the method has the following limitations that firstly, the adjustment of dose parameters is usually based on a simple linear relation or a table look-up method and cannot be accurately adapted to individual differences of different patients, secondly, the existing method usually only considers the body type or scanning position of the patient, ignores the influence of individual gene risk factors and pathological characteristics, and thirdly, the parameter optimization process is lack of theoretical guidance and mainly depends on empirical adjustment, so that the optimization effect is limited and unstable. With the development of information geometry and artificial intelligence technology, a new technical approach is provided for CT dose optimization. The information geometry maps probability distribution to Riemann manifold, can accurately represent the internal geometry of parameter space, and artificial intelligence technology can learn optimization strategy from mass data. However, there is currently no CT dose optimization system that organically combines information geometry with artificial intelligence techniques, and the advantages of both techniques cannot be fully exploited. Disclosure of Invention The invention aims to provide a CT dose self-adaptive optimization system and a CT dose self-adaptive optimization method based on information geometry and artificial intelligence, which are used for realizing accurate self-adaptive optimization of CT dose parameters and obviously reducing radiation dose on the premise of ensuring image quality by constructing information geometry manifold, calculating Fisher information matrix and applying a geometric guided genetic algorithm. The invention provides a CT dose self-adaptive optimization system, which comprises: The CT scanner is used for carrying out CT scanning according to the scanning parameters and generating CT image data; the image data processing workstation is connected with the CT scanner and is used for processing the CT image data; The high-risk group analysis device is connected with the image data processing workstation and is used for receiving the image data of the patient to be diagnosed, which is sent by the image data processing workstation, analyzing the high-risk group based on the image data of the patient to be diagnosed, generating the high-risk group layering number, and sending the high-risk group layering number to the image data processing workstation; the dose parameter self-adaptive optimization device is connected with the high-risk group analysis device and the image data processing workstation and is used for receiving the high-risk group layering, receiving pre-scanning image data, constructing an information geometric manifold based on the high-risk group layering and the pre-scanning image data, calculating optimal dose parameters on the information geometric manifold, and outputting the optimal dose parameters, wherein the information geometric manifold comprises a dose parameter manifold and an image quality manifold; The automatic dosage parameter adjusting device is connected with the dosage parameter self-adaptive optimizing device and the CT scanner and is used for receiving the optimal dosage parameter and automatically adjusting the scanning parameter of the CT scanner according to the optimal dosage parameter, and And the intelligent learning unit is connected with the dose parameter self-adaptive optimization device and the image data processing workstation and is used for updating the information geometric manifold according to the historical optimization result. Preferably, the dose parameter adaptive optimization device comprises: the manifold construction module is used for mapping the high-risk group layering number to an n-di