CN-122023597-A - Medical image reconstruction method and system for radiology department

CN122023597ACN 122023597 ACN122023597 ACN 122023597ACN-122023597-A

Abstract

The invention discloses a medical image reconstruction method and a medical image reconstruction system for radiology department, and relates to the field of medical image processing. The method comprises the steps of collecting current CT projection data of a target patient, which is lower than clinical diagnosis dose, and calling a past CT image obtained under the clinical diagnosis dose from a medical imaging system, forming a primary reconstruction image based on the current CT projection data, performing spatial registration on the past CT image and the primary reconstruction image to obtain a priori structure image, dividing a reconstruction area into different areas comprising a stable area and a suspicious change area according to differences and tissue characteristics between the priori structure image and the primary reconstruction image, generating an area weight map for distinguishing the prior constraint intensity of the different areas according to the area division result, and performing spatial self-adaptive regulation on the prior structure constraint and the data fidelity constraint based on the area weight map in the reconstruction process to obtain a medical image reconstruction result.

Inventors

YE WEN
ZHANG JIULONG
ZHAN YI

Assignees

上海市公共卫生临床中心

Dates

Publication Date: 20260512
Application Date: 20260206
Priority Date: 20251222

Claims (10)

1. A medical image reconstruction method for radiology department, comprising the steps of: S1, collecting current CT projection data of a target patient, which is lower than clinical diagnosis dosage, and calling at least one past CT image of the patient, which is obtained by the patient under the clinical diagnosis dosage, from a medical imaging system; S2, performing spatial registration on the past CT image and a primary reconstructed image formed based on the current CT projection data to obtain a priori structure image which contains priori anatomical structure and tissue characteristic information and is aligned with a current examination space; s3, dividing a reconstruction region based on the difference and tissue characteristics of the prior structural image and the primary reconstruction image to obtain a region division result at least comprising a stable region and a suspicious change region; S4, generating a region weight map related to the space position according to the region division result; And S5, performing space self-adaptive regulation and control on the prior structure constraint item and the data fidelity item based on the region weight map in the reconstruction process, adopting higher prior constraint on the stable region and adopting lower prior constraint on the suspicious change region so as to generate a medical image reconstruction result.
2. The method for reconstructing a medical image for radiology as set forth in claim 1, wherein said region segmentation comprises the step of imaging said prior structure With the primary reconstructed image Performing differential operation to obtain a differential graph Wherein: ; Based on a first threshold value Determining a stable region based on a second threshold Determining a suspicious region of change, and the first and second thresholds are based on the differential map And compensating the boundary region by utilizing the anatomical structure identification, so that the final region division result comprises a stable region, a suspicious change region and a boundary region.
3. A medical image reconstruction method for radiology as defined in claim 1, wherein the region weight map is representable as a weight function The value of the method satisfies the following conditions: Wherein, the 。
4. A medical image reconstruction method for radiology department as defined in claim 1 wherein the reconstruction process is based on a solution of an objective function consisting of a data fidelity term and a priori structural constraints term in the form of: Wherein, the The projection operator is represented by a representation of the projection operator, Representing the current CT projection data, Representing an image to be reconstructed which is to be reconstructed, Representing an image of the prior structure, In the form of a regional weight map, To measure the penalty function of the difference between the current reconstructed image and the prior structure image, the penalty function can be selected from A function of, Either a function or a Huber function. .
5. A medical image reconstruction method for radiology according to claim 1, wherein an edge protection term is introduced within the suspicious region of change, the edge protection term being representable as: for suppressing excessive smoothing of the microstructure during reconstruction, the intensity of the edge protection term is adjusted by an edge protection coefficient.
6. A medical image reconstruction method for radiology department according to claim 1 wherein the suspicious region of change is retained when it is spatially continuous and exhibits a growing or morphological evolution trend in the historical images, by filtering based on timing consistency rules of multi-stage past exam recordings, otherwise culled.
7. A medical image reconstruction method for radiology department as defined in claim 1, wherein the quality index including noise level, contrast-to-noise ratio, and structural consistency index is automatically calculated after the reconstruction is completed, and a quality control report is generated.
8. A medical image reconstruction method for radiology according to claim 7, wherein triggering reconstruction mode adjustment when at least one quality indicator in the quality control report is below a preset threshold comprises adjusting parameters of the region weight values or edge protection coefficients, or increasing iteration times, and replacing the corrected reconstructed image with the original reconstructed image as both final output or parallel output for comparison.
9. A medical image reconstruction system for radiology department, comprising: the historical image calling module is used for calling a past CT image obtained by scanning under clinical diagnosis dosage from the medical image system; the registration module is used for carrying out spatial registration on the past CT image and the primary reconstructed image to obtain an priori structure image; the region dividing module is used for generating a stable region and a suspicious change region according to the difference between the prior structural image and the primary reconstruction image; The regional weight generation module is used for generating a regional weight map according to the regional division result; the reconstruction module is used for carrying out space self-adaptive regulation and control on the prior structure constraint item and the data fidelity item based on the regional weight map in the reconstruction process so as to obtain a reconstructed image; The quality control module is used for calculating quality indexes including noise level, generating a quality control report and triggering reconstruction mode adjustment when at least one quality index is lower than a preset threshold value; And the display module is used for displaying the reconstructed image and the quality control report.
10. A computer-readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the method steps of any of claims 1 to 8.

Description

Medical image reconstruction method and system for radiology department Technical Field The invention relates to the field of medical image processing, in particular to a medical image reconstruction method and system for radiology department. Background Medical imaging is one of the core means of radiology diagnosis, wherein Computed Tomography (CT) is widely applied to disease screening, efficacy evaluation and long-term follow-up by virtue of the advantages of high imaging speed, clear display of anatomical structures and the like. However, CT imaging relies on X-ray radiation, with the increasing number of examinations, with the increasing cumulative radiation dose of patients, especially in tumor monitoring, chronic disease management and patients requiring long term periodic review, reducing radiation exposure has become an important goal of medical institutions and imaging equipment manufacturers. Under the condition of reducing the radiation dose, the influence of the reduction of the exposure and the distribution change of the signal to noise ratio often causes noise rise, artifact enhancement and detail loss of a reconstructed image, so that key clinical signs such as focus edges, tissue textures and the like can be shielded, and the judgment of a radiologist on the change trend and the disease progress is influenced. In order to improve the quality of low-dose CT images, the prior researches generally adopt a mode of noise reduction algorithm, total variation regularization method, deep learning reconstruction network and the like based on a projection domain or an image domain to inhibit noise and artifacts. However, most of these techniques are based on globally consistent constraint models, and may generate excessive smoothing effects on local areas while enhancing overall image structures and textures, especially in clinically sensitive areas such as early lesion proliferation, volume changes or slight changes in tissue density, which easily cause attenuation of the details of the changes and even erroneous erasure. In addition, existing methods in follow-up images are generally only based on current low dose data for reconstruction, and do not fully utilize the stable anatomical information and temporal change patterns contained in the patient's past images, resulting in difficulty in ensuring a balance between detail sensitivity and global noise suppression. For this purpose, a medical image reconstruction method and system for radiology department is proposed. Disclosure of Invention The invention mainly aims to provide a medical image reconstruction method and system for radiology department, which can effectively solve the problems in the background technology. In order to achieve the above purpose, the invention adopts the technical proposal that, A medical image reconstruction method for radiology department, comprising the steps of: S1, collecting current CT projection data of a target patient, which is lower than clinical diagnosis dosage, and calling at least one past CT image of the patient, which is obtained by the patient under the clinical diagnosis dosage, from a medical imaging system; S2, performing spatial registration on the past CT image and a primary reconstructed image formed based on the current CT projection data to obtain a priori structure image which contains priori anatomical structure and tissue characteristic information and is aligned with a current examination space; s3, dividing a reconstruction region based on the difference and tissue characteristics of the prior structural image and the primary reconstruction image to obtain a region division result at least comprising a stable region and a suspicious change region; S4, generating a region weight map related to the space position according to the region division result; And S5, performing space self-adaptive regulation and control on the prior structure constraint item and the data fidelity item based on the region weight map in the reconstruction process, adopting higher prior constraint on the stable region and adopting lower prior constraint on the suspicious change region so as to generate a medical image reconstruction result. Further, the region division includes, for the prior structure imageWith the primary reconstructed imagePerforming differential operation to obtain a differential graphWherein: ; Based on a first threshold value Determining a stable region based on a second thresholdDetermining a suspicious region of change, and the first and second thresholds are based on the differential mapAnd compensating the boundary region by utilizing the anatomical structure identification, so that the final region division result comprises a stable region, a suspicious change region and a boundary region. Further, the region weight map may be expressed as a weight functionThe value of the method satisfies the following conditions: Wherein, the 。 Further, the reconstruction process is based on solving an