Medical image reconstruction method and system

Abstract

The present invention relates to a medical image reconstruction method, comprising the following steps: S1, acquiring CT image data of a patient with a neurodegenerative disease, and preprocessing the image data; and S2, adding a Control-Net branch on the basis of a Stable Diffusion model, and using Stable Diffusion and Control-Net to perform image reconstruction on the preprocessed image data so as to generate a high-quality PET image. The present invention also relates to a medical image reconstruction system. The present invention ensures that specific prior information is added during image reconstruction, thereby improving the precision and reliability of image reconstruction.

Description

Medical Image Reconstruction Methods and Systems Technical Field

[0001] This invention relates to a method and system for medical image reconstruction. Background Technology

[0002] Parkinson's disease (PD) is the world's fastest-growing neurological disorder driven by an aging population. Currently, there are no blood or laboratory tests available for non-hereditary cases of PD; clinical diagnosis relies primarily on patient history and neurological examination. This disease has become an increasingly serious public health challenge, especially in countries with limited healthcare resources where patients struggle to access timely care and medication. As the disease progresses, imaging changes are often difficult to detect in the early stages; therefore, the development of early quantitative diagnostic methods is crucial for diagnosis and treatment.

[0003] Existing medical imaging technologies such as CT, MRI, and PET provide information on anatomical structures, soft tissue details, and metabolic activity, respectively. However, a single imaging modality is insufficient to provide complete disease information; therefore, multimodal image fusion, especially the combination of CT and PET images, has become an important research direction. In medical image processing, traditional image reconstruction methods typically rely on manual analysis, which is subject to subjectivity and error. With the rise of computer-aided diagnostics (CAD) technology and deep learning, quantitative analysis can provide more accurate image reconstruction and disease diagnosis.

[0004] Currently, deep learning-based image reconstruction and transformation technologies have been applied in the medical field. For example, image-to-image translation methods are used to convert CT images into PET images, combining anatomical information with metabolic activity. However, these technologies mostly rely on traditional convolutional neural networks (CNNs), which still face problems of insufficient accuracy and low computational efficiency when dealing with complex scenes and details.

[0005] In summary, existing technologies have at least the following drawbacks:

[0006] First, there is a lack of accuracy in early diagnosis: existing imaging technologies are insufficient to accurately detect early lesions of Parkinson's disease, especially subtle structural changes.

[0007] Secondly, image reconstruction is inefficient: traditional image reconstruction methods rely on manual intervention, which is inefficient and easily affected by subjective factors.

[0008] Third, insufficient integration of technologies: Although multimodal image fusion has significant advantages in theory, there are still many challenges in effectively fusing CT and PET images in practical applications. Summary of the Invention

[0009] In view of this, it is necessary to provide a method and system for medical image reconstruction.

[0010] This invention provides a medical image reconstruction method, which includes the following steps: S1, acquiring CT image data of patients with neurodegenerative diseases and performing preprocessing; S2, adding a Control-Net branch to the Stable Diffusion model, and using Stable Diffusion and Control-Net to reconstruct images from the preprocessed image data to generate high-quality PET images.

[0011] Preferably, the method further includes:

[0012] Step S3: Quantitative analysis is performed on the generated PET images to extract features of the lesion area.

[0013] Preferably, the method further includes:

[0014] Step S4: Use SPM12 to extract values ​​for each brain region to evaluate its performance in terms of diagnostic accuracy and reliability.

[0015] Preferably, step S1 includes:

[0016] CT image data of Parkinson's disease patients were collected, and image registration and standardization were performed to ensure that the data met the input requirements of the Stable Diffusion model.

[0017] Preferably, step S2 includes:

[0018] A Control-Net branch was added to the Stable Diffusion model;

[0019] All noise, prompts, and additional control images are input into Control-Net, processed, and then output to the Stable Diffusion model.

[0020] The Stable Diffusion model combines prompts to generate PET images controlled by CT.

[0021] Image reconstruction using Stable Diffusion and Control-Net.

[0022] Preferably, the Control-Net is an innovative neural network architecture designed for diffusion models, which achieves precise control of the generation process by introducing additional conditional inputs; the network structure is divided into two parts: a trainable part and a non-trainable part.

[0023] Preferably, step S3 includes:

[0024] Quantitative analysis was performed on the reconstructed PET images to extract features of the lesion areas; the features of the lesion areas included changes in brain regions and metabolic activity.

[0025] This invention provides a medical image reconstruction system, which includes a preprocessing module and an image reconstruction module, wherein:

[0026] The preprocessing module is used to acquire CT image data from patients with neurodegenerative diseases and perform preprocessing.

[0027] The image reconstruction module adds a Control-Net branch to the Stable Diffusion model, and uses Stable Diffusion and Control-Net to reconstruct images from preprocessed image data to generate high-quality PET images.

[0028] Preferably, the system further includes a quantitative analysis module:

[0029] The quantitative analysis module is used to perform quantitative analysis on the generated PET images to extract features of the lesion area.

[0030] Preferably, the system further includes an evaluation module:

[0031] The evaluation module is used to extract values ​​from each brain region using SPM12 to evaluate its performance in terms of diagnostic accuracy and reliability.

[0032] This invention utilizes deep learning technology to reconstruct and fuse CT and PET images of Parkinson's disease patients, achieving efficient and accurate detection of lesion areas. A Stable Diffusion model is used to generate high-quality medical images, especially for PET image reconstruction. Control-Net is used to finely control the generation process, ensuring that the generated images are anatomically consistent with the original CT images. Control-Net provides conditional inputs during this process to incorporate specific prior information during image reconstruction, improving the accuracy and reliability of the reconstruction. The deep learning model is combined for image feature extraction and quantitative analysis, providing accurate support for the early diagnosis of Parkinson's disease. This invention has the following advantages:

[0033] Firstly, efficient and accurate image reconstruction: The combination of Stable Diffusion and Control-Net can generate high-quality, low-noise images, improving the accuracy and detail of images.

[0034] Secondly, the accuracy of early diagnosis is improved: combining deep learning technology for quantitative analysis can identify lesion areas earlier and more accurately, especially in the early stages of Parkinson's disease.

[0035] Thirdly, automated processing: The technical solution of this invention greatly reduces manual intervention and improves diagnostic efficiency. Attached Figure Description

[0036] Figure 1 is a flowchart of the medical image reconstruction method of the present invention;

[0037] Figure 2 is a schematic diagram of the overall model provided in an embodiment of the present invention;

[0038] Figure 3 is a schematic diagram of the effect of reconstructing a high-quality PET image provided by an embodiment of the present invention;

[0039] Figure 4 is a hardware architecture diagram of the medical image reconstruction system of the present invention. Detailed Implementation

[0040] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0041] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0042] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0043] Example 1

[0044] Referring to Figure 1, it is a flowchart of a preferred embodiment of the medical image reconstruction method of the present invention.

[0045] This embodiment uses Parkinson's disease as an example for illustration, but it can also be extended to other types of neurodegenerative diseases, such as Alzheimer's disease and amyotrophic lateral sclerosis.

[0046] Please refer to Figure 2, step S1, to acquire CT image data of Parkinson's disease patients and perform preprocessing. Specifically:

[0047] CT image data of Parkinson's disease patients were collected, and image registration and standardization were performed to ensure that the data met the input requirements of the Stable Diffusion model.

[0048] Step S2 involves adding a Control-Net branch to the Stable Diffusion model. This branch is used to reconstruct the preprocessed image data using both Stable Diffusion and Control-Net to generate high-quality PET images (see Figure 3). Specifically:

[0049] This embodiment adds a Control-Net branch to the Stable Diffusion (SD) model. All noise, cue words, and additional control images are input into Control-Net, processed, and then output to the SD master control model. Finally, the SD master control model combines the cue words to generate a PET image controlled by CT.

[0050] Image reconstruction using Stable Diffusion and Control-Net: High-quality PET images are generated using the Stable Diffusion model, while Control-Net ensures structural consistency between the generated images and the original CT images. This process introduces prior conditions, ensuring that the generated images not only match the original CT in functional information but also possess accurate anatomical structures.

[0051] Control-Net is an innovative neural network architecture designed for diffusion models, achieving precise control over the generation process by introducing additional conditional inputs. Its core feature lies in dividing the network structure into two parts: ① a trainable part, used to learn deep representations of the control variables; and ② a locked part, preserving the original characteristics of the stable diffusion model. This design ensures that the model maintains the generative capability and stability of the original diffusion model while introducing conditional constraints, thereby effectively improving generation results in complex scenes.

[0052] Step S3: Quantitative analysis is performed on the generated PET images to extract features of the lesion area.

[0053] Specifically:

[0054] Quantitative analysis was performed on the reconstructed PET images to extract features of the lesion areas, and deep learning algorithms were used for early diagnosis. The features of the lesion areas included changes in brain regions and metabolic activity.

[0055] A Control-Net-based Stable Diffusion image reconstruction model was developed: the application of Control-Net in the Stable Diffusion framework was optimized for reconstructing medical images of Parkinson's disease, and feature extraction was enhanced, especially the early changes in the lesion area.

[0056] Step S4 involves using SPM12 to extract numerical values ​​from each brain region to evaluate its performance in terms of diagnostic accuracy and reliability. Specifically:

[0057] High-quality PET images generated by Stable Diffusion were used in conjunction with deep learning technology to construct an early diagnostic model for Parkinson's disease. SPM12 was used to extract numerical values ​​from various brain regions to evaluate its performance in terms of diagnostic accuracy and reliability.

[0058] Further, referring to Figure 2, this embodiment adds a Control-Net branch to the Stable Diffusion (SD) model. All noise, cue words, and additional control images are input into Control-Net, processed, and output to the SD master control model. Finally, the SD master control model combines the cue words to generate a PET image controlled by CT, as shown in Figure 3. The two branches are independent of each other, eliminating the need to retrain a completely new model, while Control-Net can also flexibly reference and process control images.

[0059] Example 2

[0060] This embodiment uses Parkinson's disease as an example for illustration, but it can also be extended to other types of neurodegenerative diseases, such as Alzheimer's disease and amyotrophic lateral sclerosis.

[0061] Referring to Figure 4, this is a hardware architecture diagram of the medical image reconstruction system 10 of the present invention. The system includes: a preprocessing module 101, an image reconstruction module 102, a quantitative analysis module 103, and an evaluation module 104.

[0062] in:

[0063] The preprocessing module 101 is used to acquire CT image data from Parkinson's disease patients and perform preprocessing. Specifically:

[0064] The preprocessing module 101 acquires CT image data of Parkinson's disease patients and performs image registration and standardization processing to ensure that the data meets the input requirements of the Stable Diffusion model.

[0065] The image reconstruction module 102 adds a Control-Net branch to the Stable Diffusion model, and uses Stable Diffusion and Control-Net to reconstruct images from preprocessed image data to generate high-quality PET images, as shown in Figure 3. Specifically:

[0066] This embodiment adds a Control-Net branch to the Stable Diffusion (SD) model. All noise, cue words, and additional control images are input into Control-Net, processed, and then output to the SD master control model. Finally, the SD master control model combines the cue words to generate a PET image controlled by CT.

[0067] Image reconstruction using Stable Diffusion and Control-Net: High-quality PET images are generated using the Stable Diffusion model, while Control-Net ensures structural consistency between the generated images and the original CT images. This process introduces prior conditions, ensuring that the generated images not only match the original CT in functional information but also possess accurate anatomical structures.

[0068] Control-Net is an innovative neural network architecture designed for diffusion models, achieving precise control over the generation process by introducing additional conditional inputs. Its core feature lies in dividing the network structure into two parts: ① a trainable part, used to learn deep representations of the control variables; and ② a locked part, preserving the original characteristics of the stable diffusion model. This design ensures that the model maintains the generative capability and stability of the original diffusion model while introducing conditional constraints, thereby effectively improving generation results in complex scenes.

[0069] The quantitative analysis module 103 is used to perform quantitative analysis on the generated PET images to extract features of the lesion area. Specifically:

[0070] The quantitative analysis module 103 performs quantitative analysis on the reconstructed PET images, extracts features of the lesion area, and uses deep learning algorithms for early diagnosis. The features of the lesion area include changes in brain regions and metabolic activity.

[0071] The quantitative analysis module 103 performs a Control-Net-based Stable Diffusion image reconstruction model: optimizes the application of Control-Net in the Stable Diffusion framework, reconstructs medical images of Parkinson's disease, and enhances feature extraction, especially early changes in the lesion area.

[0072] The evaluation module 104 is used to extract numerical values ​​from various brain regions using SPM12 to evaluate their performance in terms of diagnostic accuracy and reliability. Specifically:

[0073] High-quality PET images generated by Stable Diffusion were used in conjunction with deep learning technology to construct an early diagnostic model for Parkinson's disease. SPM12 was used to extract numerical values ​​from various brain regions to evaluate its performance in terms of diagnostic accuracy and reliability.

[0074] Further, referring to Figure 2, this embodiment adds a Control-Net branch to the Stable Diffusion (SD) model. All noise, cue words, and additional control images are input into Control-Net, processed, and output to the SD master control model. Finally, the SD master control model combines the cue words to generate a PET image controlled by CT, as shown in Figure 3. The two branches are independent of each other, eliminating the need to retrain a completely new model, while Control-Net can also flexibly reference and process control images.

[0075] In addition to the early diagnosis of neurological diseases, this invention can also be applied to other medical imaging fields that require fine image reconstruction and quantitative analysis, such as tumor detection and cardiovascular disease diagnosis.

[0076] This invention combines the Stable Diffusion model with the Control-Net deep learning framework, offering new possibilities for medical image processing. By utilizing the Stable Diffusion model fused with Control-Net, it improves the reconstruction results of Parkinson's disease medical images, accurately presenting early lesion characteristics. Particularly in the early quantitative diagnosis of the disease, it can significantly improve diagnostic accuracy and efficiency.

[0077] Although the present invention has been described with reference to the present preferred embodiments, those skilled in the art should understand that the above preferred embodiments are only used to illustrate the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A medical image reconstruction method, characterized by, The method includes the following steps: S1: Acquire CT image data of patients with neurodegenerative diseases and perform preprocessing; S2 adds a Control-Net branch to the Stable Diffusion model, which uses Stable Diffusion and Control-Net to reconstruct images from preprocessed image data to generate high-quality PET images.

2. The method of claim 1, wherein, The method further includes: Step S3: Quantitative analysis is performed on the generated PET images to extract features of the lesion area.

3. The method of claim 2, wherein, The method further includes: Step S4: Use SPM12 to extract values ​​for each brain region to evaluate its performance in terms of diagnostic accuracy and reliability.

4. The method of claim 3, wherein: Step S1 includes: CT image data of Parkinson's disease patients were collected, and image registration and standardization were performed to ensure that the data met the input requirements of the Stable Diffusion model.

5. The method of claim 4, wherein, Step S2 includes: A Control-Net branch has been added to the Stable Diffusion model; All noise, prompts, and additional control images are input into Control-Net, processed, and then output to the Stable Diffusion model. The Stable Diffusion model combines prompts to generate PET images controlled by CT. Image reconstruction using Stable Diffusion and Control-Net.

6. The method of claim 5, wherein, The Control-Net described above is an innovative neural network architecture designed for diffusion models, which achieves precise control of the generation process by introducing additional conditional inputs; the network structure is divided into two parts: a trainable part and a non-trainable part.

7. The method of claim 6, wherein, Step S3 includes: Quantitative analysis was performed on the reconstructed PET images to extract features of the lesion areas; the features of the lesion areas included changes in brain regions and metabolic activity.

8. A medical image reconstruction system, characterized by The system includes a preprocessing module and an image reconstruction module, wherein: The preprocessing module is used to acquire CT image data from patients with neurodegenerative diseases and perform preprocessing. The image reconstruction module adds a Control-Net branch to the Stable Diffusion model, and uses Stable Diffusion and Control-Net to reconstruct images from preprocessed image data to generate high-quality PET images.

9. The system of claim 8, wherein, The system also includes a quantitative analysis module: The quantitative analysis module is used to perform quantitative analysis on the generated PET images to extract features of the lesion area.

10. The system of claim 9, wherein, The system also includes an evaluation module: The evaluation module is used to extract values ​​from each brain region using SPM12 to evaluate its performance in terms of diagnostic accuracy and reliability.

Images