Synchronous Triplanar Reconstruction Integrated with Color Doppler Mapping for Precise and Rapid Localization of Thyroid Lesions

サマリー

Here we present a 5D ultrasound technique combining multi-planar 3D reconstruction and color Doppler fusion, which enables synchronous visualization of thyroid structural and functional information. By minimizing blind spots, this method allows rapid, precise localization of lesions to improve diagnostic accuracy, especially benefiting novice practitioners.

要約

This paper proposes a novel thyroid examination technique based on five-dimensional (5D) synchronous reconstruction of ultrasound data. The raw temporal sequences are reconstructed into 3D volumetric data reflecting anatomical structure. Triplanar visualization from three orthogonal planes is realized to provide a systematic inspection of the entire gland. Color Doppler imaging is integrated into each triplanar slice to map vascularity changes. This multi-modal fusion enables synchronous display of structural, functional, and blood flow information in the reconstructed 5D space. Compared to conventional scanning, this technique offers the benefits of flexible offline diagnosis, reduced dependency on scanning, enhanced intuitive interpretation, and comprehensive multi-aspect evaluation. By minimizing oversight errors, it could improve diagnostic accuracy, especially for novice practitioners. The proposed 5D fusion method allows rapid and precise localization of lesions for early detection. Future work will explore integration with biochemical markers to further improve diagnostic precision. The technique has considerable clinical value for advancing thyroid examination.

概要

Hashimoto's thyroiditis (HT), the most frequent autoimmune thyroid disorder (AITD), is the leading cause of hypothyroidism in iodine-sufficient areas of the world¹. It is characterized by lymphocytic infiltration and autoantibodies against thyroid antigens, leading to the destruction of thyroid architecture and hypothyroidism². Staging of HT aims to assess the severity and guide treatment decisions. It relies on a combination of biochemical markers such as thyroid stimulating hormone (TSH) and thyroid autoantibodies³, as well as ultrasonographic features visible on thyroid ultrasound^4,5,6.

On ultrasound examination, HT demonstrates characteristic findings, including diffusely decreased echogenicity, heterogeneous echotexture, micronodularity, and increased blood flow on color Doppler^6,7. However, conventional two-dimensional (2D) grayscale ultrasound lacks quantitative methods for systematically analyzing these features for HT staging⁸. The assessment of vascularity changes is also limited to qualitative visual inspection in 2D mode. The complex three-dimensional (3D) architecture of the thyroid gland further hampers thorough evaluation using conventional 2D slicing^9,10. These factors lead to imaging blind spots and misinterpretation, resulting in low sensitivity and specificity, especially for less experienced practitioners^11,12.

Conventional handheld ultrasound scanning integrates real-time acquisition and diagnosis. This coupled workflow reliance increases the likelihood of oversight errors during scanning. The lack of spatial localization and tracking also makes lesion identification and monitoring imprecise^12,13. Dedicated 3D ultrasound systems have emerged to address these limitations and have shown promising results^14,15. However, most 3D ultrasound technologies require complex mechanical scanning mechanisms and specialized transducers, leading to high costs and barriers to adoption.

To overcome the limitations of conventional 2D and 3D ultrasound techniques, this study proposes a novel 3D reconstruction and visualization solution tailored for thyroid examination. Using widely available handheld ultrasound, multiple 2D sweeps are first acquired to scan the entire thyroid gland. 3D volumetric reconstruction is then realized by spatial registration and fusion of the 2D sequences. Concurrently, color Doppler frames are coregistered to create vascularity maps visualizing blood flow changes. The reconstructed 3D grayscale volumes and colored vascularity maps are finally integrated into a single platform, enabling synchronized multi-planar visualization and combined structural-functional inspection.

This proposed 3D fusion technique provides a systematic and comprehensive evaluation of the complex thyroid morphology from different aspects. By minimizing blind spots and enabling global overview, it could help improve diagnostic accuracy and reduce oversight errors, especially benefiting novice practitioners. The multi-modal visualization also facilitates rapid and precise localization of lesions, holding promise for early diagnosis and treatment of thyroid nodules and tumors. Moreover, the method introduces quantitative 3D feature analysis which has not been investigated for HT staging before. With wide adoption, it has the potential to standardize and objectify the currently experience-dependent ultrasound diagnosis procedures. By synergistically integrating handheld 3D reconstruction, multi-modal fusion, quantitative feature analysis, and flexible visualization into a streamlined workflow, this low-cost, easy-to-use technique represents a diagnostically powerful leap from conventional 2D ultrasound for advancing thyroid examination.

プロトコル

This study was approved by the Institutional Review Board of Sunsimiao Hospital affiliated with the Beijing University of Chinese Medicine. The patient was recruited from the Department of Thyroid, Sunsimiao Hospital. The patient underwent a thyroid ultrasound examination and gave informed consent for the study. In this investigation, 4D ultrasound data acquired using a handheld device were utilized to reconstruct triplanar views of the thyroid gland. Furthermore, real-time synchronous color Doppler imaging was achieved. The software tools used in this research are listed in the Table of Materials.

1. Data collection and preparation

1. Using a portable handheld ultrasound device, place the linear array transducer transversely on the patient's neck to image the thyroid in the cross-sectional plane. Slide the probe slowly and steadily along the length of the thyroid while maintaining probe contact and orientation.
2. Acquire a sequence of transverse B-mode images visualizing thyroid morphology at a frame rate of 33 Hz.
3. Concurrently, apply color Doppler to detect blood flow in the gland and vessels. Scan from the superior to inferior thyroid poles to cover the entire gland. The resulting dynamic imaging sequence comprises consecutive transverse slices that form two 4D data sets.
4. Loading and browsing of 4D B-mode ultrasound data
   1. Copy all DICOM data to a customized working directory.
      NOTE: The working directory is the same in both the operating system and MATLAB. Press Enter after typing each line to run the command in MATLAB.
   2. Import the B-mode US data file into MATLAB using the dicomread function, and use the size function to view the dimensions of the data.
      1. Open MATLAB on the computer.
      2. In the Command window, type:
         VB0 = dicomread('fname.dcm');
         Where 'fname.dcm' could be replaced with the actual filename of the DICOM data. This will read in the DICOM file and store the image data in the variable VB0.
      3. To view the size of the loaded data, type:
         size(VB0),
         NOTE: The 4D data imported here had dimensions of 768 pixels x 1024 pixels x 3 x 601 layers. The 768 pixels x 1024 pixels x 3 corresponds to a standard RGB image, where each pixel is represented by three channels with 24-bit depth. The 601 layers indicate the total number of scanned slices.
   3. Call the US_B_Show function to convert the 4D matrix data into a continuous grayscale video sequence to be played back continuously for detailed examination (see Figure 1).
      1. To convert this 4D ultrasound data matrix VB0 imported in Step 1.4.2.2 using the dicomread function on the DICOM files into a continuously playing grayscale video sequence, call the US_B_Show function by typing the following command in the MATLAB Command window:
         US_B_Show(VB0)
         Where VB0 is the 4D matrix variable containing the ultrasound data imported previously.
   4. The GUI in Figure 1 shows playback buttons for pause, forward, rewind, etc.
      1. Press the play button to initiate continuous video playback of the frame sequence. Use the pause and play control tool icons for flexible navigation of any frame. Use the zoom in/out buttons to dynamically magnify or minify the images during playback and the default zoom button to reset to the original 1x view.
      2. Click on the inspect pixel values button and move the mouse over a region to overlay crosshairs with pixel coordinates and intensities for localized analysis.
         NOTE: These interactive controls enable flexible inspection of ultrasound data characteristics across both space and time.
5. Loading and browsing of 4D color Doppler ultrasound data
   1. Import the color Doppler ultrasound data file into MATLAB using the dicomread function, and use the size function to view the dimensions of the data.
      NOTE: The 4D data imported here had dimensions of 768 pixels x 1024 pixels x 3 x 331 layers. The 768 pixels x 1024 pixels x 3 corresponds to a standard RGB image, with red and blue representing blood flow in different directions. The 331 layers indicate the total number of scanned slices.
   2. Use the US_C_Show function to convert the 4D matrix data into a continuous color video sequence to be played back continuously for detailed examination (see Figure 2).
      NOTE The GUI in Figure 2 has the same set of interactive controls and operations as described previously in step 1.4.4 for Figure 1.

2. Synchronous observation of B-mode and color Doppler ultrasound

NOTE: The 4D B-Mode Ultrasound Data shown in Figure 1 and the 4D Color Doppler Ultrasound Data shown in Figure 2 contain the same absolute time stamps in the fourth dimension along the temporal axis. This field is recorded in the DICOM metadata as FrameTimeVector. Based on the time values in this field, Figure 1 and Figure 2 can be synchronized in real time.

1. After reading the two 4D files using the dicomread command, execute the Synchronize_B_C function with the two 4D matrices as inputs.
   NOTE: Figure 3 shows the resulting video that can still be played back continuously. The difference now is that the 4D B-Mode Ultrasound Data and the 4D Color Doppler Ultrasound Data are synchronized in real time within the same video frames. The GUI in Figure 3 has the same set of interactive controls and operations as described previously in step 1.4.4 for Figure 1.

3. Synchronous triplanar reconstruction for thyroid

NOTE: To enable more precise localization and quantification of lesions, this study performed triplanar reconstruction of the thyroid gland from the acquired 4D ultrasound data, with real-time interactivity. This allows clinicians to rapidly and accurately pinpoint lesions, laying a solid foundation for subsequent quantification of the affected regions.

1. Call the thyroid_triplanar function with the 4D B-mode ultrasound data from Figure 1 as input to derive the three orthogonal planes (coronal, sagittal, and axial) as shown in Figure 4.
2. The crosshair interaction in Figure 4 enables real-time inspection of different parts of the thyroid gland. Click and drag the center of the crosshair for an arbitrary 3D examination of the thyroid anatomy reconstructed from ultrasound.
   NOTE: The GUI in Figure 4 also enables adjustment of the grayscale intensity range, contrast, and brightness of the triplanar views.
3. Press and drag the left mouse button over any region of the images for real-time modification of brightness and contrast levels. Release the mouse button to confirm and finalize the adjustments.

4. Synchronous triplanar reconstruction for 3D blood flow field

NOTE: Reconstructing the synchronous triplanar views for the 3D blood flow field based on 4D color Doppler ultrasound data is also clinically important for characterizing Hashimoto's Thyroiditis (HT).

1. Call the thyroid_3D_blood function with the 4D C-mode ultrasound data from Figure 2 as input to derive the three orthogonal planes (coronal, sagittal, and axial) as shown in Figure 5.
2. The crosshair interaction in Figure 5 enables real-time inspection of different parts of the thyroid gland. Click and drag the center of the crosshair for an arbitrary 3D examination of the thyroid anatomy reconstructed from ultrasound.
   NOTE: The GUI in Figure 5 also enables adjustment of the grayscale intensity range, contrast, and brightness of the triplanar views.
3. Press and drag the left mouse button over any region of the images for real-time modification of brightness and contrast levels. Release the mouse button to confirm and finalize the adjustments.

5. Synchronization of B-mode triplanar views and color Doppler triplanar views

NOTE: Building upon the triplanar views shown in Figure 4, synchronizing the corresponding color Doppler flow images to the lesion locations would undoubtedly facilitate the diagnosis and quantification of the pathological progression in Hashimoto's Thyroiditis (HT).

1. Drag the crosshair interaction in Figure 4 to locate the region of interest, and execute US_B2C to obtain the corresponding location in the color Doppler triplanar views.
2. Drag the crosshair interaction in Figure 5 to locate the region of interest, and execute US_C2B to obtain the corresponding location in the B-mode triplanar views.
   NOTE: Figure 6 lays a solid ultrasonographic foundation for the precise localization and definitive diagnosis of Hashimoto's thyroiditis (HT) lesions.

結果

As shown in the graphical user interface (GUI) in Figure 1 and Figure 2, the ultrasound scanning sequence can be checked continuously. However, this two-dimensional examination relies heavily on the thyroidologist's anatomical knowledge to mentally reconstruct the lesion's location, which is challenging for novices and results in a lack of quantitative consistency. Figure 3 fuses the B-mode grayscale with color Doppler flow images to enable more informed inspection on the 2D scans.

To facilitate a comprehensive and reliable assessment of thyroid disorders, Figure 4 shows the 3D spatial distribution of B-mode intensities, while Figure 5 exhibits the 3D vascularity map reconstructed from color Doppler data. As illustrated in the GUI (Figure 6), synchronized visualization of the structural and functional information along three orthogonal planes is realized. Clinicians can continuously inspect the multiplanar cross-sections of thyroid grayscale images and the corresponding blood flow images. This seamless integration of complementary modalities could play a pivotal role in precisely localizing and determining the severity of thyroid pathologies.

If the 3D grayscale volumes and 3D color Doppler volumes are essentially 4D data spanning the spatial and pathological dimensions, linking their interaction across the two synchronized triplanar visualizations could empower the thyroidologist to swiftly pinpoint lesions in a unified 5D space and deliver accurate diagnoses based on the joint intensity and flow patterns.

Figure 1: Transverse B-mode ultrasound video loops. Consecutive B-mode frames acquired by continuous transverse scanning show thyroid morphology. Please click here to view a larger version of this figure.

Figure 2: Transverse color Doppler ultrasound video loops. Consecutive color Doppler frames obtained by transverse scanning reveal blood flow characteristics of the thyroid tissue. Please click here to view a larger version of this figure.

Figure 3: Synchronized B-mode and Doppler ultrasound. Integrated video synchronously exhibiting thyroid structure (grayscale) and blood flow (colored overlay). The color Doppler overlay depicts the direction and velocity of flow using a color scale-red indicates flow toward the transducer; blue indicates flow away from the transducer. Please click here to view a larger version of this figure.

Figure 4: Triplanar views extracted from B-mode ultrasound. Orthogonal coronal, sagittal, and axial planes reconstructed from 4D B-mode scans using triplanar visualization. Please click here to view a larger version of this figure.

Figure 5: Triplanar views extracted from Doppler ultrasound. Orthogonal coronal, sagittal, and axial planes reconstructed from 4D Doppler scans to map blood flow characteristics of the thyroid tissue. Please click here to view a larger version of this figure.

Figure 6: Synchronized triplanar views fusing structural and vascularity data. Fused multiplanar reconstruction synchronizing B-mode and Doppler data to enable combined morphological and functional inspection. Please click here to view a larger version of this figure.

ディスカッション

Critical steps in the protocol
While Figure 1 and Figure 2 have value for inspection and diagnosis, determining lesion location and views from other perspectives requires expert experience. For the diagnosis of Hashimoto's thyroiditis (HT), synchronizing Figure 1 and Figure 2 in real time is also an important and critical step. Protocol step 3.3 is one of the key steps where, as shown in Figure 4, the attending physician can interactively examine arbitrary cross-sections of the 3D thyroid anatomy. This is crucial for localizing lesions and identifying abnormal tissue regions. Traditionally, handheld ultrasound scanning only provides 2D transverse views. This unavoidably leads to oversight of 3D pathological details due to blind spots. Similarly, protocol step 4.3 generates the 3D blood flow map, which is also critical for pinpointing lesion locations. Protocol steps 5.1 and 5.2 synchronize the structural and functional thyroid images, equipping clinicians with more powerful digital intelligent tools for managing complex conditions.

Modifications and troubleshooting
If reconstruction artifacts occur, the acquisition sweep extent may be insufficient. Repeating scanning with extended coverage can overcome this. Parameters such as slice spacing and pixel size can also be adjusted.

Limitations of the method
Although handheld ultrasound scanning can obtain time stamps for synchronizing various modes, it lacks real-time 3D probe localization. Hence, only the transverse dimensions are precisely reconstructed in the thyroid models. Quantitative measurements on transverse planes are currently accurate, while coronal and sagittal views assist pathological localization but have unreliable quantitative scales at present.

Significance with respect to existing methods
This 5D ultrasound technique enhances conventional 2D scanning by enabling multi-planar structural examination combined with blood flow mapping in a panoramic visualized space. It overcomes limitations like operator dependence, blind spots, and diagnostic ambiguity that persist in 2D ultrasonography. The proposed workflow lays a robust foundation to standardize and transform the current experience-dependent practices for ultrasound diagnosis of thyroid diseases.

Potential applications
This method can be applied to precisely localize and quantify thyroid nodules, tumors, and inflammatory lesions such as Hashimoto's thyroiditis. It provides radiologists and surgeons with enhanced visual perspectives for evaluating pathology. The technique has considerable potential to aid diagnosis, treatment planning, and surgical guidance. Additionally, the study team plans to incorporate biochemical markers with this 5D analysis pipeline to realize AI-empowered precision diagnosis and quantification for thyroid diseases.

材料

この記事で使用された材料の一覧

名前	会社	カタログ番号	コメント
MATLAB	MathWorks	2023B	Computing and visualization
Tools for Thyroid Disease Precision Quantification	Intelligent Entropy	Thyroid-3D V1.0	Beijing Intelligent Entropy Science & Technology Co Ltd. Modeling for Thyroid Disease