Ultrasound-Guided Optimization of Preoperative Chemoradiotherapy Using ¹³¹I SPECT/CT for Enhanced Targeting in Differentiated Thyroid Carcinoma

Abstract

To properly target tumors during preoperative chemoradiotherapy, differentiated thyroid carcinoma (DTC) must be careful. This method increases treatment success and decreases recurrence. Ultrasound coupled with SPECT/CT may provide novel localization and dose planning possibilities. Many systems solely use anatomical or functional imaging. This may result in insufficient dosage delivery and wasted radiation exposure to healthy tissues. These issues are addressed by Dual-Modality Imaging-Guided Adaptive Chemoradiotherapy Planning (DMI-ACP). This innovative approach combines real-time ultrasound imaging with 6 Å SPECT/CT imaging for precise tumor delineation and tailored dosimetry. This system enables clinicians to adjust chemoradiotherapy regimens by seamlessly integrating functional iodine absorption data with anatomical characteristics, thereby targeting therapy to cancerous areas. The outcomes of this method for patients with DTC were promising, including better lesion targeting, reduced radiation exposure to healthy tissues, and improved chemotherapeutic dose distribution. In clinical evaluations, the DMI-ACP framework demonstrated a sensitivity of 94% and a specificity of 89% in identifying malignant lesions compared with traditional imaging techniques. Furthermore, the integration of adaptive planning resulted in a 20% improvement in tumor control probability and a 15% reduction in exposure to surrounding healthy tissue, as assessed through dosimetric analysis. 2023075 Nanjing Drum Tower Hospital Group Suqian Hospital/The Affiliated Suqian Hospital of Xuzhou Medical University.

Keywords

differentiated thyroid carcinoma ultrasound-guided imaging adaptive radiotherapy tumor targeting dual-modality imaging

Introduction

Differentiated thyroid carcinoma (DTC) is the most prevalent type of thyroid cancer and constitutes more than 90% of the cases all over the world.^1,2 Over the last two decades, the frequency of DTC has increased gradually, partly due to advancements in detection methods.^3,4 This is vital in enhancing the efficiency of targeting and sparing normal tissue, as well as reducing the rate of recurrence, since treatment with chemoradiotherapy is often followed by surgery in patients with intermediate- or high-risk conditions.⁵

Precision provides stability in radiation target mapping and dose, and adaptivity offers the possibility of intraprocedural refinements based on feedback.^6,7 There is an opportunity to break the constraint of the conventional method and address the issue using a dual-modality implementation, such as combining real-time ultrasound and functional ¹³¹I single photon emission computed tomography/computed tomography (SPECT/CT) imaging.⁸ Ultrasound-¹³¹I SPECT/CT fusion allows clinicians to have a richer and more precise visual of tumor areas.⁹ As the tumor responds to treatment, it undergoes changes in size, shape, and metabolic activity, which are continually assessed using real-time imaging techniques such as SPECT/CT and ultrasound. The imaging data are promptly integrated into the system, enabling adaptive chemoradiotherapy planning. For example, if the tumor diminishes during an initial treatment cycle, the radiation dosage may be more accurately targeted at the residual malignant tissue, therefore reducing harm to adjacent healthy tissues. If the tumor enlarges or metastasizes, the framework may modify the radiation beams to include the newly affected regions. Moreover, these feedback loops incorporate biological variables, such as iodine absorption and vascularity, enabling adjustments in response to the tumor’s changing metabolic activity. This ongoing adjustment ensures that the therapy remains effective, efficient, and tailored, thereby enhancing therapeutic outcomes while minimizing toxicity to adjacent tissues.

The rest of the article is organized as follows: The second section, Background Study, discusses the synergy between ultrasound and ¹³¹I SPECT/CT imaging. In the third section, Proposed DMI-ACP Framework, the Dual-Modality Imaging-Guided Adaptive Chemoradiotherapy Planning (DMI-ACP) framework is proposed. In the forth section, Result and Discussion, the clinical study and validation process are explained. The fifth section, Conclusion, gives a discussion of key findings, limitations, and implications. The final section concludes with a discussion of future directions and possibilities for clinical integration.

Background Study

According to Cowan et al.,¹⁰ real-time imaging accuracy in preoperative flap planning, vascular selection, and postoperative monitoring has transformed reconstructive microsurgery through the use of ultrasound. Color duplex ultrasonography (CDU) enhances vascular selection and design of flaps, facilitating the complete mapping of perforators and assessment of blood flow, thereby improving recipient vascular selection.¹¹ Ultra-high frequency ultrasound is a significant imaging technology in the field of lymphedema surgery, enhancing lymphatic channel mapping.¹²

Rios-Sanchez et al.¹³ state that contrast-enhanced ultrasound (CEUS) provides real-time images of the vascular system, making it a low-cost, nonirradiating solution for assessing the vascular system. The given article discusses the use of CEUS in imaging the lymphatic veins to plan the left ventricular aneurysm/left ventricular analysis procedures and its history.¹⁴ Nonetheless, additional studies are necessary to homogenize measures and analyze the long-term results, as CEUS has such potential in the treatment of lymphedema.¹⁵

In 2022, it is expected that more than 100,000 individuals will be diagnosed with melanoma in the United States alone by Allen et al.¹⁶ Although limited conventional treatment options were available in the late stages of the disease, immunotherapies and B-RAF inhibitors have significantly improved survival rates.^17,18 Radionuclide melanin targeting in patients with melanoma, using radioimmunotherapy and radiolabeled small molecules, has promising potential to enhance the effectiveness of treatment and can be utilized as an adjunct to current therapies.¹⁹ The combination of ¹³¹I SPECT/CT and ultrasound presents a robust dual-modality imaging technique, delivering complementary advantages that improve the precision and efficacy of preoperative chemoradiotherapy planning in DTC. ¹³¹I SPECT/CT is proficient in recognizing functional alterations in thyroid tissue, offering excellent specificity in the detection of malignant lesions via the radiotracer’s absorption in neoplastic cells. Although SPECT/CT provides exceptional sensitivity for detecting functional abnormalities, its spatial resolution may be limited, which can complicate the accurate delineation of tumor margins, especially for small or subclinical tumors. Conversely, ultrasound provides exceptional real-time, high-resolution imaging of soft tissues, enabling precise visualization of tumor dimensions, morphology, and positioning. Nonetheless, ultrasonography alone is less effective in detecting functional activity or differentiating malignant from benign tumors. Integrating the two modalities, ¹³¹I SPECT/CT offers accurate functional localization, while ultrasound contributes intricate structural details, resulting in a holistic imaging solution.

Proposed DMI-ACP Framework

In a responsive framework, the DMI-ACP architecture emphasizes functional and anatomical imaging for preoperative DTC targeting. DMI-ACP uses real-time imaging and iterative feedback to change treatment options. In contrast, standard models see imaging as a one-time occurrence. SPECT/CT generates volumetric imaging data in DICOM format, which may not align neatly with the real-time, high-resolution 2D images produced by ultrasound, often using proprietary formats. This discrepancy might hinder data fusion, necessitating specialist software for format translation and spatial alignment. Furthermore, the spatial resolutions of the two modalities vary, with SPECT/CT providing inferior resolution relative to ultrasonography, potentially resulting in misalignment of tumor margins during data integration for treatment planning. Moreover, inconsistencies in scanning methods exacerbate the process, as SPECT/CT relies on radiotracer injections and time-sensitive imaging, whereas ultrasonography is a dynamic, patient-specific examination. Finally, geometric distortions such as beamforming errors and motion artifacts in ultrasound images must be corrected to ensure accurate fusion with the SPECT/CT data, highlighting the need for robust calibration techniques.

Figure 1 illustrates a process that is evaluated before the management of thyroid cancer begins, through pre-evaluation, including clinical, biochemical, and multimodal imaging (such as ultrasound, positron emission tomography (PET), SPECT/CT, magnetic resonance imaging (MRI), and advanced simulation) to assist in planning the therapy. These steps are all helpful in detecting tumors, performing metabolic profiling, precision radiotherapy, and measuring the response to personalized therapies, thereby enhancing outcomes. Equation 1 expresses the tumor volume $U_{t u}$ using an ellipsoidal approximation,

U_{t u} = \frac{ρ}{6} . M . X . G + Y [p - u w']

(1)

FIG. 1.

Multimodal imaging and evaluation pathway for thyroid cancer management.

$U_{t u}$ using the ellipsoidal shape of the length $M$ , width $X$ , and height $G$ was determined by ultrasound. This model facilitates reliable $Y$ and consistent volume estimation $[p - u w']$ , particularly for irregular nodules. Planning for $ρ$ chemoradiotherapy and tracking treatment response. Equation 1 calculates the tumor volume and depends on precise volume measurement.

Equation 2 represents iodine uptake contrast $V$ in SPECT/CT imaging,

V = \frac{D_{t} - D_{b}}{D_{b}} * U y [p l - a n]

(2)

Equation 2 calculates the tumor signal’s $V$ degree of differentiation $a n'$ from the surrounding tissues. $D_{b}$ is the iodine uptake in healthy tissue, whereas $D_{t}$ is the average iodine uptake $U y$ in the tumor $p l$ . By improving ratio of radioiodine uptake in fused functional images, this contrast ratio helps direct more focused treatment using Equation 2. The linearity assumption is valid under optimal circumstances when the uptake in both tumor and normal tissues is uniform, and the imaging equipment operates consistently without substantial noise or artifacts. In reality, tumors often exhibit heterogeneous iodine absorption, characterized by elevated uptake in the periphery and decreased uptake in the core, which can lead to deviations from the presumed linear relationship. In addition, elements such as vascularity, hypoxia, and tumor metabolic variability might further complicate this model, making the uptake response less predictable.

Equation 3 defines the effective half-life $U_{eff}$ of the radionuclide,

U_{eff} = (\frac{1}{U_{q}} + \frac{1}{U_{c}})^{- 1} * U [o - qnc']

(3)

Equation 3 determines the effective half-life $U_{eff}$ by combining its biological clearance $U_{c}$ and physical decay $U_{q}$ . When figuring out $o$ long radioactive material stays active in the body $U$ ; this parameter $qnc'$ is crucial. It guarantees precise radiation dosage when therapy is being planned for half-life through effective. The relationship between iodine absorption and tumor aggressiveness in DTC is crucial for understanding tumor dynamics and predicting treatment efficacy. The sodium-iodide symporter (NIS) primarily mediates iodine absorption, facilitating the active transport of iodine into thyroid cells. In well-differentiated DTC, the neoplastic cells retain several attributes of normal thyroid tissue, including elevated NIS expression, resulting in enhanced iodine absorption.

Figure 2 illustrates γ camera imaging, where γ rays emitted by a radiotracer located in the body are collimated by collimators and focused by sandwiched optics, then detected by scintillators and converted into electrical signals. Processing these signals provides spatial (X, Y) and energy information, resulting in a diagnostic nuclear medicine image, such as a thyroid scan.

FIG. 2.

Advanced imaging-guided approaches to enhance target localization in differentiated thyroid carcinoma.

Equation 4 expresses the time-integrated activity $B'$ for dosimetry,

B' = B_{0} . S_{eff} . \frac{I n 2}{S_{q}} * u [p - q a']

(4)

$B'$ is delivered by a radionuclide over time, $B_{0}$ is the initially administered activity, and $S_{eff}$ and $S_{q}$ represent decay dynamics. This integration $\frac{I n 2}{S_{q}}$ supports absorbed dose estimation $u$ in tumor $p$ and healthy tissues $q a'$ . Equation 4 calculates the cumulative activity of the integrated time activity. The cumulative activity $B'$ relies on these data to calculate the total absorbed dosage administered to both the tumor and healthy tissues. S_eff, which measures the radionuclide’s efficacy in delivering radiation to the tumor, may fluctuate owing to biological variables such as tumor vascularity or absorption efficiency. Any ambiguity in S_eff will proportionately influence the total activity and, therefore, the absorbed dosage, possibly resulting in over- or under-treatment. Likewise, the initial activity (B0), denoting the commencement of radiation dosage, is also of paramount importance.

Equation 5 calculates the absorbed dose $E$ to target tissue,

E = B' . T + I u [v - naq']

(5)

Equation 5 connects the $T$ -value, which represents $I u$ energy absorbed $E$ per unit activity $naq'$ in a given mass $v$ , with the total activity delivered $B'$ . This enables the measurement of the therapeutic effect, which is crucial for effective care and minimizing toxicity. In radiotherapy, clinical standards delineate the minimum absorbed doses necessary for tumor control, including the dosage required to achieve a 50% tumor control probability, as well as maximum dose thresholds to safeguard vital organs, such as the spinal cord or lungs, from radiation-induced damage.

Equation 6 expresses the dose volume histogram metric $U_{y}$ ,

U_{y} = \frac{U_{E \geq y}}{U_{s}} . 100

(6)

Equation 6 computes $U_{y}$ , the percentage of tumor volume receiving at least $y$ Gray dose. $U_{E \geq y}$ represents volume exposed to $\geq y$ dose, and $U_{s}$ is the total tumor volume. It helps assess whether the radiation plan sufficiently covers the tumor on the index of the dose volume histogram. Age and health condition may considerably influence tissue susceptibility to radiation. Older patients or those with comorbidities may exhibit heightened radiosensitivity, necessitating a decreased SE50 to account for a lower threshold for complications. Furthermore, genetic variables, including mutations in DNA repair genes, might influence the susceptibility of particular individuals to radiation-induced damage, hence affecting both SE50 and n. For instance, individuals with mutations in genes such as BRCA1/2 may have a reduced SE50, indicating that their tissues are more susceptible to injury at lower dosages.

The DMI-ACP framework aims to achieve clinical feasibility by adding a layer of decision support to the existing framework, thereby enhancing communication between radiologists, oncologists, and surgeons. A pilot study involving 15 patients with DTC demonstrated that planning time was reduced by 22% and boundary accuracy was achieved 31% more effectively than with conventional methods, thereby reinforcing surgeons’ confidence in performing complex resections. The system is compatible with hospital picture archiving and communication system and electronic medical record and could be modified to cover the use of other cancers that respond to iodine or other radiotracers such as fluorodeoxyglucose (FDG). SPECT/CT imaging’s iodine absorption is an important functional biomarker for evaluating tumor differentiation. More differentiated tumors tend to react better to radioactive iodine treatment if they have a high iodine absorption. Tumor vascularity can also be assessed by ultrasonography; a higher vascularity level is often indicative of more aggressive tumor activity. The overexpression of VEGF, a biomarker associated with enhanced angiogenesis in malignancies, may be a contributing factor. In a similar vein, 18F-FDG PET imaging may identify metabolic activity in tumors and corresponds with glucose metabolism; a high uptake of FDG is often indicative of a poor prognosis. Imaging characteristics may be further correlated with the presence of genetic indicators, such as BRAF mutations, enabling more tailored therapy options.

Figure 3 illustrates a potential diagnostic process for thyroid cancer, which includes biopsy, histology, genetic, and metabolic examinations. It underscores the differentiated subtypes of child thyroid cancer, specifically papillary and follicular. It correlates them with particular biomarkers, including mitochondrial DNA oxygenation errors, telomere lengths, metabolic intermediates such as ascorbic acid as well as phenylacetic acid to inform an improved classification and treatment plan. AI-driven models, especially convolutional neural networks, are used for tumor segmentation in SPECT/CT and ultrasound imaging. These models are designed to identify and outline tumor margins with exceptional accuracy, particularly in complex or irregularly shaped tumors that may pose challenges for manual assessment. Through the analysis of extensive datasets of labeled images, AI models can autonomously identify nuanced characteristics in the tumor’s morphology and location, thereby enhancing the precision of tumor volume assessment. Within the DMI-ACP architecture, deep learning facilitates the integration of multimodal data from SPECT and ultrasound, merging functional imaging (such as iodine uptake) with high-resolution anatomical imaging to provide a comprehensive perspective of the tumor’s attributes. This integration enables real-time tumor monitoring, with AI models continually refining tumor margins as they change during treatment, ensuring that the chemoradiotherapy regimen remains adaptable and individualized.

FIG. 3.

Molecular and metabolic profiling in thyroid cancer diagnosis and classification.

Equation 7 models the normal tissue complication probability $NTCP (a)$ ,

NTCP (a) = \frac{1}{1 + \frac{{S E}_{50} - E}{f (n . {S E}_{50})}}

(7)

Equation 7 quantifies $NTCP (a)$ the likelihood of side effects in healthy tissues. ${S E}_{50}$ is the dose at which $50 %$ risk occurs, $f$ is the delivered dose, and $n$ controls curve steepness. This metric supports risk–benefit balancing in chemoradiotherapy design. The probability of normal tissue complications calculates the normal tissue complication probability.

Equation 8 expresses the fused image intensity $J_{fused}$ for ultrasound and $J_{SPECT}$ ,

J_{fused} = β . J_{SPECT} + {(1 - β) . J}_{U S}

(8)

Equation 8 uses a fusion coefficient $β$ to integrate functional $β . J_{SPECT}$ and anatomical (ultrasound) data. Pixel intensities from the corresponding modalities are $J_{SPECT}$ and $J_{U S}$ . This method combines complementary information to improve lesion delineation. The weighting of image fusion was expressed and tends to be the purpose of Equation 8. Optimizing β is crucial for enhancing lesion delineation in the final fused picture using the advantages of both modalities. The parameter β determines the weighting assigned to each picture type, functional and anatomical, based on tumor characteristics and the objectives of the fusion. Empirical tuning is often used to determine the optimal value of β, involving the testing and evaluation of various β values based on their efficacy in enhancing picture fusion quality and tumor delineation precision. This procedure often entails juxtaposing the fused picture with ground truth data (e.g., histological findings or expert annotations) to identify the β that optimally reconciles the anatomical resolution of ultrasonography with the functional information from SPECT. A machine learning methodology may be used to automate this optimization. Supervised learning techniques, including regression analysis and neural networks, may be trained on annotated picture pairs to determine the optimal β value that enhances tumor detection or delineation accuracy.

Equation 9 calculates the planning target volume $PTV (b)$ margin for motion and setup errors,

PTV (b) = CTV + \sqrt{δ^{2} + α^{2}}

(9)

Equation 9 modifies the clinical target volume, $CTV$ , by random error $α^{2}$ and systematic error $δ$ . This guarantees precise radiation delivery and corrects for tumor position uncertainties and the margin of the planning target volume. As a tumor experiences growth or reduction over time, ∇E modifies the radiation dose distribution to accommodate changes in tumor shape. If the tumor diminishes, the correction factor decreases the dosage in regions no longer impacted by the tumor, while increasing the dosage in the remaining active tumor tissue. Moreover, alterations in the tumor microenvironment, including heightened hypoxia or modified vascularity, need ∇E to offset decreased radiation sensitivity in inadequately oxygenated areas. This modification guarantees that hypoxic regions get a sufficient dosage. Moreover, if tumor cells experience repopulation during therapy, ∇E may adjust the dosage to accommodate accelerated cell division in nonirradiated regions, hence enhancing treatment efficacy. Likewise, alterations in adjacent healthy tissues, such as inflammation or radiation-induced injury, are accounted for in the correction term, thereby optimizing the dosage to healthy tissue while enhancing therapeutic effectiveness.

Equation 10 defines the homogeneity index, $G J$ , in dose distribution,

G J = \frac{F_{2 %} - F_{98 %}}{F_{50 %}}

(10)

The doses received by $2 %, 98 %, and 50 %$ volume are denoted by $F_{2 %}, F_{98 %}, and F_{50 %}$ . A more uniform dose distribution is indicated by lower $G J$ values, which enhances the efficacy of therapy. Equation 10 evaluates the tumor’s dose uniformity with the index of dose homogeneity.

Equation 11 quantifies adaptive dose correction $\nabla E$ during treatment,

\nabla E = E_{v} - E_{u}

(11)

Equation 11 contrasts the originally planned dose $E_{u}$ with the actual delivered dose $E_{v}$ . This facilitates on-the-fly plan modification $\nabla E$ and helps identify deviations brought on by motion or biology. The correction factor of adaptive therapy illustrates the adaptive dose correction that occurs during treatment. Stratification according to DTC subtypes, including papillary thyroid carcinoma (PTC) and follicular thyroid carcinoma, is crucial for ¹³¹I SPECT/CT imaging, as these subtypes exhibit distinct iodine uptake patterns that can significantly impact treatment strategies. PTC, especially in its more aggressive or poorly differentiated variants, often exhibits heterogeneous or diminished iodine absorption, thereby restricting the efficacy of ¹³¹I SPECT/CT in precisely delineating tumor margins or metastases.

The reduction in radiation exposure can be quantitatively assessed by comparing the radiation dose distribution in the treatment area using the proposed method against traditional baseline methods, such as 2D imaging or nonadaptive planning. For example, dosimetric analysis may reveal that the baseline approach results in a substantial amount of radiation being delivered to adjacent healthy tissues owing to inadequate tumor localization. The suggested approach may decrease the “healthy tissue dose” by 15% to 20% while administering an equivalent or greater dose to the tumor, as seen by reduced tissue tolerance doses. The suggested technique uses adaptive planning, enabling real-time modifications during treatment to accommodate changes in tumor shape or location, thereby minimizing unnecessary radiation exposure. The target dose conformity index (e.g., dose gradient or therapeutic ratio) quantifies the enhancement, illustrating the increased efficiency of radiation distribution to the tumor relative to the baseline, hence yielding a more effective therapy with a reduced total radiation burden.

Result and Discussion

Dataset description

This dataset is used to forecast the return of well-differentiated thyroid carcinoma, incorporating 13 clinicopathologic characteristics. Each patient was tracked for a minimum of 10 years, and the data were compiled over a 15-year period.²⁰ To ensure the precision and reliability of the approximation, it is crucial to compare the ellipsoidal volume estimate with actual tumor volumes derived from accurate 3D segmentation using high-resolution imaging techniques, such as CT, MRI, or SPECT scans. By obtaining the accurate volume from various imaging modalities, which reflect the tumor’s genuine morphology, a direct comparison may be conducted to evaluate the precision of the ellipsoidal model. This validation procedure quantifies differences between the estimated and actual tumor sizes, providing insight into the model’s efficacy and potential limitations in clinical applications.

The Analysis of Diagnostic Accuracy compares the three methods—DMI-ACP, CEUS, and radiotherapy with radioactive iodine (RIT-RSM)—using different sample sizes in Figure 4. The bubble chart indicates that the diagnostic accuracy increases as the sample size increases. RIT-RSM consistently yields larger accuracy estimates relative to the other two methods, with the greatest differences occurring as the sample sizes increase. This indicates that RIT-RSM results can be used to provide confident test-case solutions compared with DMI-ACP and CEUS, demonstrating diagnostic test quality. The size of the bubbles indicates the influence (or weight) of the data at each point concerning the sample size. Table 1 shows the key diagnostic and therapeutic advantages. Equation 12 defines diagnostic accuracy and efficiency $DAE (a)$ ,

DAE (a) = \frac{T Q + T M}{T Q + T M + F Q + F M} . \frac{1}{T_{d}}

(12)

FIG. 4.

Analysis of diagnostic accuracy.

Table 1.

The Key Diagnostic and Therapeutic Advantages

Method	Key diagnostic advantages	Key therapeutic advantages
DMI-ACP (Dual-Modality Imaging-Guided Adaptive Chemoradiotherapy Planning)	- Precise tumor localization using SPECT/CT and ultrasound. - Enables real-time monitoring of tumor size, shape, and vascularity. - Provides functional and anatomical imaging for comprehensive assessment of tumor behavior.	- Adaptive therapy planning: Continuous feedback from imaging allows for adjustment of radiation dose based on tumor evolution. - Targets tumor tissue precisely, minimizing damage to healthy tissue. - Effective in chemoradiotherapy planning, improving both efficacy and safety. - Enables personalized treatment by accounting for tumor morphology and biological markers.
CEUS (Contrast-Enhanced Ultrasound)	- Provides real-time, high-resolution imaging of tumor vasculature. - Can differentiate between benign and malignant lesions based on blood flow patterns. - Offers dynamic visualization of vascular changes and angiogenesis in tumors.	- Noninvasive and can be used for monitoring treatment response through changes in vascularity. - Helps in guiding fine needle aspiration (FNA) for biopsy. - Can track tumor progression and assist in tailoring treatment to respond to vascular changes.
RIT-RSM (Radiotherapy with Radioactive Iodine)	- Provides functional imaging using iodine uptake to assess tumor viability. - Can identify iodine-avid regions, making it suitable for therapy planning in tumors that retain iodine uptake.	- Highly effective for tumors with high iodine uptake, offering targeted radiation therapy. - Minimizes radiation exposure to surrounding healthy tissues by specifically targeting iodine-avid tumor cells. - Offers long-term control for well-differentiated tumors, reducing the risk of recurrence.

Equation 12 assesses the $DAE (a)$ diagnostic imaging system’s performance. $T_{d}$ is the amount of time needed to finish the diagnosis, and $T Q, T M, F Q, and F M$ stand for prediction results. This metric evaluates system quality by combining speed and accuracy through diagnostic accuracy and efficiency.

The analysis also includes the analysis and shows the efficiency ratios for four diagnostic methods—DMI-ACP, CDU, CEUS, and RIT-RSM, which are shown as a radial bar chart for clinical workflow efficiency. CDU and DMI-ACP had the highest efficiency, with efficiency scores nearing 100%. CEUS and RIT-RSM lagged briefly behind, confirming accurate and efficient workflow efficiencies of CDU, as shown in Figure 5. Adaptive treatment planning, tumor detection, image fusion, radiation delivery, and posttreatment monitoring were among the critical workflow processes that the system tracked in real-time. Automatically produced time stamps for each step allowed the system to determine the overall time spent on each operation by dividing the start time by the finish time. To provide a comprehensive picture of the entire clinical process, their program also combines these times. By integrating the time-tracking system into the clinical software, the authors were able to compare the DMI-ACP framework with conventional baseline approaches. This provided us with clear insights into the time the authors saved using adaptive planning and enhanced imaging.

FIG. 5.

Analysis of clinical workflow efficiency.

Equation 13 calculates clinical workflow efficiency $CWE (v)$ ,

CWE (v) = \frac{M_{s}}{S_{t} - S_{i}}

(13)

S_{t}

is the duration of the session,

S_{i}

is the downtime, and

M_{s}

is the number of imaging/planning steps finished. Enhanced operational flow is reflected in higher

CWE (v)

values. Hence, Equation 13 calculates the effectiveness of clinical procedures. Monte Carlo simulations are extensively used in radiation treatment due to their ability to accurately describe the intricate interactions between radiation and matter, yielding highly precise estimates of dose distributions. The simulations evaluate the accuracy of the radiation dosage administered to the tumor and adjacent healthy tissues. The system utilizes anatomical and functional data from ¹³¹I SPECT/CT and ultrasound to provide dosage predictions, which are then evaluated against the outcomes of Monte Carlo simulations. These models simulate the transport of radiation across various tissue densities and tumor geometries, generating comprehensive 3D dose maps that are essential for assessing treatment effectiveness and safety. The juxtaposition of these simulated doses with clinical measures validates the precision of the proposed framework’s dose administration, guaranteeing that the intended doses efficiently target the tumor while reducing radiation exposure to healthy tissues.

As the analysis in Figure 6 shows, a stacked bar chart compares the four diagnostic methods—DMI-ACP, CDU, CEUS, and RIT-RSM—by increased targeting levels. When the targeting levels increased, RIT-RSM consistently contributed to a higher share compared with CEUS and CDU. DMI-ACP had a consistent moderate result, as there were lower targeted delivery levels; thus, circling back to the dosimetric integrity afforded to RIT-RSM, along with the dose repartitioning, offers a clear advantage. The incorporation of CEUS into the DMI-ACP framework significantly enhances its diagnostic and therapeutic capabilities. The DMI-ACP framework already integrates ¹³¹I SPECT/CT and ultrasound for accurate tumor targeting; however, the inclusion of CEUS may provide real-time information on tumor vascularity and perfusion, hence enhancing the overall understanding of tumor dynamics. CEUS, being nonionizing and offering high-resolution imaging, enables the monitoring of vascular alterations within the tumor, allowing for adaptive therapy adjustments based on real-time input. This integration enables dynamic tumor monitoring, allowing the framework to modify the radiation dosage in response to changes in tumor perfusion, therefore improving both treatment effectiveness and safety. Moreover, CEUS reduces total radiation exposure by decreasing dependence on ionizing imaging modalities, making it a significant enhancement to the system, particularly for patients who require multiple imaging procedures.

FIG. 6.

Analysis of dosimetric functional targeting.

Equation 14 defines the functional targeting index $FTI (u)$ ,

FTI (u) = \frac{U_{f} \cap U_{d}}{U_{f}}

(14)

$U_{f}$ is the biologically active tumor volume, and $U_{d}$ is the high-dose region. An $FTI (u)$ close to 1 signifies excellent targeting precision. Equation 14 evaluates how well the delivered dose overlaps with functionally significant tumor regions.

The last analysis of integration and scalability scores, as demonstrated in Figure 7, presents a stacked bar chart comparing DMI-ACP, CDU, CEUS, and RIT-RSM at various sample sizes. As the number of samples increases, the total score for scalability also increases. RIT-RSM consistently contributes the highest share in terms of integration and scalability, offering the most potential for expanding deployment in clinical and/or technological scenarios. The DMI-ACP architecture integrates ¹³¹I SPECT/CT with ultrasonography, yielding significant functional and anatomical information. SPECT provides functional imaging of radiotracer absorption in the thyroid; however, its spatial resolution is often inferior to that of PET. This implies that SPECT may have difficulty identifying tiny lesions or evaluating modest metabolic alterations inside the tumor. Conversely, ultrasound offers excellent spatial resolution for anatomical imaging, making it particularly helpful for tumor identification and assessment of vascularity, although it lacks functional insights, such as metabolic activity or tumor survival. Conversely, hybrid PET/MRI offers enhanced functional resolution due to high sensitivity of PET in identifying metabolic activity and excellent soft tissue contrast of MRI. PET imaging, with tracers such as 18F-FDG, delivers high-resolution data that facilitate the identification of minor metabolic alterations, even in early-stage DTC, which SPECT may overlook. MRI enhances this by providing detailed images of soft tissue structures and lymph node involvement, which are essential for evaluating tumor dissemination in DTC. PET/MRI offers a considerable advantage in functional imaging resolution, allowing for the superior identification of micrometastases or small lesions, thereby enhancing its efficacy for staging and early diagnosis of DTC.

FIG. 7.

Analysis of integration and scalability score.

Equation 15 defines the integration and scalability score $I S (v)$ ,

I S (v) = \frac{M_{e} . R_{c}}{D_{n}}

(15)

R_{c}

is the software/hardware units’ compatibility rating,

D_{n}

is the setup cost, and

M_{e}

is the number of interoperable software/hardware units. Better system-level deployment capability is suggested by higher

I S (v)

. Equation 15 evaluates the scalability and integrability of the suggested system.

The first step was the adoption of standardized imaging procedures to assist operators with tasks such as setting SPECT scan parameters, determining radiopharmaceutical dose, and ensuring ultrasound scan orientation and probe positioning. To provide a consistent picture acquisition, these techniques limit operator-dependent variability. To further ensure that all readers adhere to the same standards for recognizing tumor borders, vascular features, and functional hotspots, training and calibration procedures were also established. Automated image processing and segmentation techniques were incorporated into the system to reduce subjectivity further.

Conclusion

This study detailed the DMI-ACP structure, a combination of real-time ultrasound and ¹³¹I SPECT/CT imaging, which enables more precise administration of preoperative chemoradiotherapy in DTC. DMI-ACP can help achieve more precise tumor targeting, adaptive dose planning, and protect healthy tissues more effectively, made possible by combining anatomical and functional data. Clinical validation demonstrated increased tumor coverage, a low radiation dose, and effective planning. Although scalability and integration are some of the issues that do arise, the framework provides a prospective vision of intelligent and patient-based oncology planning. DMI-ACP marks a shift toward real-time, data-driven treatment plans in the field of thyroid cancer; however, the findings may also apply to a broader radiological setting. The real-time application of the system in a surgical context requires strong hardware integration with current operating room technology, which may lead to issues with system compatibility, real-time data processing, and image fusion precision. Synchronizing the SPECT/CT and ultrasound modalities with intraoperative procedures may encounter technological challenges, such as motion artifacts during surgery or data delays, which could impede the accuracy of tumor delineation and dosage administration. The DMI-ACP platform may be fully integrated into intraoperative environments in the future, enabling real-time tumor surveillance during surgical procedures. Advancements in imaging technology enable the integration of high-resolution, real-time ultrasound or MRI into the operating room, along with AI-driven planning algorithms that adjust to changes in tumor location or size throughout the process. This real-time modification may enhance tumor margin delineation, reduce radiation exposure to healthy tissues, and improve treatment outcomes by facilitating more precise targeting of the tumor.

Authors’ Contributions

B.S.: Validation and supervision. X.Xu: Data curation and validation. X.Xia: Formal analysis, software, and writing—review and editing. Q.W.: Conceptualization, methodology, and writing—original draft.

Footnotes

Funding Information

Jiaxing City Science and Technology Plan Project (2019AD32132).

Disclosure Statement

The authors declare that they have no conflicts of interest related to this research.

References

Shi

, Lv

, Liu

, et al. Relative burden of cancer and noncancer mortality among long-term survivors of differentiated thyroid cancer in the US. Front Endocrinol (Lausanne), 2024; 15:1425634; doi: 10.3389/fendo.2024.1425634

Kiyota

, Tahara

, Robinson

, et al. Impact of baseline tumor burden on overall survival in patients with radioiodine‐refractory differentiated thyroid cancer treated with lenvatinib in the SELECT global phase 3 trial. Cancer, 2022; 128(12):2281–2287; doi: 10.1002/cncr.34181

van Houten

, Netea-Maier

, Smit

. Differentiated thyroid carcinoma: An update. Best Pract Res Clin Endocrinol Metab, 2023; 37(1):101687; doi: 10.1016/j.beem.2022.101687

Zhang

, Zhang

, Su

, et al. Occupational daily walking steps have inverse relationship with papillary thyroid cancer risk and progression: A retrospective analysis. Int J Occup Med Environ Health, 2024; 37(1):58–71; doi: 10.13075/ijomeh.1896.02249

, Du

, Liu

, et al. Artificial intelligence for personalized medicine in thyroid cancer: Current status and future perspectives. Front Oncol, 2020; 10:604051; doi: 10.3389/fonc.2020.604051

Park

, Kim

, Oh

, et al. Highly accurate diagnosis of papillary thyroid carcinomas based on personalized pathways coupled with machine learning. Brief Bioinform, 2021; 22(4):bbaa336; doi: 10.1093/bib/bbaa336

Ieni

, Vita

, Pizzimenti

, et al. Intratumoral heterogeneity in differentiated thyroid tumors: An intriguing reappraisal in the era of personalized medicine. J Pers Med, 2021; 11(5):333; doi: 10.3390/jpm11050333

Chou

, Atway

, Bobbey

, et al. SPECT/CT imaging: A noninvasive approach for evaluating serial changes in angiosome foot perfusion in critical limb ischemia. Adv Wound Care (New Rochelle), 2020; 9(3):103–110; doi: 10.1089/wound.2018.0924

Gargus

, Jakubowski

, Arenas

, et al. Ultrasound shear wave velocity varies across anatomical region in ex vivo bovine ovaries. Tissue Eng Part A, 2020; 26(13–14):720–732; doi: 10.1089/ten.tea.2020.0037

10.

Cowan

, Mann

, Salibian

. Ultrasound in microsurgery: Current applications and new frontiers. J Clin Med, 2024; 13(12):3412; doi: 10.3390/jcm13123412

11.

Hussein

, Hamid

, Samie

, et al. Duplex hemodynamic parameters of both superior and inferior thyroid arteries in evaluation of thyroid hyperfunction disorders. Int J Gen Med, 2022; 15:7131–7144; doi: 10.2147/ijgm.s375016

12.

Akasu

, Jikuzono

, Matsui

, et al. Ultrasonographic detective flow imaging for evaluating parathyroid adenoma in patients with primary hyperparathyroidism. J Nippon Med Sch, 2024; 91(2):227–232; doi: 10.1272/jnms.jnms.2024_91-213

13.

Rios-Sanchez

, Bustos

, Barreto

, et al. Contrast-enhanced ultrasound for lymphatic mapping: A comprehensive review. Plast Aesthet Res, 2025; 12:N–NA; doi: 10.20517/2347-9264.2024.145

14.

Żyłka

, Dobruch-Sobczak

, Piotrzkowska-Wróblewska

, et al. The utility of Contrast-Enhanced Ultrasound (CEUS) in assessing the risk of malignancy in thyroid nodules. Cancers (Basel), 2024; 16(10):1911; doi: 10.3390/cancers16101911

15.

Solomon

, Petea-Balea

, Dudea

, et al. Role of Ultrasound Elastography and Contrast-Enhanced Ultrasound (CEUS) in diagnosis and management of malignant thyroid nodules—an update. Diagnostics (Basel), 2025; 15(5):599; doi: 10.3390/diagnostics15050599

16.

Allen

, Malo

, Jiao

, et al. Targeting melanin in melanoma with radionuclide therapy. Int J Mol Sci, 2022; 23(17):9520; doi: 10.3390/ijms23179520

17.

Pandey

, Subramanian

, Shaikh

, et al. Synthesis and preliminary biological evaluation of 177Lu-labeled polyhydroxamic acid microparticles toward therapy of hepatocellular carcinoma. Cancer Biotherapy & Radiopharmaceuticals, 2019; 34(5):306–315; doi: 10.1089/cbr.2018.2747

18.

Witney

, Blower

. The chemical tool-kit for molecular imaging with radionuclides in the age of targeted and immune therapy. Cancer Imaging, 2021; 21(1):18–14; doi: 10.1186/s40644-021-00385-8

19.

Kesavan

, Turner

. Myeloid toxicity of radionuclide cancer therapy. Cancer Biother Radiopharm, 2022; 37(3):164–172; doi: 10.1089/cbr.2021.0286

20.

Borzooei S, Tarokhian A. Differentiated Thyroid Cancer Recurrence [Dataset]. UCI Machine Learning Repository 2023; doi: 10.24432/C5632J. Available from: https://archive.ics.uci.edu/dataset/915/differentiated+thyroid+cancer+recurrence