Optimizing Parathyroid Localization in Real Life: The Role of Expert‐Performed Ultrasonography

ABSTRACT

Objective

To evaluate the accuracy of expert‐performed ultrasonography (US) for preoperative localization in minimally invasive parathyroidectomy (MIP) compared with community US and other imaging modalities.

Study Design

Retrospective cohort study.

Setting

Tertiary referral center for endocrine surgery.

Methods

A retrospective, single‐center diagnostic accuracy study of 227 patients with primary hyperparathyroidism who underwent surgery and pre‐operative imaging by a dedicated neck radiologist (expert US). Sensitivity, specificity, and positive predictive value (PPV) were calculated, and were compared to other imaging modalities including community US, 99mTc‐MIBI, SPECT‐CT and 4DCT.

Results

Expert US showed high sensitivity (89%), specificity (98%), and PPV (95%), correctly localizing adenomas in 89% of cases. It outperformed the sensitivity and specificity of community US (50%/95%), MIBI (75%/95%), SPECT‐CT (82%/96%) and 4D‐CT (52%/94%), respectively, particularly in atypical cases like ectopic adenomas (25% prevalence) and multiglandular disease (6% prevalence).

Conclusions

Expert‐performed US should be considered the primary imaging modality for guiding MIP. Surgeons should consider referring patients to an expert ultrasonographer in cases of suspected ectopic adenoma, presence of multiglandular disease, before referring to 4D‐CT scans or when considering bilateral neck explorations.

1. Introduction

The success of minimally invasive parathyroidectomy (MIP) hinges on accurate preoperative localization of the hyperfunctioning parathyroid tissue [1, 2]. Traditionally, a dual‐modality approach combining ultrasonography (US) and technetium‐99m sestamibi scintigraphy (MIBI) has been recommended by guidelines, including those of the American Association of Endocrine Surgeons (AAES) [3]. However, recent improvements in imaging technologies have sparked interest in single‐modality approaches, potentially streamlining the diagnostic process and reducing health care costs.

Ultrasound, in particular, has garnered attention due to its non‐invasive nature, absence of radiation exposure, and real‐time imaging capabilities. The accuracy of US, however, is highly operator‐dependent, leading to significant variability in reported sensitivity ranging from 49% to 89%. This variability underscores the potential impact of sonographer expertise on diagnostic performance. Similar to thyroid disease, US for parathyroid localization is often done in the community. While several studies have found community US significantly inferior to expert radiologists [4, 5] or surgeon‐performed US [6, 7, 8] in thyroid disease, this has not been examined for parathyroid localization up to date.

In this context, our study aims to describe our ‘real world’ experience with the diagnostic accuracy of expert‐performed US in localizing parathyroid adenomas (PTA). Based on our clinical experience, the authors set to compare US performed by a dedicated radiologist with subspecialty expertise in head and neck imaging, with other commonly used preoperative imaging on which the surgeon often bases their surgical approach on. Additionally, we explore the efficacy of expert US in challenging scenarios such as ectopic adenomas and multiglandular disease, comparing its performance to community US and other common imaging modalities like MIBI and four‐dimensional computed tomography (4D‐CT).

By assessing the role of expertise in US performance, this study seeks to contribute to the ongoing refinement of preoperative localization strategies for primary hyperparathyroidism (PHPT), potentially improving surgical outcomes and patient care.

2. Methods

2.1. Study Design and Patients' Selection

A retrospective study was conducted at a single academic referral center, including all parathyroidectomies performed between January 2022 and June 2024. The study was approved by the Institutional Review Board (ASMC‐0019‐25) and conducted in accordance with the Declaration of Helsinki. Patients undergoing surgery for PTA were included. Exclusion criteria were parathyroid carcinoma, secondary or tertiary hyperparathyroidism, and incomplete medical records.

2.2. Data Collection

Electronic medical records were reviewed for demographic data, laboratory values, imaging reports, operative findings, and pathology results. Imaging modalities included expert‐performed US, community US, 4D‐CT, and MIBI scintigraphy with or without SPECT‐CT. Correct localization was defined as accurate lateralization and superior or inferior gland identification confirmed intraoperatively. Localization accuracy was assessed on a quadrant basis and was used as the primary endpoint.

Adenoma localization was divided into four quadrants of the thyroid gland. Failure to identify the adenoma was defined as the following: (1) incorrect localization of the specific quadrant where the adenoma was found during surgery; (2) preoperative localization of a single adenoma in cases of multiglandular disease and (3) when the adenoma was not found on imaging. Level of certainty for locating the adenoma was also examined for each imaging modality and was categorized as follows: (1) no adenoma was found; (2) low suspicion and (3) high suspicion for adenoma localization, as indicated in the imaging reports. Low suspicion was categorized when reports called for correlation with another imaging modality or were describing the findings as inconclusive, while pointing to a radiographic finding that may represent a PTA. Both low and high certainty localizations were considered positive for analysis and were classified as true positives when concordant with intraoperative findings.

An ectopic inferior parathyroid gland was defined as a gland in a location other than on\immediately adjacent to the anterior or posterolateral surface of the inferior pole of the thyroid gland. An ectopic superior parathyroid gland was defined as a gland in a location other than juxtacricothyroidal posteriorly or within the capsule of the posterior surface of the superior pole of the thyroid gland [9]. A small adenoma was defined as below 10 millimeters according to final pathology.

All enrolled patients were operated on by the same single, high‐volume surgeon (A.K.) and in a minimally invasive approach, and underwent ultrasonographic evaluation performed by two dedicated head and neck sonography specialists (A.B. and S.M.). Each possessing over 10 years of experience in head and neck imaging and conducting an average of approximately 180 neck US examinations per month. High resolution ultrasound device ‘Supersonic Imagine’ model no. RAP2102LM was used exclusively in the expert ultrasound tests, with the assistance of SL15‐4 linear array transducer. The US was performed using the ‘Thyroid<” preset, with patients examined in a supine position, upper torso elevated at a 30‐° angle, and chin extended upward. As a ‘real world’ study, other imaging modalities, apart from expert‐US, were performed at multiple centers nationwide. Community‐US examinations were conducted at various sites, resulting in nonuniform settings.

Results of expert US were compared to other imaging tests, including MIBI, with or without SPECT‐CT, 4DCT and ultrasound exams conducted by general sonographers from various medical centers across the country (referred to as community US). Surgical success was defined as identification of PTA on final pathology and normalization of calcium levels persisting for at least 6 months as defined by the AAES guidelines [3]. To mitigate potential bias, expert US was stratified into two subgroups: first, examinations performed after prior imaging with positive findings, in which the expert sonographer could theoretically have been influenced by preceding results, and second, examinations performed before any other imaging or after prior imaging with negative findings (Table 3).

Table 3.

Diagnostic performance of expert ultrasonography stratified by availability of prior imaging information.

	Prior correct imaging localization available	No prior correct imaging localization available	p value
Sensitivity	0.911	0.800	0.098
Specificity	0.984	0.965	0.197
PPV	0.956	0.889	0.113
NPV	0.967	0.933	0.120
AUC	0.948	0.883	0.103

Note: Prior correct imaging localization refers to cases in which at least one preoperative imaging modality correctly localized the adenoma.

Abbreviations: AUC, area under the curve; NPV, negative predictive value; PPV, positive predictive value.

2.3. Statistical Analysis

For normally distributed continuous variables, such as age, data were presented as mean and standard deviation, while median and range were reported for non‐normally distributed variables. Categorical variables, such as gender, were presented as numbers and percentages. The diagnostic performance of each imaging modality was assessed for the classification of parathyroid glands as normal or abnormal on a per‐quadrant basis. Each parathyroid was assigned to one of four thyroid quadrants, yielding a total of 908 evaluated quadrants across 227 patients.

Diagnostic performance of the imaging strategies was compared by calculating standard classification metrics, including sensitivity, specificity, positive predictive value (PPV), NPV, and the area under the receiver operating characteristic curve (AUC) with 95% confidence intervals. AUC values were compared pairwise using DeLong's test for correlated ROC curves. This method was chosen as it allows for robust statistical comparison of diagnostic accuracy across modalities evaluated in the same population. A p value of <0.05 was considered statistically significant.

All analyses will be performed using R (R Core Team, 2021, R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL: https://www.R-project.org/).

3. Results

Two hundred twenty‐seven patients who underwent parathyroidectomy met our inclusion criteria and enrolled into our study. Five patients (2.2%) were excluded from the study as shown in Figure 1. Demographics and laboratory data are presented in Table 1, exhibiting female predominance (161 patients, 71%) and a median age of 59 years (SD = 13, range: 18−80).

Figure 1.

Patient flow and imaging availability. Flowchart depicting patient screening, exclusions, final study cohort, and the availability of preoperative imaging modalities. The number of performed and missing examinations for 4D‐CT, SPECT‐CT, MIBI scintigraphy, and community‐performed US is shown. Each patient contributed four quadrant‐level observations, defined as right upper, right lower, left upper, and left lower. The final dataset therefore, comprised 908 quadrants. A quadrant was classified as positive for a given imaging modality when preoperative localization was concordant with intraoperative findings and confirmed by postoperative pathology. 4D‐CT, four‐dimensional computed tomography; MIBI, ^99mTc‐Sestamibi; SPECT‐CT, single‐photon emission computed tomography; US, ultrasonography.

Table 1.

Demographics and imaging distribution.

	N	Overall
Gender	227
F		162 (71%)
M		65 (29%)
Age	227
Mean (SD)		59 (13)
PTH	220
Mean (SD)		94 (60)
Calcium	225
Mean (SD)		11.21 (0.63)
Community US	167
Certain		69 (41%)
Doubt		38 (23%)
No adenoma		60 (36%)
Expert US	227
Certain		199 (88%)
Doubt		20 (8.5%)
No adenoma		8 (3.5%)
MIBI	205
Certain		139 (68%)
Doubt		36 (18%)
No adenoma		30 (15%)
SPECT‐CT	119
Certain		99 (83%)
Doubt		13 (11%)
No adenoma		7 (6%)
4D‐CT	53
Certain		16 (30%)
Doubt		18 (34%)
No adenoma		19 (36%)
Surgery Time	227
Median (IQR)		56 (46, 81)
Ectopic Adenoma	227
		57 (25%)
Multiglandular Disease	227
Single adenoma		212 (93.8%)
Double adenoma		10 (4.4%)
Triple adenoma		4 (1.8%)
Adenoma size (max)	227
Mean (SD)		15.0 (6.7)
Adenoma weight (max)	169
Mean (SD)		1.24 (2.06)

Abbreviations: 4D‐CT, four‐dimensional computed tomography; MIBI, ^99mTc‐Sestamibi; PTH, parathyroid hormone; SD, standard deviation; SPECT‐CT, single‐photon emission computed tomography; US, ultrasonography.

In total, 778 imaging studies were performed. The mean number of imaging for surgery was 3.4 ± 1.3 SD. Distribution of different imaging modalities utilization is presented in Table 1. Most commonly used initial modality was community US (94%), followed by MIBI (93%) scans. Sensitivity and specificity were calculated for each imaging modality: community US (50.3%/95.8%), expert US (89.75%/98.2%), MIBI (75%/95.8%), SPECT‐CT (82.1%/96%1) and 4D‐CT (52.5%/94.9%), respectively.

The average adenoma size on pathology was 15.0 mm (SD = 6.7, range = 6−65, Table 1). The average weight was 1.2 grams (SD = 2.1, range 0.12, max = 20.8, Table 1). For small adenomas, community US demonstrated significantly lower sensitivity compared with non‐small adenomas (22.7% vs. 56.4%, p = 0.003, respectively, Table 2), while this difference did not exist in expert US (89.3% vs. 90.9%, p = 0.7, respectively, Table 2).

Table 2.

Sensitivity for overall and small PTA localization.

	N	Overall sensitivity a	<10 mm a	≥10 mm a	p valueb
Community US	171				0.003
		89 (52.0%)	5 (22.7%)	84 (56.4%)
Expert US	237				0.7
		215 (90.7%)	25 (89.3%)	190 (90.9%)
MIBI	221				0.11
		167 (75.6%)	17 (63.0%)	150 (77.3%)
SPECT‐CT	131				0.5
		110 (84.0%)	13 (76.5%)	97 (85.1%)
4D‐CT	57				> 0.9
		31 (54.4%)	4 (50.0%)	27 (55.1%)

Note: ^a n (%); ^bPearson's Chi‐squared test; Fisher's exact test.

Abbreviations: 4D‐CT, four‐dimensional computed tomography; MIBI, ^99mTc‐Sestamibi; SPECT‐CT, single‐photon emission computed tomography; SD, standard deviation; US, ultrasonography.

As shown in Figure 2, expert US exhibited a substantially superior performance compared to community US, MIBI (with or without SPECT‐CT), and 4D‐CT (p < 0.001 in coupled analysis). The AUC was highest for expert‐US apart from combinations with 4DCT and MIBI. Differences between expert‐US alone as a combination with the community US, 4DCT, MIBI, and SPECT were statistically significant (p < 0.001). The addition of another imaging modality to the expert‐US did not improve its sensitivity in a clinically significant way, and no improvement was observed with the addition of community‐US.

Figure 2.

Comparative diagnostic performance of expert‐performed US and ancillary imaging for parathyroid localization. Receiver operating characteristic (ROC) curves illustrate the sensitivity and specificity of Expert US compared to four alternative modalities: (A) Community US, (B) MIBI scintigraphy, (C) SPECT‐CT, and (D) 4D‐CT. Each panel displays the individual performance of the comparator (yellow/pink/green/blue lines), Expert US (red lines), and the combined diagnostic yield of both modalities (purple lines). The shift of the “Both” curves toward the upper‐left corner demonstrates the incremental diagnostic value of combining Expert US with secondary imaging. Statistical comparison of the areas under the curve (AUC) was performed using DeLong's method. Full diagnostic metrics, including specific AUC values, 95% confidence intervals, and p‐values, are provided in Supporting Information S1: Table 1. 4D‐CT, four‐dimensional computed tomography; MIBI, ^99mTc‐Sestamibi; SPECT‐CT, single‐photon emission computed tomography; US, ultrasonography.

Ectopic adenoma was found in 57 (25%) cases with the paraesophageal, paravertebral and intrathyroidal being the most common locations (n = 22 [38.6%], n = 11 [19.3%] and n = 8 [14%] respectively). Multiglandular disease was identified in 14 patients (6.19%), comprising 10 cases (71.4%) with double adenomas and 4 cases (28.6%) with triple adenomas, as confirmed by pathology. Among these 14 patients, community ultrasound (US) was performed in 10 cases, demonstrating a sensitivity of only 10%, while Expert US yielding a sensitivity of 53.9%. Technetium‐99m sestamibi scans were conducted in 4 patients with a sensitivity of 30.8%, while SPECT‐CT was performed in 3 patients, achieving a sensitivity of 37.5%. Four‐dimensional CT correctly identified two adenomas, corresponding to a sensitivity of 50%.

Pre‐operative serum calcium levels averaged 11.2 mg/dL (SD = 0.63, range: 10.0−14.0 mg/dL). On post‐operative day 1, mean calcium levels decreased to 9.6 mg/dL (SD = 0.63, range: 8.2−12.1 mg/dL). Calcium normalization was achieved in 99.1% (225/227) of patients. The mean surgery duration was 68 min. There was no difference in surgery time between surgeries performed with a single imaging modality localized adenoma compared with multiple modalities (56 min (range: 47−86) vs. 55 (range: 46−77) respectively). In eight cases (3.54%) showing discordance between imaging modalities, elective four‐gland exploration was performed. In one additional case (0.47%), unplanned four‐gland exploration was required due to failure of intraoperative calcium normalization.

3.1. Effect of Prior Imaging Information on Expert US Performance

The diagnostic performance of expert‐performed US was evaluated according to the availability of prior diagnostic imaging information. In cases in which prior imaging results were available to the expert ultrasonographer, expert US demonstrated high diagnostic accuracy, with a sensitivity of 0.91, specificity of 0.98, PPV of 0.96, negative predictive value of 0.97, and an area under the curve of 0.95 (95% CI: 0.93 to 0.97).

In examinations performed without access to prior imaging information, defined as cases in which prior imaging failed to localize the adenoma or was not performed, diagnostic performance was reduced, with a sensitivity of 0.80, specificity of 0.97, PPV of 0.89, negative predictive value of 0.93, and an area under the curve of 0.88 (95% CI: 0.81 to 0.96). Sensitivity was higher when prior imaging information was available, although this difference did not reach statistical significance (p = 0.098). No statistically significant differences were observed for specificity (p = 0.197) or for overall discriminative ability based on area under the curve comparison using DeLong's test (p = 0.103).

4. Discussion

The aim of this study was to evaluate the role of expert US in detecting PTA pre‐operatively, by comparing its performance with the routinely commonly used modalities in everyday practice. Our results have shown that expert US yielded the highest overall sensitivity, specificity, PPV and NPV for detecting PTA, and outperformed also in atypical presentations such as ectopic adenoma and multiglandular disease. Even though previous studies have already demonstrated the advantage of expert US [10, 11, 12], our study includes a significantly larger cohort, operated by single surgeon and is the first study to compare community US to expert US regarding PTA localization, to the best of our knowledge.

The AAES recommends MIP in patients who appear clinically and by imaging to have a single PTA, while for patients with known or suspected multiglandular disease, bilateral neck exploration remains procedure of choice [13, 14]. Regarding preoperative localization, the AAES guidelines state that combining high resolution US and MIBI imaging enhances localization accuracy and sensitivity [15, 16]. With advances in imaging techniques, there is an increasing preference for reliance on a single imaging modality to reduce radiation exposure, shorten time to surgery, and lower costs and health care burden.

However, for a single modality to be relied upon, it must demonstrate high accuracy in localization, providing surgeons with precise anatomical guidance to improve the likelihood of successfully performing MIP. The confidence of the surgeon in the pre‐operative imaging cannot be underestimated. It is the key point which determines MIP candidacy, which driven guidelines to recommend dual imaging modality. Yet, for guidelines to be realistic and practical, a constant effort to assess “real‐life performance” is needed.

In this study, we evaluated the accuracy of expert performed US in localizing PTAs. Our results showed that in 89.8% of cases where expert identified the adenoma within a specific quadrant of the neck, the localization was correct. Overall, expert‐performed US demonstrated a high diagnostic sensitivity of 89%, specificity of 98%, and a PPV of 95%, suggesting it may be considered as a single stand‐alone imaging technique for adenoma localization.

These findings are within the upper range of previously reported results. Vitetta et al. have shown superior sensitivity of up to 84% in expert‐US in 61 patients undergoing parathyroidectomy [10]. Bunch. noted that the sensitivity of conventional US ranges from 49% to 89%, with PPV ranging from 78% to 98% [11]. This broad variability likely reflects ultrasound's inherent operator dependency—one of its primary limitations. In our study, this was evident in the marked difference between community US, which yielded a sensitivity of only 50%, compared with 89.8% of expert US. The lower sensitivity of community US joins previous works focusing on neck sonography regarding thyroid disease [4, 5, 6, 7, 8], in which it was inadequate and significantly inferior to experts (either radiologist or surgeons). This gap between the utility of community US (most common modality due to availability) and its performance requires real world data as well as alternatives. The main alternatives which have been increasingly used are surgeon performed US, new technologies (e.g., 4DCT) or expert‐performed US. While the formers have been investigated extensively, knowledge regarding expert US remains scarce.

Data on surgeon‐performed US for PTA supports the differences found between experts and community. David et al. reported a sensitivity of 87% for surgeon‐performed ultrasound in parathyroid localization [12]. Similarly, a larger study by Untch et al. involving 516 patients, also found a sensitivity of 87% and concluded that surgeon‐performed ultrasound is the optimal first‐line imaging modality for patients undergoing parathyroidectomy [17]. Notably, these studies did not report the proportion of cases with multiglandular disease in their cohorts. Additionally, the latter study did not specify the percentage of ectopic adenomas. It is also worth mentioning that the author acknowledged suboptimal MIBI performance compared to published averages, thereby emphasizing the superior performance of the surgeon‐performed US in their cohort.

Sub‐centimeter adenomas remain a diagnostic and surgical challenge and up to date. Stern et al. found that US sensitivity for sub‐centimeter adenomas was significantly reduced, requiring more experienced ultra‐sonographers [18]. Adkisson et Al. found out MIBI, and community US were significantly less accurate when adenoma's size was 13 mm or smaller [19]. In our cohort, 24 patients (10.6%) had adenoma sizing less than 1 cm. Our results support the conclusion by Stern et al. [18], as community US was significantly less sensitive in those adenomas, identifying less than quarter of the adenomas, while expert US demonstrated similar sensitivity rates. Reasons for the poor detection are operator related (most community US are performed by technicians) or device related (expert US tend to use better US models with dedicated probes and settings). While from an academic comparison perception, this comparison contains many confounders that may be explored in prospective, designed “head to head” studies, the aim of this current study was to describe real world data, hence expecting modalities ‘performance as is’.

Ectopic locations and multiglandular disease pose challenges to both diagnostic and operator staff up to date. The pivotal role of accurate pre‐operative localization cannot be over‐emphasized. In this current study, expert US outperformed all other imaging modalities. These two presentations are well‐recognized limitations of ultrasound in PTA evaluation, in addition to its operator dependence [20, 21, 22]. According to the literature, ectopic adenomas are found in approximately 16% of patients with PHPT. In our cohort, the prevalence was notably higher, with 57 patients (25%) presenting with ectopic adenomas. Similarly, the prevalence of multiglandular disease in our study was 6.2% (14 patients), compared to the 4% typically reported [23]. This higher incidence may be attributed to the referral pattern to our center, which has particular expertise in managing atypical presentations. As a result, patients with suspected ectopic or multiglandular disease are more likely to be referred to our institution for specialized evaluation and surgical management. Alternatively, it is possible that the expertise of the radiologist resulted in a higher rate of ectopic adenoma or multiglandular disease detection.

Recognizing that each imaging modality is subject to the interpreter's subjective assessment, we evaluated the level of certainty expressed in each report regarding adenoma localization, regardless of its levels of accuracy. Expert US demonstrated a high degree of certainty in localizing the adenoma in 88% of cases, compared to only 41% with community US. Notably, 4DCT exhibited the highest rate of inconclusive findings, with 34% of reports unable to definitively localize the adenoma. We hypothesize that equivocal or hesitant language in radiologic reports may prompt clinicians to seek additional preoperative imaging. Therefore, effective communication between surgeons and radiologists is essential to minimize unnecessary imaging resulting in delaying definite treatment.

The study results suggest two possible clinical implications. First, as shown in thyroid nodule management, surgeons must be familiar with the quality of their local community‐US as it may widely vary [24, 25]. In areas where community US sensitivity for PTA detection is inferior, endocrinologists and surgeons may consider using expert US or surgeon‐performed US as the initial imaging modality in the workup of PHPT, or conducting their own US following proper training, as suggested in other publications [26]. Efforts should be made to improve community US skills to detect PTA; sending feedback to community sonographers postoperatively or integrating artificial intelligence systems may help in improving detection rates, yet this should be further assessed in future studies [27].

The other implication regards ‘atypical’ challenging cases as negative previous imaging or in cases where a second imaging modality is needed to reassure atypical presentation (sub‐centimeter, ectopic or multiple PTAs). The option of expert US should be considered before escalating to SPECT‐CT or 4D‐CT, which expose patients to ionizing radiation.

Of note, a relatively low sensitivity of 4D‐CT was observed in our study (52%), which is lower than the reported sensitivity in the literature, ranging from 60% to 92% [28, 29]. One possible explanation is that 4D‐CT was not used as a first‐line imaging modality in our cohort. Instead, it was typically performed following inconclusive or discordant findings on initial imaging, such as ultrasound or MIBI scans. Furthermore, only 53 out of 227 patients in our study underwent 4D‐CT, reflecting its selective use. This targeted application in more complex cases may have contributed to the lower observed sensitivity, as these patients likely represented a diagnostically challenging subgroup and selection bias.

This study has also several limitations. First, its retrospective design inherently limits data quality and reduces control over potential confounding variables. Second, we did not assess cost‐effectiveness, which is a critical factor when considering a single‐modality approach and should be addressed in future research. Third, the number of patients who underwent 4D‐CT was relatively low (n = 53), and this imaging modality was reserved for selected cases with discordant or inconclusive findings on initial studies. As a result, the lower observed sensitivity of 4D‐CT in our analysis may reflect selection bias and the increased complexity of these cases.

Lastly, reflecting real‐life clinical practice, the study was not blinded, and prior US reports may have been available to the SPECT‐CT, 4D‐CT, and expert US teams, which may have introduced bias by influencing adenoma localization. Such exposure could have affected image interpretation either positively or negatively. Prior studies have shown that reliance on previous imaging, particularly when examinations are negative or contain false‐positive findings, may increase the risk of missed or incorrect interpretations [30, 31].

To address this potential source of bias, expert US was stratified according to the availability of prior correct imaging localization. Expert US demonstrated a numerically higher sensitivity in cases with prior correct localization compared with cases without previous correct localization (0.91 vs. 0.81), although this difference did not reach statistical significance (p = 0.098). No statistically significant differences were observed for specificity or for the area under the receiver operating characteristic curve (p = 0.197 and p = 0.103, respectively). These findings indicate insufficient evidence for a meaningful difference in diagnostic performance between the groups, and may reflect the inclusion of more diagnostically challenging cases among patients without prior correct localization.

In addition, the single‐center design may introduce referral bias, as patients evaluated at a tertiary referral institution may represent a selected population, including more complex or diagnostically challenging cases. This could influence the observed diagnostic performance of the imaging modalities. Furthermore, expert US was performed by highly experienced operators, and the results may therefore not be directly generalizable to settings with lower operator expertise. While this may limit external validity, the study reflects real‐world referral patterns and highlights the potential clinical value of expert‐performed US in specialized centers. Further multicenter studies are warranted to evaluate the reproducibility of these findings across different practice environments.

5. Conclusions

If adopted, an expert US‐first strategy has the potential to reduce radiation exposure, shorten time to surgery, and lower costs, while maintaining the feasibility of MIP, particularly in sub‐centimeter and ectopic disease. In challenging presentations, surgeons and endocrinologists should consider referral to an expert ultrasonographer prior to obtaining 4DCT imaging or when bilateral neck exploration is under consideration.

Author Contributions

Tzahi Yamin: conceptualization, data curation, resources, methodology, visualization, writing – original draft. Nir Battat: conceptualization, data curation, resources, methodology, writing – review and editing. Annat Blank: writing – review and editing, supervision. Shmuel Meirsdorf: Writing – review and editing, supervision. Bar Sara Zigdon: data curation, writing – original draft. Oded Cohen: conceptualization, methodology, supervision, writing – review and editing. Avi Khafif: conceptualization, methodology, supervision, writing – review and editing.

Funding

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.