Methods

Study design

This prospective cohort study enrolled 50 patients scheduled for endoscopic thyroidectomy between March 2023 and March 2025. To minimize selection bias, a predefined 1:1 case-matched design was employed. The patients were assigned to the experimental group (NIRAF group; n = 25) to receive intraoperative NIRAF assistance, or the control group (Non-NIRAF group; n = 25) to undergo the procedure without NIRAF. Patients were prospectively matched based on baseline demographic characteristics and biochemical markers. This study was approved by the Ethics Review Committee (GYWY-L2023-77). Written informed consent was obtained from all patients prior to the enrollment.

Participants

We screened patients aged ≥18 years who were scheduled to undergo an initial endoscopic thyroidectomy. Patients were included if they had no contraindications to general anesthesia and agreed to comply with the scheduled follow-up protocol. Patients were excluded prior to enrollment based on the following criteria: (1) preexisting parathyroid disease or abnormal baseline calcium/PTH levels; (2) recent use of medications known to affect calcium or PTH metabolism (e.g., bisphosphonates, thiazide diuretics, lithium) within 3 months prior to surgery; (3) history of prior thyroid surgery, neck surgery, or neck irradiation; (4) severe liver or kidney dysfunction; (5) history of other malignancies or active antitumor therapy; (6) pregnancy; and (7) inability to cooperate owing to language barriers. Patients who required intraoperative conversion to open thyroidectomy were excluded from the final analysis.

Techniques

All procedures were performed using 4K ultra-high-definition endoscopic systems from the same manufacturer (OptoMedic, China; OPTO-CAM214K). This system allows the surgeon to shift among three intraoperative visualization modes: standard white-light, NIRAF and a fusion mode.

Surgery

All surgeries were performed by the same team of senior physicians with >16 years of experience in thyroidectomy.

Non-NIRAF group

Patients assigned to this group underwent standard endoscopic thyroidectomy under general anesthesia. After establishing the workspace, meticulous capsular dissection was performed to safely ligate the vessels and completely mobilize the thyroid gland. During excision and central lymph node dissection, the surgeon relied on the standard endoscopic white-light mode and anatomical landmarks to identify and preserve the PGs in situ. The excised specimens were meticulously inspected ex vivo to ensure that no PGs were inadvertently removed.

NIRAF group

The foundational surgical procedures were identical to those in the Non-NIRAF group. To facilitate PG localization, the NIRAF mode was activated. Systematic NIRAF scanning of the surgical field was continuously performed before, during, and after capsular dissection to identify and preserve PGs in situ. Furthermore, all excised specimens were routinely examined ex vivo using NIRAF. If NIRAF detected any suspected PG within the excised tissue, immediate parathyroid autotransplantation was performed.

Data collection

Baseline patient characteristics, preoperative assessments, and specific surgical details were prospectively collected. Serum calcium and intact parathyroid hormone (PTH) levels were measured preoperatively, on postoperative day one, and during scheduled follow-up visits. Postoperative hypoparathyroidism and biochemical hypocalcemia were defined as intact PTH levels <15 pg/mL and serum calcium concentrations <2.10 mmol/L [6], [20]. All surgical procedures were continuously video recorded to extract and analyze specific NIRAF imaging parameters and autofluorescence characteristics (e.g., brightness, uniformity, and edge definition) of both true PGs and confounding false-positive tissues. Inadvertent parathyroidectomy was documented if parathyroid tissue was histopathologically identified in any of the resected specimens.

Statistical analysis

All enrolled participants were included in the analysis as prespecified. Missing data were not imputed. Continuous outcomes were summarized as mean ± SD if approximately normally distributed and compared using two sample _t-_tests; otherwise, outcomes were summarized as medians (minimum–maximum) and analyzed using the Mann-Whitney U test. Changes in PTH and serum calcium levels were analyzed across pre-specified time points, and between-group comparisons were performed at corresponding visits. To evaluate the diagnostic performance of NIRAF, sensitivity and specificity were calculated. Furthermore, univariate and multivariate logistic regression models were prospectively planned and employed to identify the association between baseline patient factors (such as BMI, age, and sex) and the risk of inadvertent parathyroidectomy. A two-sided p-value <0.05 was considered statistically significant.