Diagnostic Impact of Cine Clips for Thyroid Nodule Assessment on Ultrasound

Background: Thyroid ultrasound is commonly performed using a combination of static images and cine clips (video recordings). However, the exact utility and impact of cine images remains unknown. This study aimed to evaluate the impact of cine imaging on accuracy and consistency of thyroid nodule assessment, using the American College of Radiology Thyroid Reporting and Data System (ACR TI-RADS). Methods: 50 benign and 50 malignant thyroid nodules with cytopathology results were included. A reader study with 4 specialty-trained radiologists was then conducted over 3 rounds, assessing only static images in the first two rounds and both static and cine images in the third round. TI-RADS scores and the consequent management recommendations were then evaluated by comparing them to the malignancy status of the nodules. Results: Mean sensitivity for malignancy detection was 0.65 for static images and 0.67 with both static and cine images (p>0.5). Specificity was 0.20 for static images and 0.22 with both static and cine images (p>0.5). Management recommendations were similar with and without cine images. Intrareader agreement on feature assignments remained consistent across all rounds, though TI-RADS point totals were slightly higher with cine images. Conclusion: The inclusion of cine imaging for thyroid nodule assessment on ultrasound did not significantly change diagnostic performance. Current practice guidelines, which do not mandate cine imaging, are sufficient for accurate diagnosis.

💡 Research Summary

This study investigated whether adding cine (video) clips to static ultrasound images improves the diagnostic performance of thyroid nodule assessment using the ACR TI‑RADS system. The authors retrospectively selected 100 thyroid nodules (50 benign, 50 malignant) from Duke University’s pathology database between 2017 and 2020, ensuring each case had both static grayscale images and a cine clip. Four fellowship‑trained radiologists (experience ranging from 1 to 25 years) independently read the cases in three rounds. Rounds 1 and 2 presented only static transverse and longitudinal images, with a ≥2‑week washout between them to assess intra‑reader reliability. Round 3 presented the same static images together with the corresponding cine clips, mimicking current clinical practice.

For each read, radiologists assigned TI‑RADS features (composition, echogenicity, shape, margin, echogenic foci) and derived a total point score, which translated into management recommendations (no follow‑up, surveillance, or fine‑needle aspiration). Sensitivity, specificity, and the number of cases where management changed between rounds were calculated. Additionally, a previously validated convolutional neural network (trained on 1,631 external nodules) was applied to the static images as a performance benchmark.

Key findings:

• Mean sensitivity for malignancy detection was 0.65 with static images alone and 0.67 when cine clips were added (p > 0.5).
• Mean specificity was 0.20 vs. 0.22 respectively (p > 0.5).
• Management recommendations changed in an average of 11 cases between the two static‑only rounds and 12.5 cases when comparing static‑only to cine‑included reads (p = 0.19).
• TI‑RADS point totals were modestly higher in the cine round (average absolute difference 1.38 points vs. 1.06 points between the two static rounds, p < 0.0001), but this did not translate into different risk categories or clinical actions.
• Feature‑level agreement remained high across rounds (composition agreement 0.70 ± 0.08, echogenicity 0.73 ± 0.07, shape 0.71 ± 0.07, margin 0.70 ± 0.09, echogenic foci 0.70 ± 0.08).
• The deep‑learning model performed worse than the human readers on this dataset (sensitivity 0.48, specificity 0.41), suggesting the selected cases were particularly challenging.

The authors discuss several reasons why cine clips did not confer a measurable advantage. First, the static images supplied were already the optimal transverse and longitudinal views (largest diameter, central portion), which capture most TI‑RADS criteria. Second, readers may have relied primarily on the static images even when cine clips were available, as both were presented simultaneously. Third, the dataset may be biased toward larger or more complex nodules that prompted cine acquisition, making the overall task harder for both humans and AI.

Limitations include: (1) a single‑center design with only four readers, limiting generalizability; (2) lack of measurement of reading time, confidence, or preference, which could reveal efficiency or subjective benefits of cine clips; (3) the artificial 50:50 benign‑malignant prevalence, which does not reflect real‑world disease prevalence; and (4) inability to isolate whether readers actually utilized the cine information during the third round.

In conclusion, incorporating cine clips into thyroid ultrasound interpretation did not significantly improve sensitivity, specificity, or management decisions compared with static images alone. Current ACR TI‑RADS guidelines, which do not mandate cine acquisition, remain appropriate. While cine clips may still be useful for educational purposes or in particularly ambiguous cases, routine use appears unnecessary from a diagnostic‑performance standpoint, allowing institutions to streamline workflows without compromising patient care.