OBJECTIVE: In clinical practice, thyroid nodules are typically visually evaluated by expert physicians using 2D ultrasound images. Based on their assessment, a fine needle aspiration (FNA) may be recommended. However, visually classifying thyroid nodules from ultrasound images may lead to unnecessary fine needle aspirations for patients. The aim of this study is to develop an automatic thyroid ultrasound image classification system to prevent unnecessary FNAs.
METHODS: An automatic computer-aided artificial intelligence system is proposed for classifying thyroid nodules using a fine-tuned deep learning model based on the DenseNet architecture, which incorporates an attention module. The dataset comprises 591 thyroid nodule images categorized based on the Bethesda score. Thyroid nodules are classified as either requiring FNA or not. The challenges encountered in this task include managing variability in image quality, addressing the presence of artifacts in ultrasound image datasets, tackling class imbalance, and ensuring model interpretability. We employed techniques such as data augmentation, class weighting, and gradient-weighted class activation maps (Grad-CAM) to enhance model performance and provide insights into decision making.
RESULTS: Our approach achieved excellent results with an average accuracy of 0.94, F1-score of 0.93, and sensitivity of 0.96. The use of Grad-CAM gives insights on the decision making and then reinforce the reliability of the binary classification for the end-user perspective.
CONCLUSIONS: We propose a deep learning architecture that effectively classifies thyroid nodules as requiring FNA or not from ultrasound images. Despite challenges related to image variability, class imbalance, and interpretability, our method demonstrated a high classification accuracy with minimal false negatives, showing its potential to reduce unnecessary FNAs in clinical settings.

BACKGROUND: Molecular testing has been used as an adjunct to morphological evaluation in the workup of thyroid nodules. This study investigated the impact of two gene fusions, RET/PTC and THADA/IGF2BP3, that have been described as oncogenic events in thyroid neoplasms.
METHODS: We performed a retrospective, single-centered study at a McGill University teaching hospital in Montreal, Canada, from January 2016 to August 2021. We included patients who underwent surgery for thyroid nodules that pre-operatively underwent molecular testing showing either RET/PTC or THADA/IGF2BP3 gene fusion.
RESULTS: This study included 697 consecutive operated thyroid nodules assessed using molecular testing, of which five had the RET/PTC fusion and seven had the THADA/IGF2BP3 fusion. Of the five nodules in the RET/PTC group, 100% were malignant and presented as Bethesda V/VI. Eighty percent (4/5) were found to have lymph node metastasis. Twenty percent (1/5) had extrathyroidal extensions. Sixty percent (3/5) were a diffuse sclerosing variant of papillary thyroid carcinoma, and the rest were the classical variant. Of the seven THADA/IGF2BP3 nodules, all presented as Bethesda III/IV and 71.4% (5/7) were malignant based on the final pathology analysis, and 28.6% (2/7) were NIFTP. All the THADA/IGF2BP3 fusion malignancies were a follicular variant of papillary thyroid carcinoma. None had lymph node metastasis or displayed extrathyroidal extensions.
CONCLUSIONS: RET/PTC nodules presented as Bethesda V/VI and potentially had more aggressive features, whereas THADA/IGF2BP3 nodules presented as Bethesda III/IV and had more indolent behavior. This understanding may allow clinicians to develop more targeted treatment plans, such as the extent of surgery and adjuvant radioactive iodine treatment.