Machine learning model can help radiologists diagnose thyroid nodules

A new radiomics-based machine learning model can evaluate immunohistochemistry (IHC) features and CT images to predict the presence of thyroid nodules, according to a new study published in the American Journal of Roentgenology.

The technique identifies proteins and enzymes typically associated with certain thyroid cancers, wrote lead author Jiabing Gu, and colleagues. Clinicians may eventually use it to differentiate between benign and malignant thyroid nodules.

“Our proposed radiomics model based on image features had high accuracy for three IHC markers, which is useful for clinical decision making because it provides radiologists and oncologists with a potential quantitative tool for those suspected thyroid nodules,” the researchers added.

For their study, Gu and colleagues from the University of Jinan in China, enrolled 103 patients with suspected thyroid nodules who had received thyroidectomy and IHC analysis from January 2013 to 2016. All patients underwent CT imaging prior to surgery and 3D Slicer v 4.8.1 was used to analyze the images of the sample taken from surgery.

The team chose four target IHC markers for their model: cytokeratin 19, galectin 3, thyroperoxidase, and high-molecular-weight cytokeratin. They used the Kruskal-Wallis test to improve the classification performance of the model which was built from 86 reproducible and nonredundant features.

Overall, the best performing cytokeratin 19 model achieved an 84.4% accuracy in the training cohort and 80% in the validation cohort. The thyroperoxidase and galectin 3 predictive models yielded accuracies of 81.4% and 82.5% in the training cohort and 84.2% and 85.0% in the validation cohort, respectively.

According to the researchers, galectin 3 has proven to be a sensitive marker for diagnosing thyroid malignancy and the protein cytokeratin 19 has helped identify the presence of papillary thyroid cancer. The authors also noted that the positive immunoexpression of thyroperoxidase can act as an independent risk factor for thyroid malignancy.

The high-molecular-weight cytokeratin model was not accurate (65.7%) and couldn’t be validated, according to the researchers.

“Correct diagnosis of suspected nodules at an early stage is important for successful treatment, Gu and colleagues wrote. “This model can be used to predict the presence of cytokeratin 19, galectin 3, and thyroperoxidase, which can help radiologists to diagnose and identify benign and malignant thyroid nodules.”

""

Matt joined Chicago’s TriMed team in 2018 covering all areas of health imaging after two years reporting on the hospital field. He holds a bachelor’s in English from UIC, and enjoys a good cup of coffee and an interesting documentary.

Around the web

CMS has more than doubled the CCTA payment rate from $175 to $357.13. The move, expected to have a significant impact on the utilization of cardiac CT, received immediate praise from imaging specialists.

The newly cleared offering, AutoChamber, was designed with opportunistic screening in mind. It can evaluate many different kinds of CT images, including those originally gathered to screen patients for lung cancer. 

AI-enabled coronary plaque assessments deliver significant value, according to late-breaking data presented at TCT. These AI platforms have gained considerable momentum in recent months, receiving expanded Medicare coverage in addition to a new Category I CPT code.

Trimed Popup
Trimed Popup