Automated Cancer Diagnostic Method Combines Cutting-Edge Ultrasound Technology with AI

Each year, over 40,000 new thyroid cancer cases are reported. While 60-80% of patients with thyroid tumors undergo biopsies, the financial and potential physical toll of these procedures may be unnecessary for those with benign tumors. It is presently challenging for medical practitioners to accurately gauge the severity of a tumor, with different doctors having divergent opinions on a tumor's threat level.

Standard ultrasound methods, which generate images of tissues and organs based on the sound waves they reflect, are efficient at identifying thyroid tumors. However, the technology can struggle to distinguish the minute sounds emitted from small blood vessels, or microvasculature, from those of the surrounding tissue, despite microvasculature providing vital clues about a mass's cancerous nature. Although the introduction of contrast agents (chemicals easily visualized and commonly used in medical imaging procedures) allows ultrasound to display detailed images of tumor microvasculature, these substances need to be injected into patients and sometimes cause adverse side effects. While more recent ultrasound techniques can offer clearer nodule images, the ultimate evaluation still relies on the physicians' subjective judgment.

Researchers at the Mayo Clinic College of Medicine and Science (Rochester, MN, USA) have demonstrated that a pioneering cancer diagnostic method, which combines advanced ultrasound techniques with artificial intelligence (AI), can effectively diagnose thyroid cancer. This method — referred to as high-definition microvasculature imaging, or HDMI — noninvasively captures images of the minute vessels within tumors and, based on vessel characteristics, automatically categorizes the masses. The researchers believe that HDMI could potentially resolve the long-standing diagnostic challenge of assessing thyroid tumors in a clinical setting.

The researchers developed HDMI in an effort to develop an affordable, noninvasive imaging solution for evaluating thyroid tumors that could deliver quantifiable results and minimize errors. This system uses machine learning, a subset of AI, to assess high-resolution images of tumor microvasculature. The technique has already shown potential in accurately assessing breast tumors. In a recent study, the team tested HDMI on thyroid tumors in 92 patients. They captured images of the tumors using HDMI and analyzed a dozen features related to the size and shape of the microvasculature in the images, including their density and branching points. All patients in the study, in consultation with their physicians, chose to have their tumors biopsied to confirm their malignancy status. Those with tumors deemed cancerous underwent surgery for the removal of the mass.

The researchers provided their machine learning algorithms with 70% of their imaging data from the patient tumors, along with the malignancy status, to teach algorithms how to interpret various features. Through a process of trial and error, the algorithms constructed predictive models, which were then used to determine the status of tumors imaged in the remaining 30% of the data. HDMI's classifications were accurate 89% of the time, based on the clinical assessments of the biopsies and surgeries. These results suggest that HDMI could be a more reliable diagnostic method than traditional techniques and could spare numerous patients from unnecessary surgeries in the future. The researchers are now refining the method to enhance its accuracy even further. They plan to investigate its performance in diagnosing other types of cancer and whether it can assist in monitoring the effectiveness of chemotherapy on cancerous growths.

“Because HDMI allows you to objectively differentiate benign nodules from malignant ones, it could greatly improve diagnostic accuracy and reduce the number of unnecessary surgeries being done now,” said study author Azra Alizad, M.D., a professor of radiology and biomedical engineering at Mayo Clinic.

Related Links:
Mayo Clinic