Head and neck cancers present a significant treatment challenge, often requiring a combination of surgery, radiation, and chemotherapy. While these treatments can be effective in curing the disease, they are also notorious for their severe side effects. A key concern is the development of sarcopenia, or muscle wasting. This condition can lead to difficulties in eating and drinking, resulting in malnutrition and a multitude of problems, including the possible need for a feeding tube, reduced quality of life, and even earlier death. Detecting sarcopenia early is essential but traditionally has been a labor-intensive process. Doctors usually assess muscle mass through computed tomography (CT) scans, either of the abdomen or the neck. Since CT scans of the neck are common in patients with head and neck cancer, they offer an opportunity for early identification and intervention for sarcopenia. However, diagnosing sarcopenia from these scans requires a specialized expert who can differentiate muscle from other tissue, a process that can take up to 10 minutes for each scan.

Researchers from Dana-Farber Cancer Institute (Boston, MA, USA) have developed an artificial intelligence (AI) tool that can quickly and accurately diagnose sarcopenia using CT scans of the neck in patients with head and neck cancer. The application of AI streamlines what is otherwise a painstaking process prone to human error, performing the assessment in just 0.15 seconds. The development process began by training the AI model using clinical records and CT scans from 420 patients. An expert manually assessed muscle mass for each patient based on the scans and calculated a skeletal muscle index (SMI) score. This data was used to train the deep learning model, and a second dataset was utilized to validate the model's performance. Impressively, the model made clinically acceptable assessments 96.2% of the time.

The tool could have broad clinical applications. Current methods often rely on body mass index (BMI) as an indicator of health decline related to treatment. However, when the team compared the effectiveness of BMI and SMI in predicting poor outcomes, they found that SMI was a superior predictor, suggesting it could become a crucial clinical tool. The introduction of AI-based assessment means that sarcopenia could be monitored frequently throughout a patient's treatment. Early detection might prompt interventions such as nutritional support, medication, or physical therapy, potentially improving overall outcomes. It could also influence treatment decisions at the outset, as understanding a patient's muscle mass might inform a tailored, perhaps gentler, treatment strategy.

“Sarcopenia is an indicator that the patient is not doing well. A real-time tool that tells us when a patient is losing muscle mass would trigger us to intervene and do something supportive to help,” said lead author Benjamin Kann, MD, a radiation oncologist in the Department of Radiation Oncology at Dana-Farber Brigham Cancer Center. 

http://www.gzjiayumed.com/en/index.asp