Deep Learning Model Proves Useful in Head and Neck Cancer Care
Discover key findings from this NCI-funded research that assessed the effectiveness of a deep learning model that predicts survival and disease outcome in patients with head and neck squamous cell carcinoma (HNSCC).
Read the full article on JAMA Network.
Sarcopenia is the loss of muscle mass and function. It’s also a factor that can help predict outcomes in patients with HNSCC.
The process of evaluating sarcopenia is labor-intensive and impractical for large-scale use. Thus, researchers wanted to develop an image-based deep learning (DL) model that could use computed tomography data to assess sarcopenia more easily. They also wanted to know if the DL pipeline could predict patient survival and outcomes.
Researchers found that the model:
- was able to assess sarcopenia quickly and accurately.
- matched often with an expert clinician’s skeletal muscle segmentation for assessing sarcopenia.
- strongly correlated with existing findings, which state that sarcopenia is associated with poorer overall survival and longer percutaneous endoscopic gastronomy tube duration.
- could make it possible for doctors to include sarcopenia assessment in clinical decision making for patients with HNSCC. This model is easier and more efficient than existing segmentation muscle methods.