“Doctor, can you tell me what I have?”
Clinicians go above and beyond to answer this question. In difficult diagnostic cases, they perform biopsies —— extracting tiny tissue samples for examination under a microscope. However, for soft-tissue tumors such as peripheral nerve sheath tumors (PNSTs), biopsies can be challenging and are sometimes associated with pain and nerve damage.
Medical imaging can provide detailed information, but even experienced radiologists find it difficult to differentiate between malignant PNSTs (MPNSTs) and benign PNSTs (BPNSTs). A novel method to improve this differentiation is needed, and this is where Xinyi Wan and Christianne Jansma step in.
They developed machine learning models that analyze multiple types of MRI scans automatically, and compared the best model with experienced radiologists. To leverage the strengths of both AI and human expertise, they attempted to integrate the predictions from both.
Is this method ready for the clinic? Which type of MRI is best for this task? Find answers to these and more in the full paper of the study. It’s worth reading.
You can reach the paper directly here: https://www.mdpi.com/2072-6694/16/11/2039