Stimulated Raman Histology and Machine learning provides an alternative approach in brain tumor diagnosis

The study published in Nature Medicine examined the diagnostic accuracy of brain tumor image classification through artificial intelligence tool, compared with the accuracy of pathologist interpretation of conventional histologic images. The results for both methods were comparable: the artificial intelligence based diagnosis was 94.6% accurate, compared with 93.9% for the traditional pathologist-based assessment and interpretation. This research was conducted at NYU Langone Health.

The imaging technique known as stimulated Raman histology (SRH), reveals tumor infiltration in human tissue by collecting scattered laser light, illuminating essential features not typically seen in standard histologic images. Using deep convolutional neural network (CNN) with more than 2.5 million samples from 415 patients researchers trained a machine-learning algorithm to classify tissue into 13 histologic categories that represent the most common brain tumors, including malignant glioma, lymphoma, metastatic tumors, and meningioma.

Researchers note advantages with this method include quicker and almost realtime results (two and half minutes) with this method compared to traditional pathological examination (twenty to thirty minutes). This can be very useful especially during intraoperative decision making. In addition, researchers found that diagnostic errors due to this method are different from traditional pathology and combining both methods would increase diagnostic accuracy to 100%.