An advanced optical imaging approach coupled with an artificial intelligence algorithm has been demonstrated to produce accurate, real-time intraoperative diagnosis of brain tumors, according to a recent study.
The study, which has been published in Nature Medicine, examined the diagnostic accuracy of brain tumor image classification through machine learning, which was compared with the accuracy of a pathologist interpretation of histologic images. Researchers reported that the results for both methods were comparable, with the artificial intelligence-based diagnosis being 94.6% accurate and the pathologist-based interpretation being 93.9% accurate.
The imaging technique – known as stimulated Raman histology – reveals tumor infiltration in human tissues by collecting scattered laser light, which in turn illuminates essential features not typically observed in standard histologic images.
These microscopic images are subsequently processed and analyzed with artificial intelligence. According to the team, surgeons can see a predicted brain tumor diagnosis in under 2.5 minutes. Additionally, by using the same technology after resection, the surgeons can accurately detect and remove the otherwise undetectable tumor.
“As surgeons, we’re limited to acting on what we can see; this technology allows us to see what would otherwise be invisible, to improve speed and accuracy in the [operating room], and reduce the risk of misdiagnosis,” commented senior author Daniel Orringer (New York University Grossman School of Medicine, NY, USA). “With this imaging technology, cancer operations are safer and more effective than ever before.”
You might also like:
Within the study, the researchers trained a deep convolutional neural network (CNN) with more than 2.5 million samples from 415 participants to classify tissue into 13 histologic categories that represent the most common brain tumors (e.g., malignant glioma, lymphoma, metastatic tumors and meningioma).
To validate the CNN, 278 patients were enrolled who were undergoing brain tumor resection or epilepsy surgery in the prospective clinical trial. Brain tumor specimens were biopsied from patients and split intraoperatively into sister samples that were randomly assigned to either the control group or the experimental arm.
The samples within the control arm were transported to a pathology laboratory, where they went through specimen processing, slide preparation by technicians and interpretation by pathologists. The experimental arm was performed intraoperatively, from image acquisition and processing to diagnostic prediction via CNN.
The investigators noted that the diagnostic errors contained within the experimental group were unique to those found in the control group. This suggested that a pathologist using the new technique could achieve close to 100% accuracy. In addition to this, the precise diagnostic capacity from the system could also be beneficial to centers that lack access to expert neuropathologists.
“Stimulated Raman histology will revolutionize the field of neuropathology by improving decision-making during surgery and providing expert-level assessment in the hospitals where trained neuropathologists are not available,” concluded co-author of the study, Matija Snuderl (New York University Grossman School of Medicine).
Sources: Hollon TC, Pandian B, Adapa AR et al. Near real-time intraoperative brain tumor diagnosis using stimulated Raman histology and deep neural networks. Nat. Med. doi:10.1038/s41591-019-0715-9 (2019) (Epub ahead of print); https://eurekalert.org/pub_releases/2020-01/nlh-nis010320.php