Artificial Intelligence (AI) based on a combination of deep-learning algorithms and laser-imaging technology can be utilised to examine brain tissue and detect a brain tumour in near real-time according to a study published in Nature Medicine Journal on Monday. This recent AI technique can be a game-changer in intra-operative brain tumour diagnostics according to reports.
The method is a combination of “Raman histology (SRH), a label-free optical imaging method and deep convolutional neural networks (CNNs) to predict diagnosis at the bedside in near real-time in an automated fashion,” the study said.
The AI method is also much faster. The neural networks have been trained using over 2.5 million SRH images to identify brain tumours using brain tissue in under 150 seconds, according to the report. The traditional method of examining brain tissue for tumours is called a frozen section, which takes close to 30 minutes. The method, however is not quite reliable.
In addition, AI can detect more details such as spread of a tumour along nerve fibres. The method can also be useful in other surgical procedures, where doctors need to analyse other tissues while operating, including neck, breast, skin and gynaecologic surgery.
Covering gaps in skill sets
The method can help address the shortage of neuropathologists, which sometimes leads to delay in the detection and treatment.
Expert surgeons often tend to rely on more extensive methods of examination that can be done only once the surgery is completed and the process can take up to weeks at a time. The new AI tool can help fill the gap in specialisation and can help examine and detect signs of a tumour much faster than the traditional method during the surgery itself. which can reduce casualties with timely treatment.
The method can simplify intra-operative diagnosis, i.e., diagnosis during the surgery as the process using traditional methods can be complicated and labour intensive. “The existing work-flow for intra-operative diagnosis based on hematoxylin and eosin staining of processed tissue is time, resource and labor intensive,” co-author Daniel Orringer said in the study.
Another major benefit of the AI method is that it does not destroy the tissue. The same sample can be used again.
Accuracy in diagnosis
The study was carried out on brain tissue from 278 patients, which was examined during the surgery. The samples were split into two parts with one half being examined using AI while the other half was examined by a neuropathologist.
The technique was tested against manual detection with the results slightly favouring AI, with human diagnosis scoring 93.9 percent against AI’s 94.6 per cent.
More extensive tests were performed on the samples early on to gather findings and examine them against the conclusion by AI and the neuropathologist in order to ensure their accuracy. “Having an accurate intra-operative diagnosis is going to be very useful,” said Dr. Joshua Bederson, Chairman of neurosurgery for the Mount Sinai Health System, as quoted in the New York Times .
The AI was pitted against some of the best neuropathologists, including professionals from Columbia University in New York, the University of Miami and the University of Michigan, Ann Arbor. “I think that what happened with this study is that because they wanted to do a good comparison, they had the best of the best of the traditional method, which I think far exceeds what’s available in most cases,” Dr. Bederson said in the New York Times report.
AI has worked wonders in the field of healthcare with tech giants such as IBM and Google investing in AI-driven health-care start-ups such as Welltok and Ayasdi, respectively.