There are about 120 different types of tumors in the central nervous system, and the first step in their treatment is usually surgery with histology acquisition. In this process, a piece of the tumor is removed and a histological frozen section is prepared at the neuropathology department for better assessment of the tumor characteristics. With the new laser-based imaging technique "Stimulated Raman Histology" (SRH for short), neuropathologists can now produce a report within minutes.
"With the new technology, fresh, untreated tissue samples are placed on a slide during tumor surgery and can then be analyzed directly in the operating room using the SRH device," Georg Widhalm from the Department of Neurosurgery describes the process. Since the tissue is analyzed in its untreated state, it is subsequently available without restriction for further detailed analyses as part of routine diagnostics. Needle biopsies can also be terminated much more quickly if diagnostic tumor tissue has been detected using SRH. Such procedures are not only used in the diagnosis of brain tumors, but can also be applied to confirm the diagnosis of other neurological diseases such as inflammatory diseases of the blood vessels and demyelinating lesions.
Primarily developed in the USA, the SRH technique was first used in Europe at the Department of Neurosurgery of MedUni Vienna and University Hospital Vienna under the direction of Georg Widhalm and is the subject of research. A scientific study at MedUni Vienna demonstrated 99 % agreement between digital histology and conventional frozen section (Wadiura et al; 2022).
"The new technology enables surgeons to make faster decisions regarding the optimal surgical strategy in the operating room, which significantly reduces the time in the operating room for patients. In addition, the safety of the procedure is also increased," says Widhalm.
The successful use of this new technology is the result of close collaboration between several university hospitals as part of the newly established Comprehensive Center for Clinical Neurosciences and Mental Health at MedUni Vienna and University Hospital Vienna. In the case of brain tumors, there is also close cooperation with the Comprehensive Cancer Center. With the establishment of Comprehensive Centers, the Medical University of Vienna and the University Hospital Vienna promote the cooperation of various university hospitals and departments around a patient group with the aim of ensuring that new diagnostic and therapeutic procedures are applied as quickly as possible.
Other possible uses of artificial intelligence are being explored
For most brain tumors, the goal is to achieve maximum safe tumor removal. For brain-derived tumors (gliomas), delineation of tumor tissue from healthy tissue during surgery is particularly difficult, and in some cases residual tumor can therefore be observed after surgery. A new AI technology is able to more accurately detect the tumor boundary. Surgeons can thus examine tissue samples taken during surgery at the suspected tumor boundary for the presence of residual tumor tissue.
In a recent multicenter study with significant participation from the Department of Neurosurgery at MedUni Vienna and University Hospital Vienna, another AI tool based on the SRH technique was tested. The results were published in the top journal "Nature Medicine" in 2023 and show that machine learning software using specific histological features from over 2 million image datasets recognizes over 93 % of specific genetic tumor features within a few minutes. These genetic tumor features play a crucial role in the assessment and treatment of brain tumors today.
Toward digital histopathological assessment in surgery for central nervous system tumors using stimulated Raman histology.
Wadiura LI, Kiesel B, Roetzer-Pejrimovsky T, Mischkulnig M, Vogel CC, Hainfellner JA, Matula C, Freudiger CW, Orringer DA, Wöhrer A, Roessler K, Widhalm G.
Neurosurgical Focus. 2022 Dec;53(6):E12.
Artificial-intelligence-based molecular classification of diffuse gliomas using rapid, label-free optical imaging
Hollon T, Jiang C, Chowdury A, Nasir-Moin M, Kondepudi A, Aabedi A, Adapa A, Al-Holou W, Heth J, Sagher O, Lowenstein P, Castro M, Wadiura LI, Widhalm G, Neuschmelting V, Reinecke D, von Spreckelsen N, Berger MS, Hervey-Jumper SL, Golfinos JG, Snuderl M, Camelo-Piragua S, Freudiger C, Lee H, Orringer DA.
Nature Med. 2023 Apr;29(4):828-832.
doi: 10.1038/s41591-023-02252-4. Epub 2023 Mar 23.