Authors:
Françoise Bouvet
1
;
Hussein Mehidine
1
;
Bertrand Devaux
2
;
3
;
4
;
Pascale Varlet
5
;
6
;
4
and
Darine Abi Haidar
7
;
1
Affiliations:
1
Université Paris-Saclay, CNRS/IN2P3, IJCLab, 91405 Orsay, France
;
2
Pôle Neurosciences, GHU-Paris, 75014 Paris, France
;
3
Service de Neurochirurgie, Hôpital Lariboisière, 75010 Paris, France
;
4
Université de Paris, Faculté de Médecine Paris Descartes, 75006 Paris, France
;
5
Department of Neuropathology, GHU Paris-Psychiatrie et Neurosciences, Sainte-Anne Hospital, Paris, France
;
6
IMA BRAIN, INSERM U894, Centre de Psychiatrie et de Neurosciences, F-75014 Paris, France
;
7
Université de Paris, IJCLab, 91405 Orsay, France
Keyword(s):
Classification, Endogenous Fluorescence, Machine Learning, Decision Trees.
Abstract:
Delineating brain tumor margins as accurately as possible is a challenge faced by the neurosurgeon during tumor resections. The extent of resection is correlated with the survival rate of the patient while preserving healthy surrounding tissues is necessary. Real-time analysis of the endogenous fluorescence signal of brain tissues is a promising technique to answer this problem. Multimodal optical analysis has been proved to be a powerful tool to discriminate tumor samples of different grade of gliomas and meningiomas from healthy control samples. In this study, Machine Learning methods are evaluated to improve the accuracy of such discrimination. Each sample is described by 16 feature given in input to a Decision Tree based model. Once the learning step is completed, the classifier achieves a 95% correct classification on unknown samples. This study shows the potential of Machine Learning to discriminate between tumoral and non tumoral tissues based on optical parameters.