In Cytopathology : official journal of the British Society for Clinical Cytology
BACKGROUND : Intraoperative pathological diagnosis of central nervous system (CNS) tumors is essential in neuro-oncology to plan the patient management. Frozen section slides and cytological preparations provide architectural and cellular details analyzed by the pathologists to reach an intraoperative diagnosis. With the progress in artificial intelligence and machine learning fields, AI systems have significant potential in providing highly accurate real-time diagnosis in cytopathology.
OBJECTIVE : To investigate the efficiency of machine learning models in intraoperative cytological diagnosis of CNS tumors.
MATERIALS AND METHODS : We trained a deep neural network to classify 4 major brain biopsied lesions for intraoperative tissue diagnosis. Overall, 205 medical images were obtained from squash smear slides of histologically correlated cases, with 18 high-grade and 11 low-grade gliomas, 17 metastatic carcinomas, and 9 non-neoplastic pathological brain tissues. The neural network model was trained and evaluated using 5-fold cross-validation.
RESULTS : The model achieved 95% and 97% diagnostic accuracy on the patch-level classification and patient-level classification tasks, respectively.
CONCLUSIONS : We conclude that deep learning-based classification of cytological preparations may be a promising complementary method for rapid and accurate intraoperative diagnosis of CNS tumors.
Ozer Erdener, Bilecen Ali Enver, Ozer Nur Basak, Yanikoglu Berrin
2022-Dec-02
artificial intelligence, brain tumor, cytopathology, deep learning, intraoperative diagnosis, neural networks
