News

Webinar | "Artificial Intelligence in Anatomopathological Diagnosis"

Webinar | "Artificial Intelligence in Pathology Diagnosis" is provided by Zaya Artificial Intelligence

The Webinar will present the way to automatically identify with the help of Artificial Intelligence of Mycobacteria in the digital images of Ziehl Neelsen staining tissue biopsies, used in the identification of TB.

Tuberculosis is the leading cause of death from infectious disease. Ideally, the diagnosis of TB is based on the identification of Mycobacteria. Because Mycobacterium tuberculosis is a 2-4 / 0.2-0.5 micron bacterium, its search in Ziehl Nielsen (ZN) colored slides with a 40X microscopic objective (0.5 mm diameter) takes hours; Artificial intelligence can provide viable solutions in this area.

Our team includes pathologists, machine learning engineers and IT and economics experts. The training data set includes 500,000 positive and 1,000,000 negative boundary boxes from the ZN slides of 81 paraffin blocks (21 positive, 60 negative); all slides were scanned using manual and automatic scanners; 7 pathologists noted positive and negative areas. The machine learning team proposed several image augmentation techniques coupled with different computer vision architectures. The test dataset includes 23 positive and 23 negative ZN slides (46 blocks); the slides were fully scanned (WSI). 4 teams of pathologists evaluated the WSI and lamellae separately; WSI was loaded for automatic analysis.

Automatic analysis of the WSI generates a report indicating areas likely to present mycobacteria. Our architecture shows very good results (0.9020 on the ROC curve). The results of the final test (machine versus man evaluating WSI, machine versus man evaluating blades and man evaluating WSI versus blades) show over 90% accuracy in all cases.

Artificial intelligence-based software for the automatic analysis of ZN colored slides has reliable results, providing extraordinary savings in the pathologist's time and effort.

 

Speaker:

Prof. Dr. Sabina Zurac

Professor of Pathological Anatomy Faculty of Dentistry Carol Davila University of Medicine and Pharmacy Bucharest, head of discipline, vice dean of research, primary pathologist, Pathological Anatomy Service Colentina Clinical Hospital Bucharest, head of department, member of several national and international scientific organizations (European Society of Pathology, International Academy of Pathology, International Society of Dermatopathology, Romanian Society of Dermatoonocology, European Academy of Dermatovenerology, Union Medicale Balcanique), author of over 300 scientific papers presented at national and international congresses and symposia, over 150 articles published in specialized journals, 23 specialized chapters / monographs, 3 patents, researcher / project manager in 21 completed research projects, one in progress.

 

 

 

The date is Thursday, May 19, between 16:00 and 18:00 (EET).

To participate, those interested are asked to register Here