Leveraging Global Diagnosis For Tumor Localization In Dynamic Cell Imaging Of Breast Cancer Tissue Towards Fast Biopsying

D. Mandache (1) , E. Benoit a La Guillaume (2) , M.-C. Mathieu (3) , J.-C. Olivo-Marin (1) , V. Meas-Yedid (1)

1 BIA - Analyse d'images biologiques - Biological Image Analysis
2 LLTech SAS Paris
3 Cancer Campus

We propose a fast aid-to-diagnosis biopsy assessment method convenient at the point-of-care, on account of both the imaging technique and the algorithm applied. The procedure implies a pipeline of classification and localization of tumors in breast cancer biopsies imaged with a recently developed non-invasive imaging modality: Dynamic Cell Imaging (aka Dynamic Full Field Optical Coherence Tomography). This allows for fast and interpretable extemporaneous cancer detection with high confidence; we obtained a performance of 96% classification accuracy together with a coarse localization of tumors, even so for single isolated invasive cells.

Mots clés: dynamic full field optical coherence tomography; classification; segmentation; convolutional neural network; aid to diagnosis; breast cancer

Domaines: Traitement des images [eess.IV]; Vision par ordinateur et reconnaissance de formes [cs.CV]; Apprentissage [cs.LG]; Urologie et Néphrologie; Cancer; Immunothérapie

D. Mandache, E. Benoit a La Guillaume, M.-C. Mathieu, J.-C. Olivo-Marin, V. Meas-Yedid. Leveraging Global Diagnosis For Tumor Localization In Dynamic Cell Imaging Of Breast Cancer Tissue Towards Fast Biopsying. 2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI), Apr 2021, Nice, France. pp.320-323, ⟨10.1109/ISBI48211.2021.9434110⟩. ⟨pasteur-03721547⟩