Breast Carcinoma Segmentation Based on Terahertz Refractive Index Thresholding

Cassar Quentin (1) , Philipp Hillger (2) , Janusz Grzyb (2) , Ullrich Pfeiffer (2) , Gaëtan Macgrogan (3, 4) , Jean-Paul Guillet (1) , Thomas Zimmer (5) , Patrick Mounaix (1)

1 IMS - Laboratoire de l'intégration, du matériau au système
2 University of Wuppertal
3 ACTION - Actions for OnCogenesis understanding and Target Identification in ONcology
4 Institut Bergonié [Bordeaux]
5 ToNIC - Toulouse NeuroImaging Center

A self-reference imaging method, in reflection configuration, was employed to extract from inverse electromagnetic problems the refractive index map of freshly excised breast carcinoma sections. Various thresholding strategies were tested to isolate the malignant pixels from the benign ones. Correlations with pathology images were performed to determine receiver operating characteristics of considered segmentations. To enlarge the field of inquiry, additional investigations towards principal component analysis were performed.

Mots clés: terahertz; breast carcinoma; refractive index map at 560 GHz

Domaines: Imagerie médicale; Cancer

Cassar Quentin, Philipp Hillger, Janusz Grzyb, Ullrich Pfeiffer, Gaëtan Macgrogan, et al.. Breast Carcinoma Segmentation Based on Terahertz Refractive Index Thresholding. 9th international THz-BioWORKSHOP, CNR, Apr 2021, Online, Italy. pp.60-61. ⟨hal-04561591⟩