CNN-based Approach for Cervical Cancer Classification in Whole-Slide Histopathology Images

• DOI: 10.48550/arxiv.2005.13924
• Publication Date: 2020-01-01
• Authors: Ferdaous Idlahcen; Mohammed Majid Himmi; Abdelhak Mahmoudi
• Keywords: Histopathology; Digital Pathology; Transfer of learning

Cervical cancer will cause 460 000 deaths per year by 2040, approximately 90% are Sub-Saharan African women. A constantly increasing incidence in Africa making cervical a priority the World Health Organization (WHO) terms of screening, diagnosis, and treatment. Conventionally, diagnosis relies primarily on histopathological assessment, deeply error-prone procedure requiring intelligent computer-aided systems as low-cost patient safety mechanisms but lack labeled data digital pathology limits their applicability. In this study, few tissue slides from TCGA portal were pre-processed to overcome whole-slide images obstacles included our proposed VGG16-CNN classification approach. Our results achieved an accuracy 98,26% F1-score 97,9%, which confirm potential transfer learning weakly-supervised task.