direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Inhalt des Dokuments

Harshita Sharma

Image Analysis in Digital Histopathology

Image analysis in digital histopathology is a currently expanding field of research. Histological images are inherently complex in nature and contain a wide variety of visual information with different stains, magnifications and types of tissues. Our research has been performed in collaboration with Department of Pathology and IT, Institute of Pathology, Charité University Hospital, Berlin, Germany. Histological whole slide images have been analyzed using classical methods (involving sophisticated handcrafting) and deep learning methods [1], aiding different clinical and routine applications. These include cancer classification [2],[3], necrosis detection [2],[4], cell nuclei segmentation enhancement [6], determining tissue composition [7] and content-based image retrieval [8]. Deep learning methods have shown tremendous potential for the studied problems in digital histopathology [2]. Among traditional methods, graph-theoretic approaches [3],[8] have been explored due to their ability to effectively represent spatial arrangements and neighborhood relationships of tissue components [5]. The prime research objective was to develop methods assisting pathologists in computer-aided diagnosis and prognosis of diseases, and automating certain tasks which can reduce manual labor and inter-and intra-observer variability.


  • Completed doctoral studies with thesis titled "Medical image analysis of gastric cancer in digital histopathology: methods, applications and challenges" at the Technical University Berlin in April 2017.

  • Was invited and felicitated for a guest lecture on "Medical Image Analysis using Modern Description and Learning Techniques" by NIEC, Delhi, India in March 2017.


  • Presented a poster "Image Analysis of Gastric Carcinoma in Digital Histopathology" during the Computational Life Sciences Workshop (Bayer), held at Berlin, Germany in November 2016.