Deep Learning of Cell Morphology-Based Diagnosis of Sézary Syndrome – PHRT


Deep Learning of Cell Morphology-Based Diagnosis of Sézary Syndrome

Short Summary

Herein, we aim to resurge morphology-based diagnosis for hematological malignancies by overcoming current limitations through the establishment of an automated procedure integrating imaging flow cytometry and deep learning to achieve objective, ultra-high throughput and sensitive diagnosis. Specifically, imaging flow cytometry allows for imaging of single-cell morphology of hundreds of thousands of peripheral blood cells that in turn enables us to learn characteristic morphologies indicative of the presence of the disease. We focus on Sézary Syndrome, an aggressive cutaneous T cell lymphoma that is characterized by presence of tumor T cells with abnormal nucleus morphology in the peripheral blood.


Data-driven establishment of morphology-based diagnosis of Sézary Syndrome by means of imaging flow cytometry.


While we initially focus on Sézary Syndrome, our approach to define morphological peripheral blood cell signatures is likely applicable to a variety of other hematological malignancies or other diseases inducing morphological changes in the blood cell compartment, such as leukemia, or even inflammatory skin diseases.


Early diagnosis of cancer is a key determinant of patient outcome. Hematological malignancies manifest themselves in the blood and are therefore amenable to blood-based diagnostics. Traditionally, such diagnostic procedures rely on manual expert microscopical evaluation of blood cell morphology and suffer from subjectivity, limited throughput and low sensitivity.

Technology Translation

Prof. Dr. Andrew deMello

ETH Zurich


  • Prof. Dr. Manfred Claassen
  • Emmnuella Guenova (CHUV)



Funded by