ESR12 – Mihael Galinac

Host Institution:



Rupert Ecker


Renate Kain, Mark Little

Project title: Computer assisted morphometry of pathological changes in renal biopsies from patients with AAV


Application of machine learning to kidney biopsy tissue morphometry, with correlation to clinical outcome parameters, can improve diagnostic and prognostic fidelity in renal AAV.


All relating to Objective 9: Machine learning: standardised automated identification of defined changes (descriptors) in renal biopsies. Validation of machine learning: comparison of “computer” recognised and manually determined changes. First clinical outcome study: defined descriptors suited to automated analysis will be correlated to clinical outcome in a small set of patients from the MUW cohort. Clinical validation study: additional patient biopsies from RKD biobank will be included to validate automated descriptor evaluation and their suitability to predict outcome using variable clinical data sets.

Expected Results:

Development of algorithms that allow automated analysis of predictive changes in renal biopsies in AAV (M3.1 (mth 36))

Planned Secondments:

MUW: 5 mths (learn about disease context and coordinate tissue sections and clinical data)

Rupert Ecker: