Abstract #4167

ADC-Map Based Computer Aided Radiological Diagnostics (CARD) for the Initial Differential Diagnosis of Medulloblastoma versus Pilocytic Astrocytoma – A Reproducibility Study.

Urspeter Knecht¹, Nicole Porz², Beate Sick³, Elvis Murina⁴, Nuno Miguel Pedrosa de Barros⁵, Philippe Schucht⁶, Evelyn Herrmann⁷, Jan Gralla¹, Roland Wiest¹, Marwan El-Koussy⁵, and Johannes Slotboom⁸

¹Neuroradiology, University Hospital Bern, Bern, Switzerland, ²Neurosurgery, University Hospital Bern, Bern, Switzerland, ³Department of Biostatistics, Institute of Epidemiology, Biostatistics and Prevention, University of Zürich, Zürich, Switzerland, ⁴Institute for Data Analysis and Process Design, ZHAW, Switzerland, ⁵Neuroradiology, University Bern, University Hospital, Bern, Switzerland, ⁶Neurosurgery, University Hospital Bern, Bern, ⁷Radiooncology, University Bern, University Hospital, Bern, Switzerland, ⁸Neuroradiology, University Hospital Bern, Switzerland

The diagnosis of brain tumors using visual criteria is very challenging. A novel computational method for computer aided radiologic diagnostics (CARD) is described based on quantitative textural features from ADC-maps, and a machine learning algorithm (Random-Forest classification). The reproducibility of the method was examined with 3 human raters was performed, and the Fleiss'-Kappa-test revealed high inter-rater agreement of κ=0.821 (p-value<<0.001) and an intra-rater agreement of κ =0.822 (p-value<<0.001). The method significantly improves the differential diagnosis of medulloblastoma versus pilocytic-astrocytomas.

This abstract and the presentation materials are available to members only; a login is required.

Join Here