Medical image-based outcome prediction can provide information for the risk stratification and precision medicine of cancer patients. Machine learning plays an important role in the analysis of medical images and have demonstrated success in many outcome prediction tasks. However, machine learning-based outcome prediction still faces many challenges, limiting its application in the clinic. In his PhD thesis, Jianan discussed several major challenges in the clinical translation of outcome prediction algorithms and developed machine learning techniques to address them. Jianan first tackled the problem of low generalizability of radiomics models caused by inter-institution variabilities. The differences in study population, imaging equipment and imaging protocols can lead to large variations in the enhancement patterns of contrastenhanced MRIs. Jianan proposed a feature selection strategy informed by the mechanism of contrast agents to select contrast-agnostic features. The model trained with the selected features were successfully validated in multiple cohorts from multiple hospitals. This study showed the possibility of training a widely generalizable radiomics biomarker by keeping in mind the issue of overfitting and Computer Vision News 36 Congrats, Doctor Jianan! Jianan Chen has recently completed his PhD at the University of Toronto, under the supervision of Anne Martel. His research focused on improving outcome prediction for cancer patients using machine learning and liver MRI. Jianan has recently accepted a postdoc position at University College London, where he will continue to study cancer, this time using the integration of genomics and digital pathology. Congrats, Doctor Jianan!
RkJQdWJsaXNoZXIy NTc3NzU=