Treatment Response Assessment in Oncology: Clinical Challenges and Technical Advances

Our MICCAI 2020 tutorial is motivated by the need for developing radiomic and image analytics tools for post-treatment response assessment in oncology. While significant strides have recently been made in the development of radiomics tools through multiple open-source efforts (pyRadiomics, CapTk, CERR), these have been primarily seen application in improved disease characterization on diagnostic imaging. However, nearly 80-90% of over 1.6 million patients diagnosed with cancer annually in the U.S have to be re-evaluated following neoadjuvant or adjuvant chemo-, radiation, or combination therapies, to identify those with residual or progressive disease (i.e. non-responders) compared to those with stable or regressing disease (i.e. responders). Unfortunately, benign “tumor-mimicking” treatment changes (i.e. pseudo-progression, fibrosis, radiation necrosis) confound the appearance of residual disease on routine imaging. There is hence an increasing awareness of the need for specialized quantitative tools to reliably assess post-treatment changes, preferably using routine imaging to distinguish non-responders from responders. 

This goal of this tutorial is to provide a comprehensive overview of both the clinical as well as the technical challenges involved in tumor response assessment in oncology. Our tutorial will showcase a range of didactic talks presented by leading clinical experts across radiology, radiation oncology, and medical physics as well as technical experts in image modeling and machine learning approaches. The primary learning objectives are:

  • New technical advances in developing radiomic approaches and deep learning strategies for predicting response to traditional and experimental oncology therapies, as well as post-treatment response assessment towards distinguishing treatment changes from recurrent disease 
  • Computational modeling, medical physics, and mathematical oncology approaches to characterizing chemo-resistant and radio-resistant tumors on imaging
  • Oncological perspective on the need for AI and machine learning in solving challenging problems with patient stratification for clinical trials as well as prognosis and prediction of disease using imaging. 
  • Radiologist’s perspective on burning questions in response to assessment in oncology and the role of new machine learning and AI techniques in serving as decision support to improve radiology reads. Additionally, extant clinical trial datasets will also be discussed to encourage the MICCAI community to develop new and advanced tools for improved response assessment in oncology. 
  • Radiation oncology perspective on the challenges in radiation therapy treatment planning and the need for AI tools that integrate with the radiation planning suites for accurate treatment dose planning. 

We believe the MICCAI community will benefit from a uniquely structured tutorial that attempts to bridge the gap between clinicians and technical experts. By having multiple clinical and technical disciplines present individual perspectives, we believe the audience will have a more wholesome understanding of the field and how image analytics tools have to evolve to meet a set of unique challenges in this unique problem domain.

Draft program TBA

Dates: During MICCAI, 4-8 Oct 2020