Rafael Ceschin

Developing the serial quantitative imaging pipeline used in this work was part of my master’s research project as a graduate student. At the time, the phase I immunotherapy trial in this cohort had just wrapped up, and it became apparent that the question of identifying pseudoprogression close to the time of occurrence was paramount for the treatment and management of these patients. Having prior experience with quantitative diffusion imaging in pediatrics, this presented a worthwhile challenge.

Immunotherapy is an exciting new field in cancer treatment, but it also creates new challenges for diagnostic imaging. Conventional methods are not sufficient for accurate stratification and treatment management, even though imaging is often the primary source of diagnosis and planning in the absence of needle biopsies, particularly in diffuse intrinsic pontine gliomas.

Therefore, new advanced imaging techniques must be incorporated into the clinical workflow to improve patient care. While this was a pilot study, we are encouraged by the results, and we hope to further validate our findings in the upcoming phase II trial.

Two of the great benefits of publishing open-source code are the immediate feedback and high adoption rate. Since publishing this work, along with a technical overview of the pipeline in Cancer Informatics, we have had both internal and external researchers interested in applying this method to their own populations. We look forward to seeing their work published soon. Since the original publication, we have expanded our analysis to include patients diagnosed with a variety of tumor types and treatments. We will be publishing these findings in the coming months.

Our current research effort is geared toward incorporating multimodal quantitative imaging, including arterial spin-labeling and perfusion imaging, into the serial functional diffusion mapping pipeline. This will give us further granularity into the underlying microstructure of tumor tissue, allowing for more personalized treatment for each patient. Additionally, we will be developing methods for automated tumor segmentation to remove the labor-intensive task of manual tumor delineation. This will allow us to quickly analyze large-scale cohorts, increase the reproducibility of tumor segmentation, and create a purely automated tumor quantitation pipeline.

Read this article at AJNR.org …