Achievement of Gözde Ünal
Medical images improved Interview with Prof. Gozde Unal
Gozde Unal is an associate professor, and the leader of the research group on Medical Image Analysis at Sabancı University, Faculty of Engineering and Natural Sciences, Istanbul, Turkey. She recieved her PhD in ECE at North Carolina State University, USA, and she was a research scientist at Siemens Corporate Research, Princeton, USA, working on various projects in medical imaging. Her current research focuses on applying mathematical modeling to find solutions to clinically relevant problems in computational medical imaging, over various modalities such as MRI, CT, intravascular ultrasound and optical coherence tomography. Her work currently involves modeling and quantification of change in longitudinal datasets as brain tumor Magnetic Resonance Image (MRI) series in the MICAT project.
How does the MICAT software improve the monitorisation of cancer treatments?
The objective of MICAT project is to improve the quantitative analysis of the changes in tumors during monitoring brain tumor patients undergoing radiosurgery, which is a substantial task for an accurate assessment of the cancer disease. The current tumor change measures in the clinic are based on only a two-dimensional major diameter measurement of the tumor. MICAT software facilitates sophisticated measures based on both volumetric and local deformation changes of the tumor.
What have been the results of the MICAT project so far?
The techniques developed under MICAT provide outlining of the tumor surfaces in 3 dimensions from MRI, over both baseline and follow-up patient scans, as well as, a rigid and a deformable alignment of the baseline and follow-up MRI volumes in order to compute criteria for tumor change over time. Specific local criteria produced are calculated from invariants of a strain tensor based on the deformation of the tumor after the radio-surgery. For instance, the resulting segmentation technique was among the best performing algorithms at the Multimodal Tumor Segmentation challenge in a recent conference (BraTS-MICCAI).
What challenges do you still have to overcome in your research?
Challenges exist due to vast variation and heterogeneity of the tumors both in terms of geometry and appearance in identification of the tumor tissue and its contents with utmost reliability, as well as due to tumor deformations in time which complicate the registration of tumor volumes. Furthermore, software solutions in a clinical environment should work close-to real time to have practical use. Our ongoing work aims to address these issues.
What differentiates this software from the other medical imaging technologies already available?
The MICAT software will enable the physician to work with a before-therapy and after therapy MR volume of the patient, outline the tumor surface in 3-dimensions from both volumes as well as separation of the necrotic part of the tumor tissue from the rest, visualize them in 3D, compute global and local measures of tumor change for an objective assessment and evaluation of the disease status. Neither of those are currently available in the radio-oncology or radiology departments of hospitals.