The assessment of volumetric change of enhancing tissue is regarded as an important parameter used by clinicians when seeking to monitor the response of brain neoplasms to therapy. Unfortunately, direct computation of enhancing volume requires the manual tracing and segmentation of areas of enhancement typically extending over multiple images, which is time consuming, labor intensive, and therefore impractical. Substitute methods are widely utilized (such as bi-directional measurements). Such surrogate measurements for tumor volume become problematic. Our long-term goal is to seek an objective, computer aided diagnostic (CAD) methodology for automatic computation of tumor (enhancing tissue) volume relevant for clinical decision making. Our hypothesis behind this project is that volume of enhancing tissue can be accurately measured through the use of advanced computer vision techniques, which will lead to an effective CAD system able to assist radiologists analyzing MR brain images.
The specific aims of the proposed project are 1) to improve the accuracy of the measurement of enhancing tissue by constructing high resolution 3D MR images and labeling enhancing tissue using learning based computer vision techniques, and 2) to develop a CAD system for enhancing tissue volume assessment using the designed techniques and evaluate the performance of the system on assisting radiologists for image interpretation. We believe this system will be well suited for use in patients undergoing treatment protocols/clinical trials who require short term serial imaging. It will better enable radiologists to give accurate quantitative clinical information. If the proposed research is completed successfully, the determination of enhancing tissue volume will be significantly advanced. It will enable the radiologists to rapidly provide objective, accurate, reproducible, and easily reported assessment of the tumor status. This will lead to a more rapid and reproducible assessment of neoplasm and therefore, hopefully influence patient outcomes in a positive way. The proposed research will also be applicable for usage on archived studies thereby enabling the volume of enhancing tissue to be calculated on these images as well.
The goal of this research is to develop a more accurate and reproducible way to measure the amount of disease present in patients suffering from brain tumors. Measurement methods currently being used are limited in accuracy, reproducibility and efficiency and hence we propose a method, if successful, will enable the computer to identify and measure brain tumors in a more automated fashion with improved accuracy. Improvements in tumor volume assessment are important for treatment planning as well as for assessing response during clinical drug trials.
