This page is best viewed in Firefox.

Imaging: Purpose


Sample EF5 image.
Black: Off-tissue; Blue: Tissue Mask; Red Tones: EF5.

The presence of hypoxia in tumor tissue is a serious hurdle in developing viable methods of treatment. In animal tumors, hypoxic cells limit radiation response. While this effect has not been as prominently identified in human tumors, this may be the result of inadequate evaluation and monitoring of tissue for hypoxia. Many scientists also believe that hypoxic cells may be resistent to cytotoxic drugs due to their relative isolation from the blood supply. EF5 provides the necessary mechanism to evaluate and identify regions of hypoxia at both physiologic and pathologic levels.

One of the premier benefits to EF5 over other markers of its kind is the ability to calibrate the intensity of the resulting photographed image. In doing so, we can directly relate pixel intensity values to meaningful biologic measurements. This brings us to a general goal of the imaging section of our lab: Quantitative Flourescent Microscopy.

Hypoxia is present in tumor tissue over a wide dynamic range of values varying from oxic to anoxic levels. Using our calibration process, we are able to visualize and therefore identify regions in the tissue with varying levels of hypoxia. We have broken these values into 5 biologically relevant bins representing: physiologic (0-1% EF5 binding), modest hypoxia (1-3%), moderate hypoxia (3-10%), severe hypoxia (10-30%), and anoxia (30-100%). These EF5 binding levels can be correlated to pO2 values according to the following chart.



After these regions are identified, it becomes possible to look at the events which occur in these regions of differing oxygen levels and directly compare the numeric values from one slide to another. The heterogeneous nature of hypoxia in tumor tissue often poses a problem in developing methods to accurately evaluate the effect of hypoxia on biologic processes. Our ability to accurately calibrate each image to a standard range of values enables us to overcome this roadblock and draw direct comparisons between two or more images from entirely different tumors.


Oxygen Map

Our ability to quantify the oxygen level at each pixel in an image permits us to generate accurate numeric data and evaluate relationships algorithmically. The field of imaging is often a "look and see" style of data presentation where conclusions are drawn based on the appearance of the image to the human eye. While this has its place, our methods allow us to generate statistical plots of relationships between various biologic events. We can then draw statistically significant conclusions from a wider data set than would be possible simply by looking at dozens of pictures and claiming to see a trend.

Our imaging group focuses on developing useful techniques and algorithms for quantifying results observed in our photographed sections of tumor. At present we are mostly concerned with the analysis of 2D frozen sections. However, the versatility of EF5 allows us to do both PET and MRI studies as well. 3D analysis of frozen sections will also become a viable option.


Comments? Questions? E-mail kochc@mail.med.upenn.edu