Progress Report

Prostate Cancer Math Clinic - Progress Report

The Spring 1996 Math Clinic at the University of Colorado at Denver has been in progress for nine weeks. During that time the sixteen students of the class have met regularly with Dr. Firouz Daneshgari and Scott Stewart of the Division of Urology at the University of Colorado Health Sciences Center and have become acquainted with the goals and underlying problems of the clinic. What follows is a report of accomplishments and work in progress at this time.

Geometric Modeling Team This team has worked on the existing software that pre-processes tumor data for input to the three-dimensional geometric model. The goal has been to work with Scott to streamline the existing code and automate as much of the pre-processing as possible. Specifically, the existing code has been modified to incorporate color tables so that Gleason grades and prostate components are uniquely and uniformly identified for all samples. New code has also been written that creates a uniform capsule width (two pixels) in the slice data. This work is complete at this time.

A second task for this team is to identify commercial and public domain three-dimensional visualization software that can be used for displaying prostate data. The group reviewed several different packages primarily through the Internet. At the moment, we have a license to use the NCAR package called PolyPaint+. This work is in progress.

The third crucial task for this team is to provide tumor data to the other teams. Because of his familiarity with the existing codes and the tumor data sets, Scott has been absolutely essential in this task. This work is ongoing.

Biopsy Team This team has done considerable background reading and research in an attemt to define the specific problems that need to be addressed and to identify the mathematical tools that are needed. Specifically, the team now understands the SCRSCB method and the existing 3-d biopsy code that simulates it. The team has also learned enough geometric probability/tomography and to assess whether it is relevant or offers an alternative to simulations

The team is now ready to proceed on two fronts.

  1. Having decided that the existing biopsy code that relies on the idea of core regions is inappropriate, the team will write a biopsy simulation code that works on real tumor data. The code will allow for any configuration of needles and will detect a hit or miss of tumor tissue in each individual prostate data set. Running the simulation over all data sets will give an empirical probability for a specific needle configuration. The goals are to compare the SCRSCB method to other methods and to find optimal needle configurations. This work is in progress.
  2. The team will also design a geometric biopsy simulation that uses ellipsoids for prostates and randomly generated and positioned ellipsoids for tumors. The program will specify a needle configuration and then apply it in simulation to many randomly generated cases. This approach will also produce an empirical measure of effectiveness of various needle configurations. This work is in progress.
Morphometric Models Team This team has designed what we are calling a secondary data base that consists of 15 ( 20 tumor and patient variables taken from both the autopsy and clinical data sets. The goal is to find innovative ways to present and analyze these data. Specifically, the group is planning to find correlations between variables in the data set, to compare three different measures of significance (Mayo, Stanford, and Brooklyn), to investigate the point of origin question, and to devise mathematical measures of shape. This work is being done with hypothetical or preliminary data until the secondary data base can be filled with real data. The tumor variables are being extracted from the tumor data sets by Scott, the patient data from the clinical sets is available, and the patient data from the autopsies may be limited. Once the real data are avialable, the team should be able to move ahead quickly. This work is in progress.

Cellular Automaton (CA) Team While not a goal of the original Clinic agreement, it was deemed worthwhile to assign a few students to the task of developing a CA tumor growth model. At the moment a rudimentary model has been built which grows "tumors." The team is now refining the model and incorporating anatomically correct zones of the prostate in order to make it give more realistic simulations of tumor growth.

Patient Database Nevin Hahn is working independently on this project with Firouz. He has already proposed a design for the data base and is ready to enter data.