- Thursday, April 25 - We finalized plans for writing and submitting
final reports and the teams worked on specific tasks. Nevin was able to
secure about 240 new cases of patient data for prostatectomies. These
tumors have not been sectioned, but the patient data can be used
immediately by the Morphometric Team.
- Tuesday, April 23 - Firouz attended and we worked in teams. The
biopsy simulations (data and ellipsoids) are running and we are
carrying out experiments. The Morphometric Team is working with Excel,
Paradox and S+ to analyze secondary data bases.
- Thursday, April 18 - Firouz attended class as saw demonstrations of
Vis5d and the biopsy simulations. There appears to be a snag in getting
patient data from local hospitals and Firouz will need to intervene to
get the needed permission; so we may not have complete patient data for
the secondary data base. However, Scott has made files with tumor data
and the Morphometric Team can extract those data for the data base.
Firouz discussed experimental design questions for the biopsy
simulations with the Biopsy Team. For both the tumor data and the
ellipsoids, we have a set of experiments to run in which several
variables are systematically varied. During those discussions of biopsy
simulations, another "Nobel Prize" question came up. How can TV/GV ratio
be best estimated from the contents of N needle biopsy cores?
- Tuesday, April 16 - The teams worked individually on Tuesday and
made good progress. The biopsy simulations on both data sets and
ellipsoids are working and ready for production runs. The Morphometric
Team is filling out the secondary data base. Dave had imported the
autopsy data into S+. Scott worked on the 3-d visualization program
Vis5d over the weekend and was able to start producing good images of
the tumor data sets. We will work with this package and not Polypaint.
- Thursday, April 11 - Firouz attended class today, heard reports,
and gave feedback. He said it was the "best news he had heard all day."
Both biopsy simulations are nearing completion. The Morphometric Team is
still waiting for data to put in the secondary data base. Nevin
can now get patient data for the clinical tumor data base. It looks like
there are some bad autopsy data files. David is working with S+ and
starting to do some correlation plots. Korky has looked at the
significance tests and found that all three of them require data that we
do not have directly. We can compute PSA density; check to see if it is
PSA over gland volume.
k
- Tuesday, April, 9 - We worked in teams today. The biopsy simulation
on data is progressing with Cathy and Guillermo writing code that will
be merged. Anne is writing the program for geometric simulation using
ellipsoids that uses the analytical work of the Biopsy Team.
- Tuesday, April 2 and Thursday, April 4 - There was good progress on
all fronts this week. The Biopsy Team has a biospy simulation code that
is working on data sets and coding has started on a biospy simulation
for ellipsoids. The Morphometric Team made progress on defiing zones
ofthe prostate, but input on this definitions is needed from Firouz.
Scott began writing an extraction program that can identify zones in the
tumor data, although it will work initially on octants (rather than
biological zones) of the prostate. David made progrss on running S+ on
the seconday data set and Korky began looking at and implementing tests
of significance. The tests of significance may be more difficult to
implement that originally thought. For example, the Brooklyn test
requires lengthof tumo found in the needle biopsy - a variable we do not
have. And the Stanford and Mayo tests need Gleason grades, rather than
the Geason scores that we have recorded in our tumor data. We also need
to get the clinical data into a secondary database.
Work on the CA is progressing and "moldy tumors" can be seen growing!
Nevin has the skelton of the patient data base established and the data
base is ready to be filled. As for PolyPaint ... the lawyers have gotten
hold of it and I don't know when or if we will see it this semester.
- Tuesday, March 26 and Thursday, March 28 -
Returning after Spring Break, we met with
Firouz again. He spent time with the Biopsy Team which discussed the
computer simulation of biopsies using the data set. He also met with the
Morphometric Team to discuss the analysis of tumor variables and
the definitions of zones of the prostate.
- Thursday, March 14 - Firouz visited and we met primarily in two
teams. The Biopsy Team discussed the details of the biopsy simulation
code that Cathy and Guillermo will start to write. The Morphometric Team
received further guidance from Firouz and will continue to design a it
analysis and presentation of tumor/patient data now in anticipation of
the arrival of data. Firouz mentioned several specific goals for the
team to keep in mind: (1) Look for correlations in data variables, (2) address
the significance question, (3) consider the point of origin question,
(4) are there mathematical definition of shape and can one be included
as a tumor variable, (5) think of brand new hypotheses.
- Tuesday, March 12 - Firouz visited and showed us, but did not
demonstrate, the biopsy gun. We then split into teams. Firouz met
with the Morphometric Team which is designing the presentation and analysis
of data once it is available from patient data bases and the tumor data
sets. The Biopsy Team is almost ready to begin writing simulation programs for
biopsies using both real 3-d tumor data and geometrical objects. Cathy
and Guillermo will help in this effort.
- Thursday, March 7 - We met with Firouz in CN 213 and once again
essential information was revealed, at times incidentally and
unexpectedely. Nevin received some important guidance on the data base
on which he can now take the next step. The Morphmetric Team put in an
order for patient data which Firouz will try to fill; someone may get a
trip to Texas. He also seemed to approve of the approach the team is
taking to presenting and analyzing the data. The Biopsy Team was given
the most clarification when we started probing Firouz about the reason
for modeling core regions (the cylindrical regions within which the
biopsy needles are randomly located). He confessed that he never
understood the rationale for the approach, that it is a mathematical
convenience and that it would be much more realistic to consider the
actual needles. He suggested that we carry out the simulations on the
actual data and try to design optimal biopsy nethods modeling the
needles rather than the core regions.
- Tuesday, March 5 - The class met in teams today. Matt and Nevin
pushed ahead on the CA model. Cathy and Guillermo moved closer to
finishing the preprocessing code and are now working on the 2-pixel
capsule problem. Having decided that Priya's code is not producing good
results on tiger and that we should start from scratch, the Biospy Team
now has two specific problems to work on - the needle approach and the
core region approach - either of which can be applied to geometric
objects or data sets. The Morphometric Team decided to let Scott write
the code to automate the extraction of size/shape data from the 3-d data
sets. The team will then focus on building the size/shape/patient data
base and analyzing it. David will investigate S+ for this analysis.
- Tuesday and Thursday, February 27 and 29 -
We met in teams in the CU-Denver Building. The Morphometric Team
continued to work with the tumor data and to design the format for
the size/shape/patient data base. The CA Team is designing the CA
model. The Geometric Model Team is nearing completion on the
preprocessing code. The Biospsy Team is still working on the "correct
approach to the right problem," but will have some clear directions by
the next meeting!
- Thursday, February 22 - The Clinic met in teams today and
progressed on many fronts. Specifically, the slice data sets,
the interpolated slice data sets, and the biopsy model were moved to
tiger so that all students can begin to connect their work to the real
data. Cathy and Guillermo continue to work (and hope to finish soon) on
the modifications of the preprocessing code. Matt is homing in on the
Kuros visualization software package. We need to be sure that this
package provides the features that the doctors want. Anne, Nevin, Matt
and Betty Jo continued to work on the CA models using a prototype that
Matt wrote. Scott met with the Morphometric Team (Theresa, Debbie, Corky
and David) to begin interfacing the data sets with the size/shape
questions. The Biopsy Team took two paths: Leonard and Daryl will
continue to investigate geometric probability calculations, while Linda
and Susan will work on the biopsy code of Priya in order to validate it
with real tumor data.
Morphometrics Data Base:
Here is a suggestion (by Anne) of the data table that can be made from the
digitized tumors:
CASE#|#of tumors|gleason|position|volume|capsule penetration|shape|
| | | | | | |
| | | | | | |
The position entry would include which zone the tumor is located in
(transitional, periphial, central) which could possibly involve using a
computer program to locate the zones on each prostate.
This tumor/slice data base could then be paired with the patient data
base that Nevin is working on. The patient data base will include (at least)
CASE#|Age|cause of death|PSA|.....
| | | |
| | | |
Once both of these data bases are completed, analysis of the data and
comparison of different prostates can be made to see if there are any
trends in tumor location or relationships between tumor grade and PSA or
other correlations.
- Tuesday, February 20. We met with Firouz, spending the entire
session updating him on progress and asking questions. The Geometric
Models group is progressing with both the preprocessing code and
selecting a postprocessing package. We need to get the tumor data as
soon as possible. The Biopsy team was given a course correction with the
emphasis on validating the computer biopsy model -- as soon as it
becomes available. Firouz answered many questions that will provide
valuable input to the work of the CA team and the Morphometric Team.
Nevin continues to work on the patient data base.
- Tuesday and Thursday, February 13 and 15 - On these two days we met
in teams to work on individual projects. One major sturctural change
took place: the Morphometric Team was split into two teams which we will
now call the CA (cellular automaton) Team and the Morphometric Team.
This change will be reflected in the other files of this home page.
Scott's team continued to work on modifying preprocessing software
(Cathy and Guillermo) and investigating post-processing visulaization
software (Matt, Betty Jo, Scott). The Biopsy Team is now working on both
analytical (geometric probability) methods and simulations to determine
probabilities of "needle hits." The (new) Morphometric Team is reading
the first three chapters of Ewald in hopes of applying stereology
methods to the tumor data. This group needs clarification from Firouz
on specific morphometric goals and questions. The CA group is moving
ahead designing a CA model for tumor growth.
- Thursday, February 8, 1996 - We had a very productive session with
Firouz in which many questions were answered and several goals were
clarified. In addition to enlightening us on the anatomy and physiology
of the prostate, the following points were made concerning each team
project.
Morphometric Modeling Team - The idea of a CA model seemed worthwhile to
Firouz, but he stressed that so little is known about the dynamics of
tumor growth within the prostate that we have very little basis for
designing transition rules for the CA. Such a model will be useful to
teh extent that it can answer the point of origin question.
He suggested that a more immediate and tangible goal for the group would
be to focus on questions of size and shape of tumors. We need to use
the data from the 50 cases that have been modeled (using the 3-d
geometric model) and try to characterize and relate size, shape,
volume, position within prostate, Gleason grade and significant
vs. insignificant.
Biopsy Team - For the first time we understood that there is a biospy
program that accompanies the geometric model. There are three steps
that must be taken: (i) validate the biopsy program in the gemoetric
model; does it really simulate the 6-probe (Hodge) method accurately? (ii) once
validated, the biopsy program for the 6-probe method should be applied
to the 50 prostate cases that have already been modeled; (iii) to become
rich and famous, we need to propose and test a better biopsy method in
which size of needles, angle of insertion, number of probes may all be
different that currently used. The new method should improve the
detection rate, detect only significant tumors, and account for
penalties in excessive probes.
Geometric Modeling Team - It became clear that the data and modeling
tools of the 3-d geometric model underlie all of the goals of the
clinic. The model and the 50 cases should be available to all of the
teams and should support their work. We need to extract the tumor data
from the 50 cases for use by the Morphometric Team. And we need to
make the biopsy program available to the Biospy Team.
- Tuesday, February 6, 1996 - The class met in teams and worked on
individual problems. Questions were prepared for Firouz' visit on
Thursday. IDL was discovered on carbon.
- Thursday, February 1, 1996 - We met in the sixth floor conference
room as one group for about ten minutes as Matt gave a brief description
of quad trees and oct-trees. The idea is already used in the 3-d
geometric model and appears that it might be useful for morphometric
methods. We then split into teams. The Geometric Models team met in the
workstation lab and Scott demonstrated some of the preprocessing and 3-d
modeling codes. All of the students in this group have tiger accounts
and are ready to write and modify codes. The first task is to improve
the gif-< RGB code. The team is still discussing various options for
visualization software.
The Morphometric Models team began discussing a CA model for tumor
growth. It appears that a MatLab program could be used that generates
output that could be interfaced with the the visualization package that is used by the Geometric Models team. The team needs more detailed prostate
physiology to design the transition rules of the CA.
The Biopsy Team began to analyze the efficacy of the "6-pack" biopsy
method using geometric probability. Linda ran a first cut simulation of
a 2-d biopsy program.
- Tuesday, January 30, 1996 -- After some housekeeping chores and a
quick exponential growth problem, we met in teams for the remainder of
the session.
- Thursday, January 25, 1996 -- On a cold snowy afternoon that caused
a few absences, the class split into three teams for the first time.
Scott worked with the Geometric Modeling Team and began identifying
specific tasks, particularly in the preprocessing stage. Anne met with
the Morphometric Models Team and began defining problems and identifying
relevant readings. Bill met with the Biopsy Team and began discussing
the paper by Firouz, et.al. critiquing the SRSCB method.
- Tuesday, January 23, 1996 -- The class spent the first half of the
session listing questions based on reading over the weekend. Firouz
arrived in time to answer many of the questions.
- Thursday, January 18, 1996 -- Scott spent 45 minutes summarizing
the work he does on the geometric model. He described the pre-processing
that prepares the input to the computer model and the proposed
post-processing that will enhance the output of the model. Bill then
attempted to summarize the goals of the other two teams (biopsy and
morphometric models).
- Tuesday, January 16, 1996 -- About 15 students were present for the
first day of class -- a large turnout for a Math Clinic. After
introductions, Dr. Firouz Daneshgari discussed the overall purposes and
goals of the Clinic.