Mbirn: Defacing

GOALS

• Proof that defacing algorithm leaves brain untouched
• Proof that defacing algorithm removes subjects identity
• Mbirn deface: CODE DISTRIBUTION

---

---

TESTED SCENARIOS

• Mbirn deface: starting from dicom images

---

---

UPDATES

2006.03.17 (A. Bischoff-Grethe)

• Reviews for NeuroImage were reasonable, but editors felt material was not appropriate for their journal. Defacing statistical manuscript was edited to account for reviewers comments and submitted to Human Brain Mapping.

---

2005.10.24 (A. Bischoff-Grethe)

• Defacing statistical manuscript was submitted to NeuroImage.

---

2005.09.26 (A. Bischoff-Grethe)

• Defacing statistical manuscript was sent to all co-authors for review, with the request that any comments or corrections be returned to me by Monday, 5 October 2005.
• To date, fifteen of the twenty-one datasets for the behavioral study have been rendered with both left and right 3/4 portrait views.

---

2005.09.19 (A. Bischoff-Grethe)

• Currently creating the 3d rendered defaced images for the behavioral study. Images look better when presented as a 3/4 portrait view, therefore only those orientations will be used for the experiment itself. I am using AFNI's 3D Render plug-in to create the images, and I find that it is necessary to change parameters on a case by case basis to create the cleanest looking images. This variability is due to the differences in quality, age, and acquisition method of the provided T1s.

---

2005.09.12 (A. Bischoff-Grethe)

• Received statistical manuscript comments from B. Fischl. Manuscript was further edited and sent to C. Fennema-Notestine, B. Fischl, and G. Brown for additional comments. The current goal is to send to all co-authors on Wednesday, Sept. 21 for final comments.

---

2005.09.09 (A. Bischoff-Grethe)

• Statistical manuscript was sent to B. Fischl for review.

---

2005.08.24 (A. Bischoff-Grethe)

• Statistical manuscript was reviewed by C. Fennema-Notestine and returned to A. Bischoff-Grethe for additional editing. It is expected that the revised document will be sent to B. Fischl in early September.
• Behavioral experiment is in the set-up phase. Additional images are being prepared (3D renderings of defaced images) for use with the experiment. We expect to have preliminary data for the October 2005 AHM meeting.

---

2005.08.23 (T. Nguyen)

• UCI has tested the newer version of the mri_deface binary (v.1.16) and has attained similar results to the UCSD defacing. It should be noted that the data was converted to afni (BRIK) format using the to3D afni binary prior to defacing.

---

2005.08.10 (A. Bischoff-Grethe)

• UCSD determined that the cause of UCI's problems with inefficient defacing was related to the version of mri_deface. UCI had an earlier version of mri_deface than did UCSD; both of these versions are earlier than the most recent version in use at MGH. UCSD confirmed that the latest version at MGH (v 1.16) performed similarly to the current version UCSD has in-house (v 1.9).
• It was suggested to UCI that they acquire the latest version of mri_deface for their data anonymization. UCSD will confirm with Kevin Teich it is appropriate for UCSD to make v 1.16 available as the most current distribution for UCSD.

---

2005.07.25 (A. Bischoff-Grethe/C. Fennema-Notestine)

• a) Validation that defacing doesn't remove brain: complete. Manuscript in progress; ABG will send to BF for technical description revisions; discussion under revision.
• b) Validation that defacing removes facial identity: experiment programming underway; we will have preliminary data for the October 2005 AHM.
• c) UCSD working with UCI to get correct defacing output at their site. UCSD can deface UCI data correctly, however, UCI's pipeline produces different results. Source of error may be a difference in the defacing script or some difference in the anatomical volume's information available to defacing script. Currently testing 1) UCI's version of the defacing script here at UCSD and 2) the comparson of information available from image volumes created at UCSD and UCI.

---

2005.05.17 (A. Bischoff-Grethe)

• a) Validation that defacing doesn't remove brain: complete, manuscript in progress;
• b) Validation that defacing removes facial identity: behavioural data being slowly acquired, this and the analysis will have more traction once the (a) validation manuscript is out.

---

2005.03.24 (A. Bischoff-Grethe)

• Continued work on the statistical manuscript. Focus has been on the results section. The tables and graphs depicting the results should be finished today, and the text within the results section will be revised to reflect the additional information.

---

2005.03.08 (A. Bischoff-Grethe)

• Continued work on the statistical manuscript. Introduction has been fleshed out, and the details regarding the methods section are nearly complete. Plan to complete the methods section this week and begin editing the results section.

---

2005.03.03 (A. Bischoff-Grethe)

• Have begun work on the statistical manuscript. Preliminary introduction has been written, and an outline pertaining to methods, results, and discussion have been added for further development.
• Of the twenty-one datasets which can be defaced, fifteen have been visually inspected and accepted for use in the experiment. Four of the remaining datasets omitted the lower portion of the face in unaltered version and will be replaced with volumes which do include the whole face. The remaining two are awaiting visual inspection.

---

2005.02.17 (A. Bischoff-Grethe)

• Did not get to the statistical draft but plan to begin outline ASAP.
• There are now twenty-two datasets available for the behavioral paradigm. One dataset does not have a T1 image available, but the photograph will still be useful as a foil in the experiment. The twenty-one T1-weighted images have been defaced and are undergoing visual inspection. Once visual inspection has been completed, 3D renderings of the defaced images will be made and I will begin programming the experiment.

---

2005.02.03 (A. Bischoff-Grethe)

• Will start outlining the draft for the statistical manuscript next week.
• We now have twenty datasets available for the behavioral paradigm. Because it has been difficult to find additional participants, I plan to start preparing these datasets for use with a behavioral paradigm. The datasets should be ready for the next meeting.

---

2005.01.20 (A. Bischoff-Grethe)

• Hoping to start drafting the statistical manuscrip in Feb
• For the behavioral, we have 18 datasets, and I have a few more people to track down, some of whom have verbally agreed but I haven't seen to do the consent/photo. The goal is to have 25 datasets, which I'm pushing for before the next meeting.

---