Biweekly update meeting

Date: August 01, 2017

Time: 10:30 AM EST

Means: In person meeting

Attendees:

1. Ahmet Erdemir
2. Snehal Chokhandre

Agenda:

1. Discuss immediate action items from the last meeting.
2. Discuss progress of tissue testing protocol evaluation.
3. Discuss segmentation progress.
4. Decide immediate action items for the next meeting.
5. Other.

Ongoing Action Items:

• Ahmet
  • Complete reviewing the drafts of data manuscripts.
• Snehal
  • Continue filling in the dashboard for tissue testing reproducibility.
• All
  • Continue segmentation based on commitments listed in the Roadmap.

Notes:

1. Tissue testing

• Snehal conducted confined compression tests for cartilage using the 3 mm thick filter. The expectation was that there will be less instability and tilting of the filter resulting in more uniform data (no abrupt jumps). The indentor used has a 3 mm diameter head for covering as much of the filter surface as possible without risking running into compression chamber walls.
• There appears to be still some jumps in the data at lower strains but overall the behavior has improved.

• The behavior for the sample from the 48 yr old specimen appears to be repeatable (<9% error).

• The behavior of the samples from 78 yr old specimen is however inconclusive. Of the two samples tested, the first one has vastly varying behavior (with no definite trend), and the second sample has test 1 and 3 fairly repeatable but test 2 shows much higher forces.
• The cartilage strip from which the two sample were taken for the 78 yr old specimen appeared degenerated. Snehal thinks the large variations might be mostly due to the tissue itself.
• Since one specimen each for representing old and young population is not enough, Ahmet suggested a separate study with 10 specimens in each age group to conclusively make predictions / assertions about the repeatability based on specimen age.
• From the grant goals perspective, Ahmet thinks it is better to have one set of repeatability data (one specimen picked based on MRI quality and joint testing data). Essentially, samples tested from one specimen multiple times (preferably 5) for a total of 150 tissue tests done for the chosen specimen (30 tests/specimen normally). The rest or some of the 7 remaining specimens may be tested one or multiple times depending on the availability of time and resources.
• The confidence in the tissue mechanical testing data will be higher if performed multiple times and show good repeatability, even if it is just one specimen and will be very useful for specimen specific model development.
• Snehal has come across only one study so far that addressed the repeatability of experimental procedures for ligaments. This needs to be further explored.
• Snehal will focus on repeatability tests for other tissues for the 48 yr old specimen this month and by the end of August Snehal and Ahmet will make a decision on how to proceed with the grant goals.

2. Segmentation

• Snehal has segmented bones for oks006 and oks007 following segmentation specifications.
• Her goal is to complete all segmentation by end of 2017.

3. Data manuscripts

• No progress.

4. Other

• Snorkel workshop
  • Snehal attended the Snorkel workshop at Stanford University. The Snorkel platform is used for rapid development and modeling of training sets useful for information extraction from unstructured data. The goal was to be able to use the machine learning platform to extract material properties for various tissues from scientific literature. Snehal thinks that obtaining the initial training data manually might cover most of the relevant literature and Snorkel may not be necessary (due to limited literature). She will however continue to explore the platform and learn to use it effectively for other applications.