Bibliographic record
Abstract
DMRI is used for creating visual representations of the structural connectivity of the brain, also known as tractography. Research has shown that using a tissue classifier can be of great benefit to create more accurate representations of the underlying connections [1] The aim of this project was to implement an image segmentation algorithm in DIPY [2] for classifying the different tissue types of the brain using structural T1 weighted images (T1-w) and diffusion MRI images (dMRI), and to incorporate the resulting tissue probability maps for Anatomically-Constrained Tractography (ACT) [3] We used Diffusion Power Maps (DPMs), which are scalar maps that are calculated from dMRI data and have a tissue contrast similar to the T1-w. By performing the tissue classification on dMRI derived scalar maps, the T1-w to dMRI registration step can be avoided. We used a Bayesian approach for the segmentation in a similar fashion than the methods proposed in [4] and [5] by applying the Max-imum-A-Posteriori (MAP) procedure. The prior probability was modeled with Markov Random Fields (MRF). The MRF distribution was modeled as a Gibbs distribution. We used the Expectation Maximization (EM) algorithm to update the tissue labels at each site and to update the parameters of the log-likelihood in all iterations. The first row of Fig. 21 shows the tissue classification on T1-w, the initial segmentation based on maximum likelihood and the final segmentation after 10 iterations and beta=0.1. Beta determines the weight of the neighborhood in the MRF model. These two parameters were tuned and validated by permuting 42 different combinations and calculating the Jaccard index between the segmentation of the proposed method against manually segmented brains from the IBSR dataset [http://www.nitrc.org/projects/ibsr]. The second row of Fig. 21 shows the probability maps of the three main tissue classes of the brain. The top row of Fig. 22 shows on the left the Diffusion Power Map (DPM), followed by its tissue classification and the streamlines from the corpus callosum reconstructed with ACT. The bottom row of Fig. 22 shows the tissue probability maps of the segmentation performed on a DPM. Example segmentations on T1 images Example segmentations on Difussion Power Maps (DPM) We developed a segmentation algorithm based on a Bayesian framework by using the MAP-MRF approach and EM. The algorithm was tested on T1-w as well as on DPMs [6] The tissue specific probability maps from both the T1-w and the DPMs were then used for ACT. We were able to successfully run ACT with the tissue probability maps derived from the DPMs. More information about this project can be found at: https://github.-com/villalonreina/dipy/tree/pve None. Julio E. Villalon-Reina and Eleftherios Garyfallidis performed the project and wrote the report. The authors would like to thank the organizers and attendees of the 2015 OHBM Hackathon. Julio E. Villalon-Reina was funded by Google Summer of Code 2015.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.002 | 0.004 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.040 | 0.034 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".