Data compatibility across point probe Raman spectroscopy systems and its impact on statistical models portability (Conference Presentation)
Bibliographic record
Abstract
Raman spectroscopy is an optical technique that can assess a sample’s molecular content by probing its vibrational modes and has been used over the last decades to diagnose multiple types of cancer. The standard method used to build the classification models, based on machine learning algorithms, is the source of two majors limitations: the small size of the collected training datasets and the issue of portability of statistical models across imaging systems and medical centers. Model portability can be adressed by using a spectrum processing method that totally removes the hardware influence from the processed Raman measurements. We focus here on the results of two experiments conducted to evaluate the reproductibility of Raman measurements made with nine different point-probe systems. For the first experiment, we used a nylon phantom to assess inter-systems differences and applied the data processing method which lowered the inter-systems deviation for the processed nylon peaks under 3%. Furthermore, system #1 was used in vivo in a human brain surgery to acquire 15 Raman measurements from normal and tumor tissue. We evaluated the deviation between classes and found that it was superior to the 3% inter-systems reproductibility for 10 Raman peaks associated with proteins, lipids and nucleic acids. The second experiment was done with the system #1 as a master system and systems #2 to #9 as slave systems. The master system was used to build a Support Vector Machine classification model to discriminate white matter from grey matter on fixed ex vivo monkey brain slices. The model was exported from master to slaves performing a diagnosis accuracy consistently over 95%. The reported results indicate the possibility to succesfully export statistical model from one system to another and to greatly increase the size of dataset using multiple imaging systems.
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 | 0.035 | 0.160 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.004 | 0.004 |
| Open science | 0.002 | 0.003 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.004 | 0.001 |
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".