Deep learning architectures for spatial-frequency 3D fluorescence in oral cancer surgery models
Bibliographic record
Abstract
Fluorescence-guided surgery systems employed during oral cancer resection help detect the lateral margin yet fail to quantify the deep margins of the tumor prior to resection. Without comprehensive quantification of three-dimensional tumor margins, complete resection remains challenging. While interoperative techniques to assess the deep margin exist, they are limited in precision, leaving an unmet need for a system that can quantify depth. Our group is developing a deep learning (DL)-enabled fluorescence spatial frequency domain imaging (SFDI) system to address this limitation. The SFDI system captures fluorescence (F) and reflectance (R) images that contain information on tissue optical properties (OP) and depth sensitivity across spatial frequencies. Coupling DL with SFDI imaging allows for the near-real time construction of depth and concentration maps. Here, we compare three DL architectures that use SFDI images as inputs: i) F+OP, where OP (absorption and scattering) are obtained analytically from reflectance images; ii) F+R; iii) F/R. Training the three models required 10,000 tumor samples; synthetic tumors derived from composite spherical harmonics circumvented the need for patient data. The synthetic tumors were passed to a diffusion-theory light propagation model to generate a dataset of artificial SFDI images for DL training. Two oral cancer models derived from MRI of patient tongue tumors are used to evaluate DL performance in: i) in silico SFDI images ii) optical phantoms. These studies evaluate how system performance is affected by the SFDI input data and DL architectures. Future studies are required to assess system performance in vivo.
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) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
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".