Should end-to-end deep learning replace handcrafted radiomics?
Bibliographic record
Abstract
learning.Representation learning involves training a neural network to learn a parsimonious yet comprehensive representation of the useful information present in the images.This information can then be used as latent deep radiomic features to build models corresponding to different classification or prediction tasks.Unlike handcrafted features, not only the values, but the definitions of the features involved in deep radiomics themselves depend on the training data and model architecture used to train a classification or prediction model.While deep learning is gaining ground thanks to the increasing availability of open datasets and shared deep learning models (https://nmmitools.org,https://monai.io),one might ask whether researchers should focus efforts exclusively on deep learning rather than radiomics.This short paper examines the respective positions of handcrafted and end-to-end deep radiomics in relation to key factors to be considered when developing and implementing classification or outcome prediction models.It has been inspired by the content of a moderated debate held during the SNMMI 2023 meeting involving two advocates of handcrafted radiomics and two advocates of deep radiomics.
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 | 0.023 | 0.096 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.008 |
| Scholarly communication | 0.004 | 0.009 |
| Open science | 0.002 | 0.003 |
| Research integrity | 0.025 | 0.039 |
| Insufficient payload (model declined to judge) | 0.005 | 0.010 |
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".