Mirror: Embedded AI Mirror for Thermal Tumor Screening and Remote Monitoring
Bibliographic record
Abstract
This paper presents mIRror, a modular intelligent mirror system designed to support autonomous tumor screening through thermal imaging in both clinical and remote settings.The system integrates embedded long-wave infrared (LWIR) sensing, real-time thermal preprocessing, lightweight convolutional neural inference, and onboard result visualization within a compact edge device.Its design addresses key challenges in accessibility, repeatability, and privacy, offering an alternative to conventional imaging methods in low-resource or decentralized environments.The clinical workflow is structured around standardized thermal acquisition protocols that ensure consistent patient positioning, controlled environmental conditions, and reproducible image capture.The processing chain encompasses thermal calibration, region-of-interest segmentation, and real-time classification executed locally on embedded hardware.This configuration enables low-latency inference without reliance on external servers, preserving data confidentiality and supporting future scalability.To validate the system's diagnostic module, six lightweight convolutional models-including MobileNetV2, MobileNetV3 (Small and Large), EfficientNetB0, EfficientNetV2B0, and NASNetMobile-were trained and evaluated using the publicly available DMR-IR dataset.Models were assessed using AUC-ROC, precision-recall metrics, and statistical significance testing via the Friedman test and Nemenyi post-hoc analysis.MobileNetV3-Large demonstrated superior performance (AUC = 0.99) with consistent interpretability through Smooth Grad-CAM++ visualizations.While hardware prototyping is ongoing, these results provide a proof-of-concept for mIRror's embedded diagnostic capabilities.The modular architecture is designed to support future extensions, including federated learning for secure collaborative training and digital twin integration for individualized monitoring.Together, these components position mIRror as a scalable platform for AI-assisted thermal diagnostics.
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 | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.000 |
| Insufficient payload (model declined to judge) | 0.005 | 0.002 |
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".