MétaCan
Menu
← Back to cohort
Record W7055726336

Détection automatique des complications suivant les chirurgies orthopédiques des membres inférieurs par imagerie thermique

2024· other· fr· W7055726336 on OpenAlexfundno aff

Bibliographic record

VenuePolyPublie (École Polytechnique de Montréal) · 2024
Typeother
Languagefr
FieldEngineering
TopicMagnetic Field Sensors Techniques
Canadian institutionsnot available
FundersAlliance de recherche numérique du CanadaNatural Sciences and Engineering Research Council of CanadaGovernment of Canada
KeywordsVideo recordingStatistical analysisAgrégation
DOInot available

Abstract

fetched live from OpenAlex

RÉSUMÉ: On sait, depuis de nombreuses années, que la température surfacique du corps humain est un indicateur clé de son état de santé. L’imagerie thermique se veut un outil efficace pour visualiser les distributions d’intensités sur la peau, ouvrant de nouvelles possibilités de diagnostic. De plus, la symétrie bilatérale des températures sur les membres opposées peut être utilisée pour détecter certaines maladies. Avec l’intérêt grandissant pour les technologies de diagnostic assistées par ordinateur et l’intégration de l’apprentissage automatique profond, une modélisation efficace de ces distributions de température serait souhaitable. En apprenant les représentations de texture et en comparant leurs asymétries, nous émettons l’hypothèse que les vecteurs latents de texture entre les thermographies des parties du corps opposées permettent de décrire efficacement les différences de température qui peuvent signaler une anomalie dans le domaine thermique. Le modèle développé, baptisé TADA-SAE (Texture Anomaly Detection using Swapping Autoencoder), est testé sur l’ensemble de données privées OrthoPOT (Orthopedics Post-Operative Thermograms) pour la détection de complications post-chirurgicales suivant une arthroplastie totale du genou ou de la hanche et sur l’ensemble public DMR-IR (Database for Mastology Research with Infrared Image) pour la détection du cancer du sein. Un algorithme de forêt d’isolation iForest entraîné sur les attributs de symétrie entre les textures et des données cliniques atteint un score AUROC de 0.842 sur OrthoPOT et un score AUROC compétitif avec l’état de l’art de 0.988 sur DMR-IR tout en utilisant moins de paramètres que les méthodes traditionnelles. Ainsi, les attributs de texture et leurs différences entre les parties du corps opposées sont pertinentes pour la détection d’anomalies médicales dans le domaine thermique. Notre méthode est la première à utiliser des attributs de textures appris de manière auto-supervisée et leurs asymétries pour la détection d’anomalies médicales avec l’imagerie thermique, démontrant des résultats prometteurs qui pourraient ouvrir la voie à d’autres tâches dans le domaine médical où la symétrie bilatérale peut être exploitée. ABSTRACT: It has been known for many years that the surface temperature of the human body is a key indicator of its health status. Thermal imaging is an effective tool to visualize intensity distributions on the skin, opening new diagnostic possibilities. In addition, the bilateral symmetry of temperatures on opposite limbs can be used to detect certain diseases. With the growing interest in computer-aided diagnostic technologies and the integration of deep machine learning, an efficient modeling of these temperature distributions would be desirable. By learning texture representations and comparing their asymmetries, we hypothesize that the latent texture vectors between thermographs of opposite body parts can effectively describe temperature differences that may signal an anomaly in the thermal domain. The developed model, named TADA-SAE (Texture Anomaly Detection using Swapping Autoencoder), is tested on the private OrthoPOT (Orthopedics Post-Operative Thermograms) dataset for the detection of post-surgical complications following total knee or hip arthroplasty and on the public DMR-IR (Database for Mastology Research with Infrared Image) dataset for the detection of breast cancer. An isolation forest algorithm iForest trained on the symmetry attributes between textures and clinical data achieves 0.842 AUROC on OrthoPOT and a competitive 0.988 AUROC score on DMR-IR while using fewer parameters than traditional methods. Thus, texture attributes and their differences between opposite body parts are relevant for the detection of medical anomalies in the thermal domain. Our method is the first to use self-supervised learned texture attributes and their asymmetries for medical anomaly detection with thermal imaging, demonstrating promising results that could pave the way for other tasks in the medical domain where bilateral symmetry can be exploited.

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 imitation

Not 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.

metaresearch head score (Codex)0.000
metaresearch head score (Gemma)0.003
Version: metacan-v3-hybrid-931329e0061cValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Bench or experimental · Consensus signal: none
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.007
Threshold uncertainty score0.013

Distilled classifier scores by category (both heads)

CategoryCodexGemma
Metaresearch0.0000.003
Meta-epidemiology (narrow)0.0010.000
Meta-epidemiology (broad)0.0010.001
Bibliometrics0.0010.001
Science and technology studies0.0000.000
Scholarly communication0.0010.000
Open science0.0000.000
Research integrity0.0010.000
Insufficient payload (model declined to judge)0.0010.001

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.

Opus teacher head0.012
GPT teacher head0.240
Teacher spread0.227 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.

The models applied no category: nothing in the taxonomy fit this work.
Study designBench or experimental
Domainnot available
GenreEmpirical

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".

Quick stats

Citations0
Published2024
Admission routes1
Has abstractyes

Explore more

Same venuePolyPublie (École Polytechnique de Montréal)→Same topicMagnetic Field Sensors Techniques→French-language works237,207→