Comparison of EfficientNet CNN models for multi-label chest X-ray disease diagnosis
Bibliographic record
Abstract
The analysis of chest X-ray images, which are critical for the early diagnosis of many diseases, is a difficult and time-consuming process due to the multiple labeling requirements and similar looking pathologies. In traditional methods, expert physicians analyze high-resolution chest X-ray images to diagnose these diseases using observational methods, a process that can lead to human error and hence misdiagnosis or underdiagnosis. In this study, we aim to autonomously detect 14 different diseases that significantly affect human health and some cases even lead to death using chest X-ray images in a multi-class manner using deep learning techniques. Previous studies on chest X-ray images focus on a single disease or have low success rates, and the architectures presented in previous studies have high computational costs. The novelty of this work is that it presents a hybrid lightweight, fast and attention-based architecture with high classification performance. In this study, we used the ChestX-Ray14 dataset consisting of 112,104 labeled chest X-ray images of 14 disease classes. Eight deep learning architectures (EfficientNetB0-B7) and coordinate attention mechanism are used in the training and testing processes. The proposed EfficientNetB7 architecture achieved an average overall classification performance with an AUC value of 0.8265. The EfficientNet enhanced with coordinate attention architecture achieved a classification success with an AUC value of 0.8309. Moreover, when the proposed architecture and the individual disease classes are considered separately, higher classification success is achieved for eight of the 14 diseases in the dataset. Finally, the results of this study outperformed the classification performance of other similar studies in the literature in terms of AUC score. The results obtained in our study show that the proposed deep learning based lightweight and fast architecture can support radiologists in decision making in disease diagnosis. The use of autonomous disease diagnosis systems can support the protection of human health by preventing incomplete or erroneous diagnoses.
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
Direct model labels (unvalidated)
Per-model category and study-design labels from the labeling rounds. They are machine output, unvalidated, and the disagreement between models ships as data. No study design here is MEDLINE-validated yet.
| Model arm | Categories | Study design | Confidence |
|---|---|---|---|
| gemma | no category Domain: not available · Genre: Empirical About the Canadian research system: no · About a Canadian topic: no | Simulation or modeling | low |
| gpt | no category Domain: not available · Genre: Empirical About the Canadian research system: no · About a Canadian topic: no | Simulation or modeling | high |
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 | 0.002 | 0.003 |
| Meta-epidemiology (narrow) | 0.002 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 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, unvalidatedLabeled directly by 2 models reading the full record.
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".