Attention-based Saliency Maps Improve Interpretability of Pneumothorax Classification

Wollek, Alessandro; Graf, R.; Čečatka, Saša; Fink, Nicola; Willem, Theresa; Sabel, Bastian O.; Lasser, Tobias

doi:10.1148/ryai.220187

Cited by 13 publications

(11 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Instead, including this "real-world" data not only in the training process, but also into the model validation, will lead to more robust ML models and ultimately improve clinical acceptance One limitation of this work is that we use the CheXnet model as a representative for other chest x-ray classification models. While a DenseNet-121 is still considered as state-of -the-art for chest x-ray classification, 11,29 further research is necessary to determine if our findings translate to other architectures. Furthermore, we only tested our model on chest x-ray images, even though our approach remains relevant to all multi-label OOD detection data sets.…”

Section: Discussionmentioning

confidence: 93%

“…3 Considering the increasing demand for imaging, while the number of radiologists remains insufficient, the aforementioned and similar models can help improve medical patient care, for example, by screening acquired radiographs for critical findings prior to radiologist interpretation. [6][7][8][9][10][11] Then, patients with time sensitive illnesses will receive treatment earlier, potentially saving their lives. What all of these chest x-ray classifiers have seen, once trained, validated and tested, are chest x-rays of a certain type, the in-distribution (ID) images.…”

Section: Introductionmentioning

confidence: 99%

“…[13][14][15] In our scenario, prioritization based on false, highconfidence, OOD x-rays can lead to longer waiting times for other patients with time critical conditions, like a pneumothorax, potentially risking their life until the error is discovered and resolved. Moreover, repeated instances of such misreporting will -if not balanced with transparency measures sufficiently, for example, using saliency maps 11 -quickly lead physicians to distrust the model, eventually leading them to refrain from using it. [16][17][18] Therefore, in recent years, several methods, have been proposed to detect OOD samples.…”

Section: Introductionmentioning

confidence: 99%

“…Due to the early and profound digitization of imaging techniques, machine learning in radiology can already show convincing successes, such as the detection of certain critical pathologies of the lung on x‐ray images with performance non‐inferior to radiologists 3 . Considering the increasing demand for imaging, while the number of radiologists remains insufficient, the aforementioned and similar models can help improve medical patient care, for example, by screening acquired radiographs for critical findings prior to radiologist interpretation 6–11 . Then, patients with time sensitive illnesses will receive treatment earlier, potentially saving their lives.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Out‐of‐distribution detection with in‐distribution voting using the medical example of chest x‐ray classification

Wollek,

Willem,

Ingrisch

et al. 2023

Medical Physics

Self Cite

View full text Add to dashboard Cite

BackgroundDeep learning models are being applied to more and more use cases with astonishing success stories, but how do they perform in the real world? Models are typically tested on specific cleaned data sets, but when deployed in the real world, the model will encounter unexpected, out‐of‐distribution (OOD) data.PurposeTo investigate the impact of OOD radiographs on existing chest x‐ray classification models and to increase their robustness against OOD data.MethodsThe study employed the commonly used chest x‐ray classification model, CheXnet, trained on the chest x‐ray 14 data set, and tested its robustness against OOD data using three public radiography data sets: IRMA, Bone Age, and MURA, and the ImageNet data set. To detect OOD data for multi‐label classification, we proposed in‐distribution voting (IDV). The OOD detection performance is measured across data sets using the area under the receiver operating characteristic curve (AUC) analysis and compared with Mahalanobis‐based OOD detection, MaxLogit, MaxEnergy, self‐supervised OOD detection (SS OOD), and CutMix.ResultsWithout additional OOD detection, the chest x‐ray classifier failed to discard any OOD images, with an AUC of 0.5. The proposed IDV approach trained on ID (chest x‐ray 14) and OOD data (IRMA and ImageNet) achieved, on average, 0.999 OOD AUC across the three data sets, surpassing all other OOD detection methods. Mahalanobis‐based OOD detection achieved an average OOD detection AUC of 0.982. IDV trained solely with a few thousand ImageNet images had an AUC 0.913, which was considerably higher than MaxLogit (0.726), MaxEnergy (0.724), SS OOD (0.476), and CutMix (0.376).ConclusionsThe performance of all tested OOD detection methods did not translate well to radiography data sets, except Mahalanobis‐based OOD detection and the proposed IDV method. Consequently, training solely on ID data led to incorrect classification of OOD images as ID, resulting in increased false positive rates. IDV substantially improved the model's ID classification performance, even when trained with data that will not occur in the intended use case or test set (ImageNet), without additional inference overhead or performance decrease in the target classification. The corresponding code is available at https://gitlab.lrz.de/IP/a‐knee‐cannot‐have‐lung‐disease.

show abstract

Section: Discussionmentioning

confidence: 93%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Out‐of‐distribution detection with in‐distribution voting using the medical example of chest x‐ray classification

Wollek,

Willem,

Ingrisch

et al. 2023

Medical Physics

Self Cite

View full text Add to dashboard Cite

show abstract

“…Similarly, the release of the chest X-ray 14 data set [12] sparked the development of chest X-ray classification models like CheXnet [13]. While the number of images used by modern deep learning architectures has in-creased over the years, large publicly available data sets required for new architectures such as Vision Transformers [14] are missing in radiology, thus limiting the use of advanced models [15] and inhibiting advances in automated chest X-ray diagnosis.…”

Section: Introductionmentioning

confidence: 99%

German CheXpert Chest X-ray Radiology Report Labeler

Wollek,

Hyska,

Sedlmeyr

et al. 2024

Rofo

Self Cite

View full text Add to dashboard Cite

Purpose The aim of this study was to develop an algorithm to automatically extract annotations from German thoracic radiology reports to train deep learning-based chest X-ray classification models. Materials and Methods An automatic label extraction model for German thoracic radiology reports was designed based on the CheXpert architecture. The algorithm can extract labels for twelve common chest pathologies, the presence of support devices, and “no finding”. For iterative improvements and to generate a ground truth, a web-based multi-reader annotation interface was created. With the proposed annotation interface, a radiologist annotated 1086 retrospectively collected radiology reports from 2020–2021 (data set 1). The effect of automatically extracted labels on chest radiograph classification performance was evaluated on an additional, in-house pneumothorax data set (data set 2), containing 6434 chest radiographs with corresponding reports, by comparing a DenseNet-121 model trained on extracted labels from the associated reports, image-based pneumothorax labels, and publicly available data, respectively. Results Comparing automated to manual labeling on data set 1: “mention extraction” class-wise F1 scores ranged from 0.8 to 0.995, the “negation detection” F1 scores from 0.624 to 0.981, and F1 scores for “uncertainty detection” from 0.353 to 0.725. Extracted pneumothorax labels on data set 2 had a sensitivity of 0.997 [95 % CI: 0.994, 0.999] and specificity of 0.991 [95 % CI: 0.988, 0.994]. The model trained on publicly available data achieved an area under the receiver operating curve (AUC) for pneumothorax classification of 0.728 [95 % CI: 0.694, 0.760], while the models trained on automatically extracted labels and on manual annotations achieved values of 0.858 [95 % CI: 0.832, 0.882] and 0.934 [95 % CI: 0.918, 0.949], respectively. Conclusion Automatic label extraction from German thoracic radiology reports is a promising substitute for manual labeling. By reducing the time required for data annotation, larger training data sets can be created, resulting in improved overall modeling performance. Our results demonstrated that a pneumothorax classifier trained on automatically extracted labels strongly outperformed the model trained on publicly available data, without the need for additional annotation time and performed competitively compared to manually labeled data. Key points:

show abstract

Is Visual Explanation with Grad-CAM More Reliable for Deeper Neural Networks? A Case Study with Automatic Pneumothorax Diagnosis

Qiu,

Rivaz,

Xiao

2023

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Attention-based Saliency Maps Improve Interpretability of Pneumothorax Classification

Cited by 13 publications

References 23 publications

Out‐of‐distribution detection with in‐distribution voting using the medical example of chest x‐ray classification

Out‐of‐distribution detection with in‐distribution voting using the medical example of chest x‐ray classification

German CheXpert Chest X-ray Radiology Report Labeler

Is Visual Explanation with Grad-CAM More Reliable for Deeper Neural Networks? A Case Study with Automatic Pneumothorax Diagnosis

Contact Info

Product

Resources

About