Mask of truth: model sensitivity to unexpected regions of medical images

The development of larger models for medical image analysis has led to increased performance. However, it also affected our ability to explain and validate model decisions. Models can use non-relevant parts of images, also called spurious correlations or shortcuts, to obtain high performance on benchmark datasets but fail in real-world scenarios. In this work, we challenge the capacity of convolutional neural networks (CNN) to classify chest X-rays and eye fundus images while masking out clinically relevant parts of the image. We show that all models trained on the PadChest dataset, irrespective of the masking strategy, are able to obtain an Area Under the Curve (AUC) above random. Moreover, the models trained on full images obtain good performance on images without the region of interest (ROI), even superior to the one obtained on images only containing the ROI. We also reveal a possible spurious correlation in the Chaksu dataset while the performances are more aligned with the expectation of an unbiased model. We go beyond the performance analysis with the usage of the explainability method SHAP and the analysis of embeddings. We asked a radiology resident to interpret chest X-rays under different masking to complement our findings with clinical knowledge. Our code is available at https://github.com/TheoSourget/MMC_Masking and https://github.com/TheoSourget/MMC_Masking_EyeFundus

Source Matters: Source Dataset Impact on Model Robustness in Medical Imaging

Transfer learning has become an essential part of medical imaging classification algorithms, often leveraging ImageNet weights. However, the domain shift from natural to medical images has prompted alternatives such as RadImageNet, often demonstrating comparable classification performance. However, it remains unclear whether the performance gains from transfer learning stem from improved generalization or shortcut learning. To address this, we investigate potential confounders -- whether synthetic or sampled from the data -- across two publicly available chest X-ray and CT datasets. We show that ImageNet and RadImageNet achieve comparable classification performance, yet ImageNet is much more prone to overfitting to confounders. We recommend that researchers using ImageNet-pretrained models reexamine their model robustness by conducting similar experiments. Our code and experiments are available at https://github.com/DovileDo/source-matters.

Towards actionability for open medical imaging datasets: lessons from community-contributed platforms for data management and stewardship

Medical imaging datasets are fundamental to artificial intelligence (AI) in healthcare. The accuracy, robustness and fairness of diagnostic algorithms depend on the data (and its quality) on which the models are trained and evaluated. Medical imaging datasets have become increasingly available to the public, and are often hosted on Community-Contributed Platforms (CCP), including private companies like Kaggle or HuggingFace. While open data is important to enhance the redistribution of data's public value, we find that the current CCP governance model fails to uphold the quality needed and recommended practices for sharing, documenting, and evaluating datasets. In this paper we investigate medical imaging datasets on CCPs and how they are documented, shared, and maintained. We first highlight some differences between medical imaging and computer vision, particularly in the potentially harmful downstream effects due to poor adoption of recommended dataset management practices. We then analyze 20 (10 medical and 10 computer vision) popular datasets on CCPs and find vague licenses, lack of persistent identifiers and storage, duplicates and missing metadata, with differences between the platforms. We present "actionability" as a conceptual metric to reveal the data quality gap between characteristics of data on CCPs and the desired characteristics of data for AI in healthcare. Finally, we propose a commons-based stewardship model for documenting, sharing and maintaining datasets on CCPs and end with a discussion of limitations and open questions.

Data usage and citation practices in medical imaging conferences

Medical imaging papers often focus on methodology, but the quality of the algorithms and the validity of the conclusions are highly dependent on the datasets used. As creating datasets requires a lot of effort, researchers often use publicly available datasets, there is however no adopted standard for citing the datasets used in scientific papers, leading to difficulty in tracking dataset usage. In this work, we present two open-source tools we created that could help with the detection of dataset usage, a pipeline \url{https://github.com/TheoSourget/Public_Medical_Datasets_References} using OpenAlex and full-text analysis, and a PDF annotation software \url{https://github.com/TheoSourget/pdf_annotator} used in our study to manually label the presence of datasets. We applied both tools on a study of the usage of 20 publicly available medical datasets in papers from MICCAI and MIDL. We compute the proportion and the evolution between 2013 and 2023 of 3 types of presence in a paper: cited, mentioned in the full text, cited and mentioned. Our findings demonstrate the concentration of the usage of a limited set of datasets. We also highlight different citing practices, making the automation of tracking difficult.

Augmenting Chest X-ray Datasets with Non-Expert Annotations

The advancement of machine learning algorithms in medical image analysis requires the expansion of training datasets. A popular and cost-effective approach is automated annotation extraction from free-text medical reports, primarily due to the high costs associated with expert clinicians annotating chest X-ray images. However, it has been shown that the resulting datasets are susceptible to biases and shortcuts. Another strategy to increase the size of a dataset is crowdsourcing, a widely adopted practice in general computer vision with some success in medical image analysis. In a similar vein to crowdsourcing, we enhance two publicly available chest X-ray datasets by incorporating non-expert annotations. However, instead of using diagnostic labels, we annotate shortcuts in the form of tubes. We collect 3.5k chest drain annotations for CXR14, and 1k annotations for 4 different tube types in PadChest. We train a chest drain detector with the non-expert annotations that generalizes well to expert labels. Moreover, we compare our annotations to those provided by experts and show "moderate" to "almost perfect" agreement. Finally, we present a pathology agreement study to raise awareness about ground truth annotations. We make our annotations and code available.

Detecting Shortcuts in Medical Images -- A Case Study in Chest X-rays

The availability of large public datasets and the increased amount of computing power have shifted the interest of the medical community to high-performance algorithms. However, little attention is paid to the quality of the data and their annotations. High performance on benchmark datasets may be reported without considering possible shortcuts or artifacts in the data, besides, models are not tested on subpopulation groups. With this work, we aim to raise awareness about shortcuts problems. We validate previous findings, and present a case study on chest X-rays using two publicly available datasets. We share annotations for a subset of pneumothorax images with drains. We conclude with general recommendations for medical image classification.

Memory-aware curriculum federated learning for breast cancer classification

For early breast cancer detection, regular screening with mammography imaging is recommended. Routinary examinations result in datasets with a predominant amount of negative samples. A potential solution to such class-imbalance is joining forces across multiple institutions. Developing a collaborative computer-aided diagnosis system is challenging in different ways. Patient privacy and regulations need to be carefully respected. Data across institutions may be acquired from different devices or imaging protocols, leading to heterogeneous non-IID data. Also, for learning-based methods, new optimization strategies working on distributed data are required. Recently, federated learning has emerged as an effective tool for collaborative learning. In this setting, local models perform computation on their private data to update the global model. The order and the frequency of local updates influence the final global model. Hence, the order in which samples are locally presented to the optimizers plays an important role. In this work, we define a memory-aware curriculum learning method for the federated setting. Our curriculum controls the order of the training samples paying special attention to those that are forgotten after the deployment of the global model. Our approach is combined with unsupervised domain adaptation to deal with domain shift while preserving data privacy. We evaluate our method with three clinical datasets from different vendors. Our results verify the effectiveness of federated adversarial learning for the multi-site breast cancer classification. Moreover, we show that our proposed memory-aware curriculum method is beneficial to further improve classification performance. Our code is publicly available at: https://github.com/ameliajimenez/curriculum-federated-learning.

Curriculum learning for annotation-efficient medical image analysis: scheduling data with prior knowledge and uncertainty

Convolutional neural networks (CNNs) for multi-class classification require training on large, representative, and high quality annotated datasets. However, in the field of medical imaging, data and annotations are both difficult and expensive to acquire. Moreover, they frequently suffer from highly imbalanced distributions, and potentially noisy labels due to intra- or inter-expert disagreement. To deal with such challenges, we propose a unified curriculum learning framework to schedule the order and pace of the training samples presented to the optimizer. Our novel framework reunites three strategies consisting of individually weighting training samples, reordering the training set, or sampling subsets of data. The core of these strategies is a scoring function ranking the training samples according to either difficulty or uncertainty. We define the scoring function from domain-specific prior knowledge or by directly measuring the uncertainty in the predictions. We perform a variety of experiments with a clinical dataset for the multi-class classification of proximal femur fractures and the publicly available MNIST dataset. Our results show that the sequence and weight of the training samples play an important role in the optimization process of CNNs. Proximal femur fracture classification is improved up to the performance of experienced trauma surgeons. We further demonstrate the benefits of our unified curriculum learning method for three controlled and challenging digit recognition scenarios: with limited amounts of data, under class-imbalance, and in the presence of label noise.

Medical-based Deep Curriculum Learning for Improved Fracture Classification

Current deep-learning based methods do not easily integrate to clinical protocols, neither take full advantage of medical knowledge. In this work, we propose and compare several strategies relying on curriculum learning, to support the classification of proximal femur fracture from X-ray images, a challenging problem as reflected by existing intra- and inter-expert disagreement. Our strategies are derived from knowledge such as medical decision trees and inconsistencies in the annotations of multiple experts, which allows us to assign a degree of difficulty to each training sample. We demonstrate that if we start learning "easy" examples and move towards "hard", the model can reach a better performance, even with fewer data. The evaluation is performed on the classification of a clinical dataset of about 1000 X-ray images. Our results show that, compared to class-uniform and random strategies, the proposed medical knowledge-based curriculum, performs up to 15% better in terms of accuracy, achieving the performance of experienced trauma surgeons.

Towards an Interactive and Interpretable CAD System to Support Proximal Femur Fracture Classification

Fractures of the proximal femur represent a critical entity in the western world, particularly with the growing elderly population. Such fractures result in high morbidity and mortality, reflecting a significant health and economic impact on our society. Different treatment strategies are recommended for different fracture types, with surgical treatment still being the gold standard in most of the cases. The success of the treatment and prognosis after surgery strongly depends on an accurate classification of the fracture among standard types, such as those defined by the AO system. However, the classification of fracture types based on x-ray images is difficult as confirmed by low intra- and inter-expert agreement rates of our in-house study and also in the previous literature. The presented work proposes a fully automatic computer-aided diagnosis (CAD) tool, based on current deep learning techniques, able to identify, localize and finally classify proximal femur fractures on x-rays images according to the AO classification. Results of our experimental evaluation show that the performance achieved by the proposed CAD tool is comparable to the average expert for the classification of x-ray images into types ''A'', ''B'' and ''normal'' (precision of 89%), while the performance is even superior when classifying fractures versus ''normal'' cases (precision of 94%). In addition, the integration of the proposed CAD tool into daily clinical routine is extensively discussed, towards improving the interface between humans and AI-powered machines in supporting medical decisions.