Background Deep neural networks (DNNs) are widely investigated in medical image classification to achieve automated support for clinical diagnosis. It is necessary to evaluate the robustness of medical DNN tasks against adversarial attacks, as high-stake decision-making will be made based on the diagnosis. Several previous studies have considered simple adversarial attacks. However, the vulnerability of DNNs to more realistic and higher risk attacks, such as universal adversarial perturbation (UAP), which is a single perturbation that can induce DNN failure in most classification tasks has not been evaluated yet. Methods We focus on three representative DNN-based medical image classification tasks (i.e., skin cancer, referable diabetic retinopathy, and pneumonia classifications) and investigate their vulnerability to the seven model architectures of UAPs. Results We demonstrate that DNNs are vulnerable to both nontargeted UAPs, which cause a task failure resulting in an input being assigned an incorrect class, and to targeted UAPs, which cause the DNN to classify an input into a specific class. The almost imperceptible UAPs achieved > 80% success rates for nontargeted and targeted attacks. The vulnerability to UAPs depended very little on the model architecture. Moreover, we discovered that adversarial retraining, which is known to be an effective method for adversarial defenses, increased DNNs’ robustness against UAPs in only very few cases. Conclusion Unlike previous assumptions, the results indicate that DNN-based clinical diagnosis is easier to deceive because of adversarial attacks. Adversaries can cause failed diagnoses at lower costs (e.g., without consideration of data distribution); moreover, they can affect the diagnosis. The effects of adversarial defenses may not be limited. Our findings emphasize that more careful consideration is required in developing DNNs for medical imaging and their practical applications.

Background. Deep neural networks (DNNs) are widely investigated in medical image classification to achieve automated support for clinical diagnosis. It is necessary to evaluate the robustness of medical DNN tasks against adversarial attacks, as high-stake decision making will be made based on the diagnosis. Several previous studies have considered simple adversarial attacks. However, the vulnerability of DNNs to more realistic and higher risk attacks have not been evaluated yet, i.e., universal adversarial perturbation (UAP), which is a single perturbation that can induce DNN failure in most classification tasks.Methods. We focus on three representative DNN-based medical image classification tasks (i.e., skin cancer, referable diabetic retinopathy, and pneumonia classifications) and investigate their vulnerability of DNNs with various model architectures to UAPs.Results. We demonstrate that the DNNs are vulnerable to both nontargeted UAPs, which cause a task failure resulting in an input being assigned an incorrect class, and to targeted UAPs, which cause the DNN to classify an input into a specific class. The almost imperceptible UAPs achieved > 80% success rates for nontargeted and targeted attacks. The vulnerability to UAPs barely depended on model architecture. Moreover, we discovered that adversarial retraining, which is known to be an effective method for adversarial defenses, increased the robustness of DNNs against UAPs in only limited cases. Conclusion. Unlike previous assumptions, the results indicate that DNN-based clinical diagnosis is easier to deceive because of adversarial attacks. Adversaries can result in failed diagnoses at lower costs (e.g., without consideration of data distribution); moreover, they can affect the diagnosis. The effects of adversarial defenses may be not limited. Our findings emphasize that more careful consideration is required in developing DNNs for medical imaging and their practical applications.

Background. Deep neural networks (DNNs) are widely investigated in medical image classification to achieve automated support for clinical diagnosis. It is necessary to evaluate the robustness of medical DNN tasks against adversarial attacks, as high-stake decision-making will be made based on the diagnosis. Several previous studies have considered simple adversarial attacks. However, the vulnerability of DNNs to more realistic and higher risk attacks, such as universal adversarial perturbation (UAP), which is a single perturbation that can induce DNN failure in most classification tasks has not been evaluated yet.Methods. We focus on three representative DNN-based medical image classification tasks (i.e., skin cancer, referable diabetic retinopathy, and pneumonia classifications) and investigate their vulnerability to the seven model architectures of UAPs.Results. We demonstrate that DNNs are vulnerable to both nontargeted UAPs, which cause a task failure resulting in an input being assigned an incorrect class, and to targeted UAPs, which cause the DNN to classify an input into a specific class. The almost imperceptible UAPs achieved > 80% success rates for nontargeted and targeted attacks. The vulnerability to UAPs depended very little on the model architecture. Moreover, we discovered that adversarial retraining, which is known to be an effective method for adversarial defenses, increased DNNs’ robustness against UAPs in only very few cases.Conclusion. Unlike previous assumptions, the results indicate that DNN-based clinical diagnosis is easier to deceive because of adversarial attacks. Adversaries can cause failed diagnoses at lower costs (e.g., without consideration of data distribution); moreover, they can affect the diagnosis. The effects of adversarial defenses may not be limited. Our findings emphasize that more careful consideration is required in developing DNNs for medical imaging and their practical applications.

Transfer learning from natural images is used in deep neural networks (DNNs) for medical image classification to achieve a computer-aided clinical diagnosis. Although the adversarial vulnerability of DNNs hinders practical applications owing to the high stakes of diagnosis, adversarial attacks are expected to be limited because training datasets (medical images), which are often required for adversarial attacks, are generally unavailable in terms of security and privacy preservation. Nevertheless, in this study, we demonstrated that adversarial attacks are also possible using natural images for medical DNN models with transfer learning, even if such medical images are unavailable; in particular, we showed that universal adversarial perturbations (UAPs) can also be generated from natural images. UAPs from natural images are useful for both non-targeted and targeted attacks. The performance of UAPs from natural images was significantly higher than that of random controls. The use of transfer learning causes a security hole, which decreases the reliability and safety of computer-based disease diagnosis. Model training from random initialization reduced the performance of UAPs from natural images; however, it did not completely avoid vulnerability to UAPs. The vulnerability of UAPs to natural images is expected to become a significant security threat.

Transfer learning from natural images is well used in deep neural networks (DNNs) for medical image classification to achieve computer-aided clinical diagnosis. Although the adversarial vulnerability of DNNs hinders practical applications owing to the high stakes of diagnosis, adversarial attacks are expected to be limited because training data — which are often required for adversarial attacks — are generally unavailable in terms of security and privacy preservation. Nevertheless, we hypothesized that adversarial attacks are also possible using natural images because pre-trained models do not change significantly after fine-tuning. We focused on three representative DNN-based medical image classification tasks (i.e., skin cancer, referable diabetic retinopathy, and pneumonia classifications) and investigated whether medical DNN models with transfer learning are vulnerable to universal adversarial perturbations (UAPs), generated using natural images. UAPs from natural images are useful for both non-targeted and targeted attacks. The performance of UAPs from natural images was significantly higher than that of random controls, although slightly lower than that of UAPs from training images. Vulnerability to UAPs from natural images was observed between different natural image datasets and between different model architectures. The use of transfer learning causes a security hole, which decreases the reliability and safety of computer-based disease diagnosis. Model training from random initialization (without transfer learning) reduced the performance of UAPs from natural images; however, it did not completely avoid vulnerability to UAPs. The vulnerability of UAPs from natural images will become a remarkable security threat.