Search for: All records

Alzheimer’s disease (AD) is the most prevalent neurodegenerative disease, yet its current treatments are limited to stopping disease progression. Moreover, the effectiveness of these treatments remains uncertain due to the heterogeneity of the disease. Therefore, it is essential to identify disease subtypes at a very early stage. Current data-driven approaches can be used to classify subtypes during later stages of AD or related disorders, but making predictions in the asymptomatic or prodromal stage is challenging. Furthermore, the classifications of most existing models lack explainability, and these models rely solely on a single modality for assessment, limiting the scope of their analysis. Thus, we propose a multimodal framework that utilizes early-stage indicators, including imaging, genetics, and clinical assessments, to classify AD patients into progression-specific subtypes at an early stage. In our framework, we introduce a tri-modal co-attention mechanism (Tri-COAT) to explicitly capture cross-modal feature associations. Data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) (slow progressing = 177, intermediate = 302, and fast = 15) were used to train and evaluate Tri-COAT using a 10-fold stratified cross-testing approach. Our proposed model outperforms baseline models and sheds light on essential associations across multimodal features supported by known biological mechanisms. The multimodal design behind Tri-COAT allows it to achieve the highest classification area under the receiver operating characteristic curve while simultaneously providing interpretability to the model predictions through the co-attention mechanism. 

Purpose: To determine if saliency maps in radiology artificial intelligence (AI) are vulnerable to subtle perturbations of the input, which could potentially lead to misleading interpretations, using Prediction-Saliency Correlation (PSC) for evaluating the sensitivity and robustness of saliency methods. Materials and Methods: In this retrospective study, locally trained deep learning models and a research prototype provided by a commercial vender were systematically evaluated on 191,229 chest radiographs from the CheXpert dataset(1,2) and 7,022 MRI images of human brain tumor classification dataset(3). Two radiologists performed a reader study on 270 chest radiographs pairs. A model-agnostic approach for computing the PSC coefficient was used to evaluate the sensitivity and robustness of seven commonly used saliency methods. Results: Leveraging locally trained model parameters, we revealed the saliency methods’ low sensitivity (maximum PSC = 0.25, 95% CI: 0.12, 0.38) and weak robustness (maximum PSC = 0.12, 95% CI: 0.0, 0.25) on the CheXpert dataset. Without model specifics, we also showed that the saliency maps from a commercial prototype could be irrelevant to the model output (area under the receiver operating characteristic curve dropped by 8.6% without affecting the saliency map). The human observer studies confirmed that is difficult for experts to identify the perturbed images, who had less than 44.8% correctness. Conclusion: Popular saliency methods scored low PSC values on the two datasets of perturbed chest radiographs, indicating weak sensitivity and robustness. The proposed PSC metric provides a valuable quantification tool for validating the trustworthiness of medical AI explainability. Abbreviations: AI = artificial intelligence, PSC = prediction-saliency correlation, AUC = area under the receiver operating characteristic curve, SSIM = structural similarity index measure. Summary: Systematic evaluation of saliency methods through subtle perturbations in chest radiographs and brain MRI images demonstrated low sensitivity and robustness of those methods, warranting caution when using saliency methods that may misrepresent changes in AI model prediction. 

Domain generalization (DG) aims to train a model to perform well in unseen domains under different distributions. This paper considers a more realistic yet more challenging scenario, namely Single Domain Generalization (Single-DG), where only a single source domain is available for training. To tackle this challenge, we first try to understand when neural networks fail to generalize? We empirically ascertain a property of a model that correlates strongly with its generalization that we coin as model sensitivity. Based on our analysis, we propose a novel strategy of Spectral Adversarial Data Augmentation (SADA) to generate augmented images targeted at the highly sensitive frequencies. Models trained with these hard-to-learn samples can effectively suppress the sensitivity in the frequency space, which leads to improved generalization performance. Extensive experiments on multiple public datasets demonstrate the superiority of our approach, which surpasses the state-of-the-art single-DG methods by up to 2.55%. The source code is available at https://github.com/DIAL-RPI/Spectral-Adversarial-Data-Augmentation.