What limits performance of weakly supervised deep learning for chest CT classification?

Weakly supervised learning with noisy data has drawn attention in the medical imaging community due to the sparsity of high-quality disease labels. However, little is known about the limitations of such weakly supervised learning and the effect of these constraints on disease classification performance. In this paper, we test the effects of such weak supervision by examining model tolerance for three conditions. First, we examined model tolerance for noisy data by incrementally increasing error in the labels within the training data. Second, we assessed the impact of dataset size by varying the amount of training data. Third, we compared performance differences between binary and multi-label classification. Results demonstrated that the model could endure up to 10% added label error before experiencing a decline in disease classification performance. Disease classification performance steadily rose as the amount of training data was increased for all disease classes, before experiencing a plateau in performance at 75% of training data. Last, the binary model outperformed the multilabel model in every disease category. However, such interpretations may be misleading, as the binary model was heavily influenced by co-occurring diseases and may not have learned the specific features of the disease in the image. In conclusion, this study may help the medical imaging community understand the benefits and risks of weak supervision with noisy labels. Such studies demonstrate the need to build diverse, large-scale datasets and to develop explainable and responsible AI.

Large Intestine 3D Shape Refinement Using Point Diffusion Models for Digital Phantom Generation

Accurate 3D modeling of human organs plays a crucial role in building computational phantoms for virtual imaging trials. However, generating anatomically plausible reconstructions of organ surfaces from computed tomography scans remains challenging for many structures in the human body. This challenge is particularly evident when dealing with the large intestine. In this study, we leverage recent advancements in geometric deep learning and denoising diffusion probabilistic models to refine the segmentation results of the large intestine. We begin by representing the organ as point clouds sampled from the surface of the 3D segmentation mask. Subsequently, we employ a hierarchical variational autoencoder to obtain global and local latent representations of the organ's shape. We train two conditional denoising diffusion models in the hierarchical latent space to perform shape refinement. To further enhance our method, we incorporate a state-of-the-art surface reconstruction model, allowing us to generate smooth meshes from the obtained complete point clouds. Experimental results demonstrate the effectiveness of our approach in capturing both the global distribution of the organ's shape and its fine details. Our complete refinement pipeline demonstrates remarkable enhancements in surface representation compared to the initial segmentation, reducing the Chamfer distance by 70%, the Hausdorff distance by 32%, and the Earth Mover's distance by 6%. By combining geometric deep learning, denoising diffusion models, and advanced surface reconstruction techniques, our proposed method offers a promising solution for accurately modeling the large intestine's surface and can easily be extended to other anatomical structures.

Data diversity and virtual imaging in AI-based diagnosis: A case study based on COVID-19

Many studies have investigated deep-learning-based artificial intelligence (AI) models for medical imaging diagnosis of the novel coronavirus (COVID-19), with many reports of near-perfect performance. However, variability in performance and underlying data biases raise concerns about clinical generalizability. This retrospective study involved the development and evaluation of artificial intelligence (AI) models for COVID-19 diagnosis using both diverse clinical and virtually generated medical images. In addition, we conducted a virtual imaging trial to assess how AI performance is affected by several patient- and physics-based factors, including the extent of disease, radiation dose, and imaging modality of computed tomography (CT) and chest radiography (CXR). AI performance was strongly influenced by dataset characteristics including quantity, diversity, and prevalence, leading to poor generalization with up to 20% drop in receiver operating characteristic area under the curve. Model performance on virtual CT and CXR images was comparable to overall results on clinical data. Imaging dose proved to have negligible influence on the results, but the extent of the disease had a marked affect. CT results were consistently superior to those from CXR. Overall, the study highlighted the significant impact of dataset characteristics and disease extent on COVID assessment, and the relevance and potential role of virtual imaging trial techniques on developing effective evaluation of AI algorithms and facilitating translation into diagnostic practice.

Virtual vs. Reality: External Validation of COVID-19 Classifiers using XCAT Phantoms for Chest Computed Tomography

Research studies of artificial intelligence models in medical imaging have been hampered by poor generalization. This problem has been especially concerning over the last year with numerous applications of deep learning for COVID-19 diagnosis. Virtual imaging trials (VITs) could provide a solution for objective evaluation of these models. In this work utilizing the VITs, we created the CVIT-COVID dataset including 180 virtually imaged computed tomography (CT) images from simulated COVID-19 and normal phantom models under different COVID-19 morphology and imaging properties. We evaluated the performance of an open-source, deep-learning model from the University of Waterloo trained with multi-institutional data and an in-house model trained with the open clinical dataset called MosMed. We further validated the model's performance against open clinical data of 305 CT images to understand virtual vs. real clinical data performance. The open-source model was published with nearly perfect performance on the original Waterloo dataset but showed a consistent performance drop in external testing on another clinical dataset (AUC=0.77) and our simulated CVIT-COVID dataset (AUC=0.55). The in-house model achieved an AUC of 0.87 while testing on the internal test set (MosMed test set). However, performance dropped to an AUC of 0.65 and 0.69 when evaluated on clinical and our simulated CVIT-COVID dataset. The VIT framework offered control over imaging conditions, allowing us to show there was no change in performance as CT exposure was changed from 28.5 to 57 mAs. The VIT framework also provided voxel-level ground truth, revealing that performance of in-house model was much higher at AUC=0.87 for diffuse COVID-19 infection size >2.65% lung volume versus AUC=0.52 for focal disease with <2.65% volume. The virtual imaging framework enabled these uniquely rigorous analyses of model performance.

Quality or Quantity: Toward a Unified Approach for Multi-organ Segmentation in Body CT

Organ segmentation of medical images is a key step in virtual imaging trials. However, organ segmentation datasets are limited in terms of quality (because labels cover only a few organs) and quantity (since case numbers are limited). In this study, we explored the tradeoffs between quality and quantity. Our goal is to create a unified approach for multi-organ segmentation of body CT, which will facilitate the creation of large numbers of accurate virtual phantoms. Initially, we compared two segmentation architectures, 3D-Unet and DenseVNet, which were trained using XCAT data that is fully labeled with 22 organs, and chose the 3D-Unet as the better performing model. We used the XCAT-trained model to generate pseudo-labels for the CT-ORG dataset that has only 7 organs segmented. We performed two experiments: First, we trained 3D-UNet model on the XCAT dataset, representing quality data, and tested it on both XCAT and CT-ORG datasets. Second, we trained 3D-UNet after including the CT-ORG dataset into the training set to have more quantity. Performance improved for segmentation in the organs where we have true labels in both datasets and degraded when relying on pseudo-labels. When organs were labeled in both datasets, Exp-2 improved Average DSC in XCAT and CT-ORG by 1. This demonstrates that quality data is the key to improving the model's performance.

Co-occurring Diseases Heavily Influence the Performance of Weakly Supervised Learning Models for Classification of Chest CT

Despite the potential of weakly supervised learning to automatically annotate massive amounts of data, little is known about its limitations for use in computer-aided diagnosis (CAD). For CT specifically, interpreting the performance of CAD algorithms can be challenging given the large number of co-occurring diseases. This paper examines the effect of co-occurring diseases when training classification models by weakly supervised learning, specifically by comparing multi-label and multiple binary classifiers using the same training data. Our results demonstrated that the binary model outperformed the multi-label classification in every disease category in terms of AUC. However, this performance was heavily influenced by co-occurring diseases in the binary model, suggesting it did not always learn the correct appearance of the specific disease. For example, binary classification of lung nodules resulted in an AUC of < 0.65 when there were no other co-occurring diseases, but when lung nodules co-occurred with emphysema, the performance reached AUC > 0.80. We hope this paper revealed the complexity of interpreting disease classification performance in weakly supervised models and will encourage researchers to examine the effect of co-occurring diseases on classification performance in the future.

Interpretable Mammographic Image Classification using Cased-Based Reasoning and Deep Learning

When we deploy machine learning models in high-stakes medical settings, we must ensure these models make accurate predictions that are consistent with known medical science. Inherently interpretable networks address this need by explaining the rationale behind each decision while maintaining equal or higher accuracy compared to black-box models. In this work, we present a novel interpretable neural network algorithm that uses case-based reasoning for mammography. Designed to aid a radiologist in their decisions, our network presents both a prediction of malignancy and an explanation of that prediction using known medical features. In order to yield helpful explanations, the network is designed to mimic the reasoning processes of a radiologist: our network first detects the clinically relevant semantic features of each image by comparing each new image with a learned set of prototypical image parts from the training images, then uses those clinical features to predict malignancy. Compared to other methods, our model detects clinical features (mass margins) with equal or higher accuracy, provides a more detailed explanation of its prediction, and is better able to differentiate the classification-relevant parts of the image.

IAIA-BL: A Case-based Interpretable Deep Learning Model for Classification of Mass Lesions in Digital Mammography

Interpretability in machine learning models is important in high-stakes decisions, such as whether to order a biopsy based on a mammographic exam. Mammography poses important challenges that are not present in other computer vision tasks: datasets are small, confounding information is present, and it can be difficult even for a radiologist to decide between watchful waiting and biopsy based on a mammogram alone. In this work, we present a framework for interpretable machine learning-based mammography. In addition to predicting whether a lesion is malignant or benign, our work aims to follow the reasoning processes of radiologists in detecting clinically relevant semantic features of each image, such as the characteristics of the mass margins. The framework includes a novel interpretable neural network algorithm that uses case-based reasoning for mammography. Our algorithm can incorporate a combination of data with whole image labelling and data with pixel-wise annotations, leading to better accuracy and interpretability even with a small number of images. Our interpretable models are able to highlight the classification-relevant parts of the image, whereas other methods highlight healthy tissue and confounding information. Our models are decision aids, rather than decision makers, aimed at better overall human-machine collaboration. We do not observe a loss in mass margin classification accuracy over a black box neural network trained on the same data.

Multi-Label Annotation of Chest Abdomen Pelvis Computed Tomography Text Reports Using Deep Learning

To develop a high throughput multi-label annotator for body Computed Tomography (CT) reports that can be applied to a variety of diseases, organs, and cases. First, we used a dictionary approach to develop a rule-based algorithm (RBA) for extraction of disease labels from radiology text reports. We targeted three organ systems (lungs/pleura, liver/gallbladder, kidneys/ureters) with four diseases per system based on their prevalence in our dataset. To expand the algorithm beyond pre-defined keywords, an attention-guided recurrent neural network (RNN) was trained using the RBA-extracted labels to classify the reports as being positive for one or more diseases or normal for each organ system. Confounding effects on model performance were evaluated using random or pre-trained embedding as well as different sizes of training datasets. Performance was evaluated using the receiver operating characteristic (ROC) area under the curve (AUC) against 2,158 manually obtained labels. Our model extracted disease labels from 261,229 radiology reports of 112,501 unique subjects. Pre-trained models outperformed random embedding across all diseases. As the training dataset size was reduced, performance was robust except for a few diseases with relatively small number of cases. Pre-trained Classification AUCs achieved > 0.95 for all five disease outcomes across all three organ systems. Our label-extracting pipeline was able to encompass a variety of cases and diseases by generalizing beyond strict rules with exceptional accuracy. As a framework, this model can be easily adapted to enable automated labeling of hospital-scale medical data sets for training image-based disease classifiers.