Dzung L. Pham

Samuel W. Remedios, Shuo Han, Yuan Xue et al.

In 2D multi-slice magnetic resonance (MR) acquisition, the through-plane signals are typically of lower resolution than the in-plane signals. While contemporary super-resolution (SR) methods aim to recover the underlying high-resolution volume, the estimated high-frequency information is implicit via end-to-end data-driven training rather than being explicitly stated and sought. To address this, we reframe the SR problem statement in terms of perfect reconstruction filter banks, enabling us to identify and directly estimate the missing information. In this work, we propose a two-stage approach to approximate the completion of a perfect reconstruction filter bank corresponding to the anisotropic acquisition of a particular scan. In stage 1, we estimate the missing filters using gradient descent and in stage 2, we use deep networks to learn the mapping from coarse coefficients to detail coefficients. In addition, the proposed formulation does not rely on external training data, circumventing the need for domain shift correction. Under our approach, SR performance is improved particularly in "slice gap" scenarios, likely due to the constrained solution space imposed by the framework.

Dzung Pham, Shreyas Kulkarni, Amir Houmansadr

Federated learning has emerged as a promising privacy-preserving solution for machine learning domains that rely on user interactions, particularly recommender systems and online learning to rank. While there has been substantial research on the privacy of traditional federated learning, little attention has been paid to the privacy properties of these interaction-based settings. In this work, we show that users face an elevated risk of having their private interactions reconstructed by the central server when the server can control the training features of the items that users interact with. We introduce RAIFLE, a novel optimization-based attack framework where the server actively manipulates the features of the items presented to users to increase the success rate of reconstruction. Our experiments with federated recommendation and online learning-to-rank scenarios demonstrate that RAIFLE is significantly more powerful than existing reconstruction attacks like gradient inversion, achieving high performance consistently in most settings. We discuss the pros and cons of several possible countermeasures to defend against RAIFLE in the context of interaction-based federated learning. Our code is open-sourced at https://github.com/dzungvpham/raifle.

Dzung Pham, Peter Kairouz, Niloofar Mireshghallah et al.

Large language models (LLMs) are increasingly being used to protect sensitive user data. However, current LLM-based privacy solutions assume that these models can reliably detect personally identifiable information (PII), particularly named entities. In this paper, we challenge that assumption by revealing systematic failures in LLM-based privacy tasks. Specifically, we show that modern LLMs regularly overlook human names even in short text snippets due to ambiguous contexts, which cause the names to be misinterpreted or mishandled. We propose AMBENCH, a benchmark dataset of seemingly ambiguous human names, leveraging the name regularity bias phenomenon, embedded within concise text snippets along with benign prompt injections. Our experiments on modern LLMs tasked to detect PII as well as specialized tools show that recall of ambiguous names drops by 20--40% compared to more recognizable names. Furthermore, ambiguous human names are four times more likely to be ignored in supposedly privacy-preserving summaries generated by LLMs when benign prompt injections are present. These findings highlight the underexplored risks of relying solely on LLMs to safeguard user privacy and underscore the need for a more systematic investigation into their privacy failure modes.

Zhangxing Bian, Ahmed Alshareef, Shuwen Wei et al.

Magnetic Resonance Imaging with tagging (tMRI) has long been utilized for quantifying tissue motion and strain during deformation. However, a phenomenon known as tag fading, a gradual decrease in tag visibility over time, often complicates post-processing. The first contribution of this study is to model tag fading by considering the interplay between $T_1$ relaxation and the repeated application of radio frequency (RF) pulses during serial imaging sequences. This is a factor that has been overlooked in prior research on tMRI post-processing. Further, we have observed an emerging trend of utilizing raw tagged MRI within a deep learning-based (DL) registration framework for motion estimation. In this work, we evaluate and analyze the impact of commonly used image similarity objectives in training DL registrations on raw tMRI. This is then compared with the Harmonic Phase-based approach, a traditional approach which is claimed to be robust to tag fading. Our findings, derived from both simulated images and an actual phantom scan, reveal the limitations of various similarity losses in raw tMRI and emphasize caution in registration tasks where image intensity changes over time.

Samuel W. Remedios, Zihao Wu, Camilo Bermudez et al.

Multiple instance learning (MIL) is a supervised learning methodology that aims to allow models to learn instance class labels from bag class labels, where a bag is defined to contain multiple instances. MIL is gaining traction for learning from weak labels but has not been widely applied to 3D medical imaging. MIL is well-suited to clinical CT acquisitions since (1) the highly anisotropic voxels hinder application of traditional 3D networks and (2) patch-based networks have limited ability to learn whole volume labels. In this work, we apply MIL with a deep convolutional neural network to identify whether clinical CT head image volumes possess one or more large hemorrhages (> 20cm$^3$), resulting in a learned 2D model without the need for 2D slice annotations. Individual image volumes are considered separate bags, and the slices in each volume are instances. Such a framework sets the stage for incorporating information obtained in clinical reports to help train a 2D segmentation approach. Within this context, we evaluate the data requirements to enable generalization of MIL by varying the amount of training data. Our results show that a training size of at least 400 patient image volumes was needed to achieve accurate per-slice hemorrhage detection. Over a five-fold cross-validation, the leading model, which made use of the maximum number of training volumes, had an average true positive rate of 98.10%, an average true negative rate of 99.36%, and an average precision of 0.9698. The models have been made available along with source code to enabled continued exploration and adaption of MIL in CT neuroimaging.

Samuel Remedios, Snehashis Roy, Justin Blaber et al.

Machine learning models are becoming commonplace in the domain of medical imaging, and with these methods comes an ever-increasing need for more data. However, to preserve patient anonymity it is frequently impractical or prohibited to transfer protected health information (PHI) between institutions. Additionally, due to the nature of some studies, there may not be a large public dataset available on which to train models. To address this conundrum, we analyze the efficacy of transferring the model itself in lieu of data between different sites. By doing so we accomplish two goals: 1) the model gains access to training on a larger dataset that it could not normally obtain and 2) the model better generalizes, having trained on data from separate locations. In this paper, we implement multi-site learning with disparate datasets from the National Institutes of Health (NIH) and Vanderbilt University Medical Center (VUMC) without compromising PHI. Three neural networks are trained to convergence on a computed tomography (CT) brain hematoma segmentation task: one only with NIH data,one only with VUMC data, and one multi-site model alternating between NIH and VUMC data. Resultant lesion masks with the multi-site model attain an average Dice similarity coefficient of 0.64 and the automatically segmented hematoma volumes correlate to those done manually with a Pearson correlation coefficient of 0.87,corresponding to an 8% and 5% improvement, respectively, over the single-site model counterparts.

Dzung L. Pham, Snehashis Roy

A key feature of magnetic resonance (MR) imaging is its ability to manipulate how the intrinsic tissue parameters of the anatomy ultimately contribute to the contrast properties of the final, acquired image. This flexibility, however, can lead to substantial challenges for segmentation algorithms, particularly supervised methods. These methods require atlases or training data, which are composed of MR image and labeled image pairs. In most cases, the training data are obtained with a fixed acquisition protocol, leading to suboptimal performance when an input data set that requires segmentation has differing contrast properties. This drawback is increasingly significant with the recent movement towards multi-center research studies involving multiple scanners and acquisition protocols. In this work, we propose a new framework for supervised segmentation approaches that is robust to contrast differences between the training MR image and the input image. Our approach uses a generative simulation model within the segmentation process to compensate for the contrast differences. We allow the contrast of the MR image in the training data to vary by simulating a new contrast from the corresponding label image. The model parameters are optimized by a cost function measuring the consistency between the input MR image and its simulation based on a current estimate of the segmentation labels. We provide a proof of concept of this approach by combining a supervised classifier with a simple simulation model, and apply the resulting algorithm to synthetic images and actual MR images.

Snehashis Roy, John A. Butman, Leighton Chan et al.

Traumatic brain injury (TBI) is caused by a sudden trauma to the head that may result in hematomas and contusions and can lead to stroke or chronic disability. An accurate quantification of the lesion volumes and their locations is essential to understand the pathophysiology of TBI and its progression. In this paper, we propose a fully convolutional neural network (CNN) model to segment contusions and lesions from brain magnetic resonance (MR) images of patients with TBI. The CNN architecture proposed here was based on a state of the art CNN architecture from Google, called Inception. Using a 3-layer Inception network, lesions are segmented from multi-contrast MR images. When compared with two recent TBI lesion segmentation methods, one based on CNN (called DeepMedic) and another based on random forests, the proposed algorithm showed improved segmentation accuracy on images of 18 patients with mild to severe TBI. Using a leave-one-out cross validation, the proposed model achieved a median Dice of 0.75, which was significantly better (p<0.01) than the two competing methods.

Snehashis Roy, John A. Butman, Daniel S. Reich et al.

Multiple Sclerosis (MS) is an autoimmune disease that leads to lesions in the central nervous system. Magnetic resonance (MR) images provide sufficient imaging contrast to visualize and detect lesions, particularly those in the white matter. Quantitative measures based on various features of lesions have been shown to be useful in clinical trials for evaluating therapies. Therefore robust and accurate segmentation of white matter lesions from MR images can provide important information about the disease status and progression. In this paper, we propose a fully convolutional neural network (CNN) based method to segment white matter lesions from multi-contrast MR images. The proposed CNN based method contains two convolutional pathways. The first pathway consists of multiple parallel convolutional filter banks catering to multiple MR modalities. In the second pathway, the outputs of the first one are concatenated and another set of convolutional filters are applied. The output of this last pathway produces a membership function for lesions that may be thresholded to obtain a binary segmentation. The proposed method is evaluated on a dataset of 100 MS patients, as well as the ISBI 2015 challenge data consisting of 14 patients. The comparison is performed against four publicly available MS lesion segmentation methods. Significant improvement in segmentation quality over the competing methods is demonstrated on various metrics, such as Dice and false positive ratio. While evaluating on the ISBI 2015 challenge data, our method produces a score of 90.48, where a score of 90 is considered to be comparable to a human rater.