Defending against Reconstruction Attacks through Differentially Private Federated Learning for Classification of Heterogeneous Chest X-ray Data.

Privacy regulations and the physical distribution of heterogeneous data are often primary concerns for the development of deep learning models in a medical context. This paper evaluates the feasibility of differentially private federated learning for chest X-ray classification as a defense against data privacy attacks. To the best of our knowledge, we are the first to directly compare the impact of differentially private training on two different neural network architectures, DenseNet121 and ResNet50. Extending the federated learning environments previously analyzed in terms of privacy, we simulated a heterogeneous and imbalanced federated setting by distributing images from the public CheXpert and Mendeley chest X-ray datasets unevenly among 36 clients. Both non-private baseline models achieved an area under the receiver operating characteristic curve (AUC) of 0.94 on the binary classification task of detecting the presence of a medical finding. We demonstrate that both model architectures are vulnerable to privacy violation by applying image reconstruction attacks to local model updates from individual clients. The attack was particularly successful during later training stages. To mitigate the risk of a privacy breach, we integrated Rényi differential privacy with a Gaussian noise mechanism into local model training. We evaluate model performance and attack vulnerability for privacy budgets ε∈{1,3,6,10}. The DenseNet121 achieved the best utility-privacy trade-off with an AUC of 0.94 for ε=6. Model performance deteriorated slightly for individual clients compared to the non-private baseline. The ResNet50 only reached an AUC of 0.76 in the same privacy setting. Its performance was inferior to that of the DenseNet121 for all considered privacy constraints, suggesting that the DenseNet121 architecture is more robust to differentially private training.

Tubular shape aware data generation for segmentation in medical imaging.

PURPOSE: Chest X-ray is one of the most widespread examinations of the human body. In interventional radiology, its use is frequently associated with the need to visualize various tube-like objects, such as puncture needles, guiding sheaths, wires, and catheters. Detection and precise localization of these tube-like objects in the X-ray images are, therefore, of utmost value, catalyzing the development of accurate target-specific segmentation algorithms. Similar to the other medical imaging tasks, the manual pixel-wise annotation of the tubes is a resource-consuming process. METHODS: In this work, we aim to alleviate the lack of annotated images by using artificial data. Specifically, we present an approach for synthetic generation of the tube-shaped objects, with a generative adversarial network being regularized with a prior-shape constraint. Namely, our model uses Frangi-based regularization to draw synthetic tubes in the predefined fake mask regions and, then, uses the adversarial component to preserve the global realistic appearance of the synthesized image. RESULTS: Our method eliminates the need for the paired image-mask data and requires only a weakly labeled dataset, with fine-tuning on a small paired sample (10-20 images) proving sufficient to reach the accuracy of the fully supervised models. CONCLUSION: We report the applicability of the approach for the task of segmenting tubes and catheters in the X-ray images, whereas the results should also hold for the other acquisition modalities and image computing applications that contain tubular objects.

Impact of CT reconstruction algorithm on auto-segmentation performance.

Model-based iterative reconstruction (MBIR) reduces CT imaging dose while maintaining image quality. However, MBIR reduces noise while preserving edges which may impact intensity-based tasks such as auto-segmentation. This work evaluates the sensitivity of an auto-contouring prostate atlas across multiple MBIR reconstruction protocols and benchmarks the results against filtered back projection (FBP). Images were created from raw projection data for 11 prostate cancer cases using FBP and nine different MBIR reconstructions (3 protocols/3 noise reduction levels) yielding 10 reconstructions/patient. Five bony structures, bladder, rectum, prostate, and seminal vesicles (SVs) were segmented using an auto-segmentation pipeline that renders 3D binary masks for analysis. Performance was evaluated for volume percent difference (VPD) and Dice similarity coefficient (DSC), using FBP as the gold standard. Nonparametric Friedman tests plus post hoc all pairwise comparisons were employed to test for significant differences (P < 0.05) for soft tissue organs and protocol/level combinations. A physician performed qualitative grading of 396 MBIR contours across the prostate, bladder, SVs, and rectum in comparison to FBP using a six-point scale. MBIR contours agreed with FBP for bony anatomy (DSC ≥ 0.98), bladder (DSC ≥ 0.94, VPD < 8.5%), and prostate (DSC = 0.94 ± 0.03, VPD = 4.50 ± 4.77% (range: 0.07-26.39%). Increased variability was observed for rectum (VPD = 7.50 ± 7.56% and DSC = 0.90 ± 0.08) and SVs (VPD and DSC of 8.23 ± 9.86% range (0.00-35.80%) and 0.87 ± 0.11, respectively). Over the all protocol/level comparisons, a significant difference was observed for the prostate VPD between BSPL1 and BSTL2 (adjusted P-value = 0.039). Nevertheless, 300 of 396 (75.8%) of the four soft tissue structures using MBIR were graded as equivalent or better than FBP, suggesting that MBIR offered potential improvements in auto-segmentation performance when compared to FBP. Future work may involve tuning organ-specific MBIR parameters to further improve auto-segmentation performance. Running title: Impact of CT Reconstruction Algorithm on Auto-segmentation Performance.

Electromagnetic tracking for thermal ablation and biopsy guidance: clinical evaluation of spatial accuracy.

PURPOSE: To evaluate the spatial accuracy of electromagnetic needle tracking and demonstrate the feasibility of ultrasonography (US)-computed tomography (CT) fusion during CT- and US-guided biopsy and radiofrequency ablation procedures. MATERIALS AND METHODS: The authors performed a 20-patient clinical trial to investigate electromagnetic needle tracking during interventional procedures. The study was approved by the institutional investigational review board, and written informed consent was obtained from all patients. Needles were positioned by using CT and US guidance. A commercial electromagnetic tracking device was used in combination with prototype internally tracked needles and custom software to record needle positions relative to previously obtained CT scans. Position tracking data were acquired to evaluate the tracking error, defined as the difference between tracked needle position and reference standard needle position on verification CT scans. Registration between tracking space and image space was obtained by using reference markers attached to the skin ("fiducials"), and different registration methods were compared. The US transducer was tracked to demonstrate the potential use of real-time US-CT fusion for imaging guidance. RESULTS: One patient was excluded from analysis because he was unable to follow breathing instructions during the acquisition of CT scans. Nineteen of the 20 patients were evaluable, demonstrating a basic tracking error of 5.8 mm +/- 2.6, which improved to 3.5 mm +/- 1.9 with use of nonrigid registrations that used previous internal needle positions as additional fiducials. Fusion of tracked US with CT was successful. Patient motion and distortion of the tracking system by the CT table and gantry were identified as sources of error. CONCLUSIONS: The demonstrated spatial tracking accuracy is sufficient to display clinically relevant preprocedural imaging information during needle-based procedures. Virtual needles displayed within preprocedural images may be helpful for clandestine targets such as arterial phase enhancing liver lesions or during thermal ablations when obscuring gas is released. Electromagnetic tracking may help improve imaging guidance for interventional procedures and warrants further investigation, especially for procedures in which the outcomes are dependent on accuracy.

Long-term follow-up after complete ablation of Barrett's esophagus with argon plasma coagulation.

AIM: To report the long-term outcome of patients after complete ablation of non-neoplastic Barrett's esophagus (BE) with respect to BE relapse and development of intraepithelial neoplasia or esophageal adenocarcinoma. METHODS: In 70 patients with histologically proven non-neoplastic BE, complete BE ablation was achieved by argon plasma coagulation (APC) and high-dose proton pump inhibitor therapy (120 mg omeprazole daily). Sixty-six patients (94.4%) underwent further surveillance endoscopy. At each surveillance endoscopy four-quadrant biopsies were taken from the neo-squamous epithelium at 2 cm intervals depending on the pre-treatment length of BE mucosa beginning at the neo-Z-line, and from any endoscopically suspicious lesion. RESULTS: The median follow-up of 66 patients was 51 mo (range 9-85 mo) giving a total of 280.5 patient years. A mean of 6 biopsies were taken during surveillance endoscopies. In 13 patients (19.7%) tongues or islands suspicious for BE were found during endoscopy. In 8 of these patients (12.1%) non-neoplastic BE relapse was confirmed histologically giving a histological relapse rate of 3% per year. In none of the patients, intraepithelial neoplasia nor an esophageal adenocarcinoma was detected. Logistic regression analysis identified endoscopic detection of islands or tongues as the only positive predictor of BE relapse (P = 0.0004). CONCLUSION: The long-term relapse rate of non-neoplastic BE following complete ablation with high-power APC is low (3% per year).

Real-time interactive viewing of 4D kinematic MR joint studies.

Assessment of soft tissue in normal and abnormal joint motion today gets feasible by acquiring time series of 3D MRI images. However, slice-by-slice viewing of such 4D kinematic images is cumbersome, and does not allow appreciating the movement in a convenient way. Simply presenting slice data in a cine-loop will be compromised by through-plane displacements of anatomy and "jerks" between frames, both of which hamper visual analysis of the movement. To overcome these limitations, we have implemented a demonstrator for viewing 4D kinematic MRI datasets. It allows to view any user defined anatomical structure from any viewing perspective in real-time. Smoothly displaying the movement in a cine-loop is realized by image post processing, fixing any user defined anatomical structure after image acquisition.