Domain adaptation using AdaBN and AdaIN for high-resolution IVD mesh reconstruction from clinical MRI.

PURPOSE: Deep learning has firmly established its dominance in medical imaging applications. However, careful consideration must be exercised when transitioning a trained source model to adapt to an entirely distinct environment that deviates significantly from the training set. The majority of the efforts to mitigate this issue have predominantly focused on classification and segmentation tasks. In this work, we perform a domain adaptation of a trained source model to reconstruct high-resolution intervertebral disc meshes from low-resolution MRI. METHODS: To address the outlined challenges, we use MRI2Mesh as the shape reconstruction network. It incorporates three major modules: image encoder, mesh deformation, and cross-level feature fusion. This feature fusion module is used to encapsulate local and global disc features. We evaluate two major domain adaptation techniques: adaptive batch normalization (AdaBN) and adaptive instance normalization (AdaIN) for the task of shape reconstruction. RESULTS: Experiments conducted on distinct datasets, including data from different populations, machines, and test sites demonstrate the effectiveness of MRI2Mesh for domain adaptation. MRI2Mesh achieved up to a 14% decrease in Hausdorff distance (HD) and a 19% decrease in the point-to-surface (P2S) metric for both AdaBN and AdaIN experiments, indicating improved performance. CONCLUSION: MRI2Mesh has demonstrated consistent superiority to the state-of-the-art Voxel2Mesh network across a diverse range of datasets, populations, and scanning protocols, highlighting its versatility. Additionally, AdaBN has emerged as a robust method compared to the AdaIN technique. Further experiments show that MRI2Mesh, when combined with AdaBN, holds immense promise for enhancing the precision of anatomical shape reconstruction in domain adaptation.

Diabetic foot ulcers segmentation challenge report: Benchmark and analysis.

Monitoring the healing progress of diabetic foot ulcers is a challenging process. Accurate segmentation of foot ulcers can help podiatrists to quantitatively measure the size of wound regions to assist prediction of healing status. The main challenge in this field is the lack of publicly available manual delineation, which can be time consuming and laborious. Recently, methods based on deep learning have shown excellent results in automatic segmentation of medical images, however, they require large-scale datasets for training, and there is limited consensus on which methods perform the best. The 2022 Diabetic Foot Ulcers segmentation challenge was held in conjunction with the 2022 International Conference on Medical Image Computing and Computer Assisted Intervention, which sought to address these issues and stimulate progress in this research domain. A training set of 2000 images exhibiting diabetic foot ulcers was released with corresponding segmentation ground truth masks. Of the 72 (approved) requests from 47 countries, 26 teams used this data to develop fully automated systems to predict the true segmentation masks on a test set of 2000 images, with the corresponding ground truth segmentation masks kept private. Predictions from participating teams were scored and ranked according to their average Dice similarity coefficient of the ground truth masks and prediction masks. The winning team achieved a Dice of 0.7287 for diabetic foot ulcer segmentation. This challenge has now entered a live leaderboard stage where it serves as a challenging benchmark for diabetic foot ulcer segmentation.

PIRET-A Platform for Treatment Planning in Electroporation-Based Therapies.

Tissue electroporation is the basis of several therapies. Electroporation is performed by briefly exposing tissues to high electric fields. It is generally accepted that electroporation is effective where an electric field magnitude threshold is overreached. However, it is difficult to preoperatively estimate the field distribution because it is highly dependent on anatomy and treatment parameters. OBJECTIVE: We developed PIRET, a platform to predict the treatment volume in electroporation-based therapies. METHODS: The platform seamlessly integrates tools to build patient-specific models where the electric field is simulated to predict the treatment volume. Patient anatomy is segmented from medical images and 3D reconstruction aids in placing the electrodes and setting up treatment parameters. RESULTS: Four canine patients that had been treated with high-frequency irreversible electroporation were retrospectively planned with PIRET and with a workflow commonly used in previous studies, which uses different general-purpose segmentation (3D Slicer) and modeling software (3Matic and COMSOL Multiphysics). PIRET outperformed the other workflow by 65 minutes (× 1.7 faster), thanks to the improved user experience during treatment setup and model building. Both approaches computed similarly accurate electric field distributions, with average Dice scores higher than 0.93. CONCLUSION: A platform which integrates all the required tools for electroporation treatment planning is presented. Treatment plan can be performed rapidly with minimal user interaction in a stand-alone platform. SIGNIFICANCE: This platform is, to the best of our knowledge, the most complete software for treatment planning of irreversible electroporation. It can potentially be used for other electroporation applications.

Prediction of Lung Nodule Progression with an Uncertainty-Aware Hierarchical Probabilistic Network.

Predicting whether a lung nodule will grow, remain stable or regress over time, especially early in its follow-up, would help doctors prescribe personalized treatments and better surgical planning. However, the multifactorial nature of lung tumour progression hampers the identification of growth patterns. In this work, we propose a deep hierarchical generative and probabilistic network that, given an initial image of the nodule, predicts whether it will grow, quantifies its future size and provides its expected semantic appearance at a future time. Unlike previous solutions, our approach also estimates the uncertainty in the predictions from the intrinsic noise in medical images and the inter-observer variability in the annotations. The evaluation of this method on an independent test set reported a future tumour growth size mean absolute error of 1.74 mm, a nodule segmentation Dice's coefficient of 78% and a tumour growth accuracy of 84% on predictions made up to 24 months ahead. Due to the lack of similar methods for providing future lung tumour growth predictions, along with their associated uncertainty, we adapted equivalent deterministic and alternative generative networks (i.e., probabilistic U-Net, Bayesian test dropout and Pix2Pix). Our method outperformed all these methods, corroborating the adequacy of our approach.

EView: An electric field visualization web platform for electroporation-based therapies.

BACKGROUND AND OBJECTIVES: Electroporation is the phenomenon by which cell membrane permeability to ions and macromolecules is increased when the cell is briefly exposed to high electric fields. In electroporation-based treatments, such exposure is typically performed by delivering high voltage pulses across needle electrodes in tissue. For a given tissue and pulsing protocol, an electric field magnitude threshold exists that must be overreached for treatment efficacy. However, it is hard to preoperatively infer the treatment volume because the electric field distribution intricately depends on the electrodes' positioning and length, the applied voltage, and the electric conductivity of the treated tissues. For illustrating such dependencies, we have created EView (https://eview.upf.edu), a web platform that estimates the electric field distribution for arbitrary needle electrode locations and orientations and overlays it on 3D medical images. METHODS: A client-server approach has been implemented to let the user set the electrode configuration easily on the web browser, whereas the simulation is computed on a dedicated server. By means of the finite element method, the electric field is solved in a 3D volume. For the sake of simplicity, only a homogeneous tissue is modeled, assuming the same properties for healthy and pathologic tissues. The non-linear dependence of tissue conductivity on the electric field due to the electroporation effect is modeled. The implemented model has been validated against a state of the art finite element solver, and the server has undergone a heavy load test to ensure reliability and to report execution times. RESULTS: The electric field is rapidly computed for any electrode and tissue configuration, and alternative setups can be easily compared. The platform provides the same results as the state of the art finite element solver (Dice = 98.3 ± 0.4%). During the high load test, the server remained responsive. Simulations are computed in less than 2 min for simple cases consisting of two electrodes and take up to 40 min for complex scenarios consisting of 6 electrodes. CONCLUSIONS: With this free platform we provide expert and non-expert electroporation users a way to rapidly model the electric field distribution for arbitrary electrode configurations.

Fetal MRI Synthesis via Balanced Auto-Encoder Based Generative Adversarial Networks.

Machine learning approaches for image analysis require large amounts of training imaging data. As an alternative, the use of realistic synthetic data reduces the high cost associated to medical image acquisition, as well as avoiding confidentiality and privacy issues, and consequently allows the creation of public data repositories for scientific purposes. Within the context of fetal imaging, we adopt an auto-encoder based Generative Adversarial Network for synthetic fetal MRI generation. The proposed architecture features a balanced power of the discriminator against the generator during training, provides an approximate convergence measure, and enables fast and robust training to generate high-quality fetal MRI in axial, sagittal and coronal planes. We demonstrate the feasibility of the proposed approach quantitatively and qualitatively by segmenting relevant fetal structures to assess the anatomical fidelity of the simulation, and performing a clinical verisimilitude study distinguishing the simulated data from the real images. The results obtained so far are promising, which makes further investigation on this new topic worthwhile.

Weighted regularized statistical shape space projection for breast 3D model reconstruction.

The use of 3D imaging has increased as a practical and useful tool for plastic and aesthetic surgery planning. Specifically, the possibility of representing the patient breast anatomy in a 3D shape and simulate aesthetic or plastic procedures is a great tool for communication between surgeon and patient during surgery planning. For the purpose of obtaining the specific 3D model of the breast of a patient, model-based reconstruction methods can be used. In particular, 3D morphable models (3DMM) are a robust and widely used method to perform 3D reconstruction. However, if additional prior information (i.e., known landmarks) is combined with the 3DMM statistical model, shape constraints can be imposed to improve the 3DMM fitting accuracy. In this paper, we present a framework to fit a 3DMM of the breast to two possible inputs: 2D photos and 3D point clouds (scans). Our method consists in a Weighted Regularized (WR) projection into the shape space. The contribution of each point in the 3DMM shape is weighted allowing to assign more relevance to those points that we want to impose as constraints. Our method is applied at multiple stages of the 3D reconstruction process. Firstly, it can be used to obtain a 3DMM initialization from a sparse set of 3D points. Additionally, we embed our method in the 3DMM fitting process in which more reliable or already known 3D points or regions of points, can be weighted in order to preserve their shape information. The proposed method has been tested in two different input settings: scans and 2D pictures assessing both reconstruction frameworks with very positive results.

A multiscale imaging and modelling dataset of the human inner ear.

Understanding the human inner ear anatomy and its internal structures is paramount to advance hearing implant technology. While the emergence of imaging devices allowed researchers to improve understanding of intracochlear structures, the difficulties to collect appropriate data has resulted in studies conducted with few samples. To assist the cochlear research community, a large collection of human temporal bone images is being made available. This data descriptor, therefore, describes a rich set of image volumes acquired using cone beam computed tomography and micro-CT modalities, accompanied by manual delineations of the cochlea and sub-compartments, a statistical shape model encoding its anatomical variability, and data for electrode insertion and electrical simulations. This data makes an important asset for future studies in need of high-resolution data and related statistical data objects of the cochlea used to leverage scientific hypotheses. It is of relevance to anatomists, audiologists, computer scientists in the different domains of image analysis, computer simulations, imaging formation, and for biomedical engineers designing new strategies for cochlear implantations, electrode design, and others.

Automated cement segmentation in vertebroplasty.

Vertebroplasty is a minimally invasive procedure with many benefits; however, the procedure is not without risks and potential complications, of which leakage of the cement out of the vertebral body and into the surrounding tissues is one of the most serious. Cement can leak into the spinal canal, venous system, soft tissues, lungs and intradiscal space, causing serious neurological complications, tissue necrosis or pulmonary embolism. We present a method for automatic segmentation and tracking of bone cement during vertebroplasty procedures, as a first step towards developing a warning system to avoid cement leakage outside the vertebral body. We show that by using active contours based on level sets the shape of the injected cement can be accurately detected. The model has been improved for segmentation as proposed in our previous work by including a term that restricts the level set function to the vertebral body. The method has been applied to a set of real intra-operative X-ray images and the results show that the algorithm can successfully detect different shapes with blurred and not well-defined boundaries, where the classical active contours segmentation is not applicable. The method has been positively evaluated by physicians.

Accurate and robust reconstruction of a surface model of the proximal femur from sparse-point data and a dense-point distribution model for surgical navigation.

Constructing a 3-D surface model from sparse-point data is a nontrivial task. Here, we report an accurate and robust approach for reconstructing a surface model of the proximal femur from sparse-point data and a dense-point distribution model (DPDM). The problem is formulated as a three-stage optimal estimation process. The first stage, affine registration, is to iteratively estimate a scale and a rigid transformation between the mean surface model of the DPDM and the sparse input points. The estimation results of the first stage are used to establish point correspondences for the second stage, statistical instantiation, which stably instantiates a surface model from the DPDM using a statistical approach. This surface model is then fed to the third stage, kernel-based deformation, which further refines the surface model. Handling outliers is achieved by consistently employing the least trimmed squares (LTS) approach with a roughly estimated outlier rate in all three stages. If an optimal value of the outlier rate is preferred, we propose a hypothesis testing procedure to automatically estimate it. We present here our validations using four experiments, which include 1) leave-one-out experiment, 2) experiment on evaluating the present approach for handling pathology, 3) experiment on evaluating the present approach for handling outliers, and 4) experiment on reconstructing surface models of seven dry cadaver femurs using clinically relevant data without noise and with noise added. Our validation results demonstrate the robust performance of the present approach in handling outliers, pathology, and noise. An average 95-percentile error of 1.7-2.3 mm was found when the present approach was used to reconstruct surface models of the cadaver femurs from sparse-point data with noise added.

A comparison study assessing the feasibility of ultrasound-initialized deformable bone models.

This article presents a feasibility and evaluation study for using 2D ultrasound in conjunction with our statistical deformable bone model within the scope of computer-assisted surgery. The final aim is to provide the surgeon with enhanced 3D visualization for surgical navigation in orthopedic surgery without the need for preoperative CT or MRI scans. We unified our earlier work to combine several automatic methods for statistical bone shape prediction and ultrasound segmentation and calibration to provide the intended rapid and accurate visualization. We compared the use of a tracked digitizing pointer and ultrasound for acquiring landmarks and bone surface points for the estimation of two cast proximal femurs.