Search results

This study aims to summarize the critical issues in medical federated learning and applicable solutions. Also, detailed explanations of how federated learning techniques can be applied to the medical field are presented. About 80 reference studies described in the field were reviewed, and the federated learning framework currently being developed by the research team is provided. This paper will help researchers to build an actual medical federated learning environment.

Since machine learning techniques emerged, more efficient analysis was possible with a large amount of data. However, data regulations have been tightened worldwide, and the usage of centralized machine learning methods has become almost infeasible. Federated learning techniques have been introduced as a solution. Even with its powerful structural advantages, there still exist unsolved challenges in federated learning in a real medical data environment. This paper aims to summarize those by category and presents possible solutions.

This paper provides four critical categorized issues to be aware of when applying the federated learning technique to the actual medical data environment, then provides general guidelines for building a federated learning environment as a solution.

Existing studies have dealt with issues such as heterogeneity problems in the federated learning environment itself, but those were lacking on how these issues incur problems in actual working tasks. Therefore, this paper helps researchers understand the federated learning issues through examples of actual medical machine learning environments.

There are over 40 million amputees globally with more than 185,000 Americans losing their limbs every year. For most of the world, prosthetic devices remain too expensive and uncomfortable. This paper aims to outline advancements made by a multidisciplinary research group, interested in advancing the restoration of human motion through accessible lower limb prostheses.

Customization, comfort and functionality are the most important metrics reported by prosthetists and patients. The work of this paper presents the design and manufacturing of a custom made, cost-effective and functional three-dimensional (3D) printed transtibial prosthesis monocoque design. The design of the prosthesis integrates 3D imaging, modelling and optimization techniques coupled with additive manufacturing.

The successful fabrication of a functional monocoque prosthesis through 3D printing indicates the workflow may be a solution to the worldwide accessibility crisis. The digital workflow developed in this work offers great potential for providing prosthetic devices to rural communities, which lack access to skilled prosthetic physicians. The authors found that using the workflow together with 3D printing, this study can create custom monocoque prostheses (Figure 16). These prostheses are comfortable, functional and properly aligned. In comparison with traditional prosthetic devices, the authors slowered the average cost, weight and time of production by 95%, 55% and 95%, respectively.

This novel digital design and manufacturing workflow has the potential to democratize and globally proliferate access to prosthetic devices, which restore the patient’s mobility, quality of life and health. LIMBER’s toolbox can reach places where proper prosthetic and orthotic care is not available. The digital workflow reduces the cost of making custom devices by an order of magnitude, enabling broader reach, faster access and improved comfort. This is particularly important for children who grow quickly and need new devices every few months or years, timely access is both physically and psychologically important.

In this manuscript, the authors show the application of digital design techniques for fabricating prosthetic devices. The proposed workflow implements several advantageous changes and, most importantly, digitally blends the three components of a transtibial prosthesis into a single, 3D printable monocoque device. The development of a novel unibody transtibial device that is properly aligned and adjusted digitally, greatly reduces the number of visits an amputee must make to a clinic to have a certified prosthetist adjust and modify their prosthesis. The authors believe this novel workflow has the potential to ease the worldwide accessibility crisis for prostheses.

This paper aims to report on the findings of an investigation to compare three different three-dimensional printing (3DP) or additive manufacturing technologies [i.e. fused deposition modeling (FDM), stereolithography (SLA) and material jetting (MJ)] and four different equipment (FDM, SLA, MJP 2600 and Object 260) in terms of their dimensional process capability (dimensional accuracy and surface roughness). It provides a comprehensive and comparative understanding about the level of attainable dimensional accuracy, repeatability and surface roughness of commonly used 3DP technologies. It is expected that these findings will help other researchers and industrialists in choosing the right technology and equipment for a given 3DP application.

A benchmark model of 5 × 5 cm with several common and challenging features, such as around protrusion and hole, flat surface, micro-scale ribs and micro-scale long channels was designed and printed repeatedly using four different equipment of three different 3DP technologies. The dimensional accuracy of the printed models was measured using non-contact digital measurement methods. The surface roughness was evaluated using a digital profilometer. Finally, the surface quality and edge sharpness were evaluated under a reflected light ZEISS microscope with a 50× magnification objective.

The results show that FDM technology with the used equipment results in a rough surface and loose dimensional accuracy. The SLA printer produced a smoother surface, but resulted in the distortion of thin features (<1 mm). MJ printers, on the other hand, produced comparable surface roughness and dimensional accuracy. However, ProJet MJP 3600 produced sharper edges when compared to the Objet 260 that produced round edges.

This paper, for the first time, provides a comprehensive comparison of three different commonly used 3DP technologies in terms of their dimensional capability and surface roughness without farther post-processing. Thus, it offers a reliable guideline for design consideration and printer selection based on the target application.

Crystallization is the process widely used for components separation and solids purification. The systems for crystallization process evaluation applied so far, involve numerous non-invasive tomographic measurement techniques which suffers from some reported problems. The purpose of this paper is to show the abilities of three-dimensional Electrical Capacitance Tomography (3D ECT) in the context of non-invasive and non-intrusive visualization of crystallization processes. Multiple aspects and problems of ECT imaging, as well as the computer model design to work with the high relative permittivity liquids, have been pointed out.

To design the most efficient (from a mechanical and electrical point of view) 3D ECT sensor structure, the high-precise impedance meter was applied. The three types of sensor were designed, built, and tested. To meet the new concept requirements, the dedicated ECT device has been constructed.

It has been shown that the ECT technique can be applied to the diagnosis of crystallization. The crystals distribution can be identified using this technique. The achieved measurement resolution allows detecting the localization of crystals. The usage of stabilized electrodes improves the sensitivity of the sensor and provides the images better suitable for further analysis.

The dedicated 3D ECT sensor construction has been proposed to increase its sensitivity in the border area, where the crystals grow. Regarding this feature, some new algorithms for the potential field distribution and the sensitivity matrix calculation have been developed. The adaptation of the iterative 3D image reconstruction process has also been described.

At airport security checkpoints, baggage screening is aimed to prevent transportation of prohibited and potentially dangerous items. Observing the projection images generated by X-rays scanner is a critical method. However, when multiple objects are stacked on top of each other, distinguishing objects only by a two-dimensional picture is difficult, which prompts the demand for more precise imaging technology to be investigated for use. Reconstructing from 2D X-ray images to 3D-computed tomography (CT) volumes is a reliable solution.

To more accurately distinguish the specific contour shape of items when stacked, multi-information fusion network (MFCT-GAN) based on generative adversarial network (GAN) and U-like network (U-NET) is proposed to reconstruct from two biplanar orthogonal X-ray projections into 3D CT volumes. The authors use three modules to enhance the reconstruction qualitative and quantitative effects, compared with the original network. The skip connection modification (SCM) and multi-channels residual dense block (MRDB) enable the network to extract more feature information and learn deeper with high efficiency; the introduction of subjective loss enables the network to focus on the structural similarity (SSIM) of images during training.

On account of the fusion of multiple information, MFCT-GAN can significantly improve the value of quantitative indexes and distinguish contour explicitly between different targets. In particular, SCM enables features more reasonable and accurate when expanded into three dimensions. The appliance of MRDB can alleviate problem of slow optimization during the late training period, as well as reduce the computational cost. The introduction of subjective loss guides network to retain more high-frequency information, which makes the rendered CT volumes clearer in details.

The authors' proposed MFCT-GAN is able to restore the 3D shapes of different objects greatly based on biplanar projections. This is helpful in security check places, where X-ray images of stacked objects need to be distinguished from the presence of prohibited objects. The authors adopt three new modules, SCM, MRDB and subjective loss, as well as analyze the role the modules play in 3D reconstruction. Results show a significant improvement on the reconstruction both in objective and subjective effects.

A user-friendly HTML-based open-source software has been developed for structural shielding design of medical X-ray imaging facilities. Based on values published by the NCRP Report N° 147 the software allows thickness calculations for different materials used in conventional X-ray rooms, mammography rooms and computed tomography rooms, diminishing errors resulting from the use of curves. The software focuses on the optimization principle by considering workload distributions instead of applying all the workload at a single high operating potential. The input data can be those recommended by the NCRP Report N° 147 or, if the facility has its own data, they can be used instead. With the implemented methodology, the code validation was performed by comparison of the results with a study case provided by the Report. The software application is available in two languages (English and Spanish) and provides the accuracy of the method presented, as well as assisting the physicist in shielding computations in a user-friendly manner. This software tool is available upon request to the corresponding author.