Emergence of "Big Data" and Its Potential and Current Limitations in Medical Imaging

Development of the quantitative PET prostate phantom (Q3P) for improved quality assurance of 18F-PSMA PET imaging in metastatic prostate cancer

BACKGROUND

Phantoms are commonly used to evaluate and compare the performance of imaging systems given the known ground truth. Positron emission tomography (PET) scanners are routinely validated using the NEMA image quality phantom, in which lesions are modeled using 10 to 37 mm fillable spheres. The NEMA phantom neglects, however, to model focal (3-10-mm), high-uptake lesions that are increasingly observed in prostate-specific membrane antigen (PSMA) PET images. PSMA-targeting radiopharmaceuticals allow for enhanced detection of metastatic prostate cancers. As such, there is significant need to develop an updated phantom which considers both the quantitative and lesion detectability of this new paradigm in oncological PET imaging.

PURPOSE

In this work, we present the Quantitative PET Prostate Phantom (Q3P); a portable and modular phantom that can be used to improve and harmonize imaging protocols for 18F-PSMA PET scans.

METHODS

A one-piece cylindrical phantom was designed effectively in two halves, which we call modules. Module 1 was designed to mimic lesions in the presence of background, and Module 2 mimicked very high contrast conditions (i.e., very low background) that can be observed in 18F-PSMA PET scans. Shell-less radioactive spheres (3-16-mm) were cast using epoxy resin mixed with sodium-22 (22Na), a long half-life positron emitter with positron range similar to 18F. To establish realistic lesion contrast, the 22Na spheres were mounted in a cylindrical chamber that can be filled with an 18F background (module 1). Thirteen exchangeable spherical cavity inserts (3-37-mm) were machined in two parts and solvent welded together, and filled with 18F (50 kBq/mL) to model lesions with very high contrast (module 2). Five 2.5-min PET scans were acquired on a 5-ring GE Discovery MI PET/CT scanner (General Electric, USA). Lesions were segmented using 41% of SUVmax fixed thresholding (41% FT) and recovery coefficients (RCs) were computed from 5 noise realizations.

RESULTS

The manufactured phantom is portable (5.7 kg) and scan preparation takes less than 40 min. The total 22Na activity is 250 kBq, allowing it to be shipped as an exempt package under International Atomic Energy Agency (IAEA) regulations. Recovery coefficients, computed using PSF modeling and no post-reconstruction smoothing, were 130.3% (16 mm), 147.1% (10 mm), 87.2% (6 mm), and 7.0% (3 mm) for RCmax, which decreased to 91.1% (16 mm), 90.6% (10 mm), 53.2% (6 mm), and 3.6% (3 mm) for RCmean in the 22Na spheres. Comparatively, 18F sphere recovery was 110.7% (17 mm), 123.6% (10 mm), 106.5% (7 mm), and 23.3% (3 mm) for RCmax, which was reduced to 76.7% (17 mm), 77.7% (10 mm), 66.8% (7 mm), and 13.5% (3 mm), for RCmean.

CONCLUSIONS

A standardized imaging phantom was developed for lesion quantification assessment in 18F-PSMA PET images. The phantom is configurable, providing users with the opportunity to modify background activity levels or sphere sizes according to clinical demands. Distributed to the community, the Q3P phantom has the potential to enable better assessment of lesion quantification and harmonization of 18F-PSMA PET imaging, which may lead to more robust predictive metrics and better outcome prediction in metastatic prostate cancer.

Ethical Data Collection for Medical Image Analysis: a Structured Approach

Due to advancements in technology such as data science and artificial intelligence, healthcare research has gained momentum and is generating new findings and predictions on abnormalities leading to the diagnosis of diseases or disorders in human beings. On one hand, the extensive application of data science to healthcare research is progressing faster, while on the other hand, the ethical concerns and adjoining risks and legal hurdles those data scientists may face in the future slow down the progression of healthcare research. Simply put, the application of data science to ethically guided healthcare research appears to be a dream come true. Hence, in this paper, we discuss the current practices, challenges, and limitations of the data collection process during medical image analysis (MIA) conducted as part of healthcare research and propose an ethical data collection framework to guide data scientists to address the possible ethical concerns before commencing data analytics over a medical dataset.

Near Lossless Compression for 3D Radiological Images Using Optimal Multilinear Singular Value Decomposition (3D-VOI-OMLSVD)

Storage and transmission of high-compression 3D radiological images that create high-quality reconstruction upon decompression are critical necessities for effective and efficient teleradiology. To cater to this need, we propose a near lossless 3D image volume compression method based on optimal multilinear singular value decomposition called "3D-VOI-OMLSVD." The proposed strategy first eliminates any blank 2D image slices from the 3D image volume and uses the selective bounding volume (SBV) to identify and extract the volume of Interest (VOI). Following this, the VOI is decomposed with an optimal multilinear singular value decomposition (OMLSVD) to obtain the corresponding core tensor, factor matrices, and singular values that are compressed with adaptive binary range coder (ABRC), integrated as an entropy encoder. The compressed file can be transferred or transmitted and then decompressed in order to reconstruct the original image. The resultant decompressed VOI is acquired by reversing the above process and then fusing it with the background, using the bound volume coordinates associated with the compressed 3D image. The proposed method performance was tested on a variety of 3D radiological images with different imaging modalities and dimensions using quantitative evaluation metrics such as the compression rate (CR), bit rate (BR), peak signal to noise ratio (PSNR), and structural similarity index (SSIM). Furthermore, we also investigate the impact of VOI extraction on the model performance, before comparing it with two popular compression methods, namely JPEG and JPEG2000. Our proposed method, 3D-VOI-OMLSVD, displayed a high CR value, with a maximum of 37.31, and a low BR, with the lowest reported to be 0.21. The SSIM score was consistently high, with an average performance of 0.9868, while using < 1 second for decoding the image. We observe that with VOI extraction, the compression rate increases manifold, and bit rate drops significantly, and thus reduces the encoding and decoding time to a great extent. Compared to JPEG and JPEG2000, our method consistently performs better in terms of higher CR and lower BR. The results indicate that the proposed compression methodology performs consistently to create high-quality image compressions, and overall gives a better outcome when compared against two state-of-the-art and widely used methods, JPEG and JPEG2000.

Application of Deep Learning in Histopathology Images of Breast Cancer: A Review

With the development of artificial intelligence technology and computer hardware functions, deep learning algorithms have become a powerful auxiliary tool for medical image analysis. This study was an attempt to use statistical methods to analyze studies related to the detection, segmentation, and classification of breast cancer in pathological images. After an analysis of 107 articles on the application of deep learning to pathological images of breast cancer, this study is divided into three directions based on the types of results they report: detection, segmentation, and classification. We introduced and analyzed models that performed well in these three directions and summarized the related work from recent years. Based on the results obtained, the significant ability of deep learning in the application of breast cancer pathological images can be recognized. Furthermore, in the classification and detection of pathological images of breast cancer, the accuracy of deep learning algorithms has surpassed that of pathologists in certain circumstances. Our study provides a comprehensive review of the development of breast cancer pathological imaging-related research and provides reliable recommendations for the structure of deep learning network models in different application scenarios.

Development of scalable lymphatic system in the 4D XCAT phantom: Application to quantitative evaluation of lymphoma PET segmentations

BACKGROUND

Digital anthropomorphic phantoms, such as the 4D extended cardiac-torso (XCAT) phantom, are actively used to develop, optimize, and evaluate a variety of imaging applications, allowing for realistic patient modeling and knowledge of ground truth. The XCAT phantom defines the activity and attenuation for a simulated patient, which includes a complete set of organs, muscle, bone, and soft tissue, while also accounting for cardiac and respiratory motion. However, the XCAT phantom does not currently include the lymphatic system, critical for evaluating medical imaging tasks such as sentinel node detection, node density measurement, and radiation dosimetry.

PURPOSE

In this study, we aimed to develop a scalable lymphatic system in the XCAT phantom, to facilitate improved research of the lymphatic system in medical imaging. Using this scalable lymphatic system, we modeled the lymph node conglomerate pathology that is characteristically observed in primary mediastinal B-cell lymphoma (PMBCL). As an extended application, we evaluated positron emission tomography (PET) image quantification of metabolic tumor volume (MTV) and total lesion glycolysis (TLG) of these simulated lymphomas, though the phantoms may be applied to other imaging modalities and study design paradigms (e.g., image quality, detection).

METHODS

A template model for the lymphatic system was developed based on anatomical data from the Visible Human Project of the National Library of Medicine. The segmented nodes and vessels were fit with non-uniform rational basis spline surfaces, and multichannel large deformation diffeomorphic metric mapping was used to propagate the template to different XCAT anatomies. To model conglomerates observed in PMBCL, lymph nodes were enlarged, converged within the mediastinum, and tracer concentration was increased. We used the phantoms as inputs to a PET simulation tool, which generated images using ordered subsets expectation maximization reconstruction with 2-8 mm Gaussian filters. Fixed thresholding (FT) and gradient segmentation were used to determine MTV and TLG. Percent bias (%Bias) and coefficient of variation (COV) were computed as measures of accuracy and precision, respectively, for each MTV and TLG measurement.

RESULTS

Using the methodology described above, we introduced a scalable lymphatic system in the XCAT phantom, which allows for the radioactivity and attenuation ground truth to be generated in 116 ± 2.5 s using a 2.3 GHz processor. Within the Rhinoceros interface, lymph node anatomy and function were modified to create a cohort of 10 phantoms with lymph node conglomerates. Using the lymphoma phantoms to evaluate PET quantification of MTV, mean %Bias values were -9.3%, -41.3%, and 20.9%, while COV values were 4.08%, 7.6%, and 3.4% using 25% FT, 40% FT, and gradient segmentations, respectively. Comparatively for TLG, mean %Bias values were -27.4%, -45.8%, and -16.0%, while COV values were 1.9%, 5.7%, and 1.4%, for the 25% FT, 40% FT, and gradient segmentations, respectively.

CONCLUSIONS

In this work, we upgraded the XCAT phantom to include a lymphatic system, comprised of a network of 276 scalable lymph nodes and corresponding vessels. As an application, we created a cohort of phantoms with lymph node conglomerates to evaluate lymphoma quantification in PET imaging, which highlights an important application of this work.

Impact of big data resources on clinicians’ activation of prior medical knowledge

Background

Activating prior medical knowledge in diagnosis and treatment is an important basis for clinicians to improve their care ability. However, it has not been systematically explained whether and how various big data resources affect the activation of prior knowledge in the big data environment faced by clinicians.

Objective

The aim of this study is to contribute to a better understanding on how the activation of prior knowledge of clinicians is affected by a wide range of shared and private big data resources, to reveal the impact of big data resources on clinical competence and professional development of clinicians.

Method

Through the comprehensive analysis of extant research results, big data resources are classified as big data itself, big data technology and big data services at the public and institutional levels. A survey was conducted on clinicians and IT personnel in Chinese hospitals. A total of 616 surveys are completed, involving 308 medical institutions. Each medical institution includes a clinician and an IT personnel. SmartPLS version 2.0 software package was used to test the direct impact of big data resources on the activation of prior knowledge. We further analyze their indirect impact of those big data resources without direct impact.

Results

(1) Big data quality environment at the institutional level and the big data sharing environment at the public level directly affect activation of prior medical knowledge; (2) Big data service environment at the institutional level directly affects activation of prior medical knowledge; (3) Big data deployment environment at the institutional level and big data service environment at the public level have no direct impact on activation of prior knowledge of clinicians, but they have an indirect impact through big data quality environment and service environment at the institutional level and the big data sharing environment at the public level.

Conclusions

Big data technology, big data itself and big data service at the public level and institutional level interact and influence each other to activate prior medical knowledge. This study highlights the implications of big data resources on improvement of clinicians’ diagnosis and treatment ability.

Quantifying cilia beat frequency using high-speed video microscopy: Assessing frame rate requirements when imaging different ciliated tissues

Motile cilia are found in numerous locations throughout our body and play a critical role in various physiological processes. The most commonly used method to assess cilia motility is to quantify cilia beat frequency (CBF) via video microscopy. However, a large heterogeneity exists within published literature regarding the framerate used to image cilia motility for calculating CBF. The aim of this study was to determine the optimal frame rate required to image cilia motility for CBF assessment, and if the Nyquist theorem may be used to set this rate. One‐second movies of cilia were collected at >600 fps from mouse airways and ependyma at room‐temperature or 37°C. Movies were then down‐sampled to 30–300 fps. CBF was quantified for identical cilia at different framerates by either manual counting or automated MATLAB script. Airway CBF was significantly impaired in 30 fps movies, while ependymal CBF was significantly impaired in both 60 and 30 fps movies. Pairwise comparison showed that video framerate should be at least 150 fps to accurately measure CBF, with minimal improvement in CBF accuracy in movies >150 fps. The automated script was also found to be less accurate for measuring CBF in lower fps movies than manual counting, however, this difference disappeared in higher framerate movies (>150 fps). In conclusion, our data suggest the Nyquist theorem is unreliable for setting sampling rate for CBF measurement. Instead, sampling rate should be 3–4 times faster than CBF for accurate CBF assessment. Especially if CBF calculation is to be automated.

Automatic Breast Tumor Screening of Mammographic Images with Optimal Convolutional Neural Network

Mammography is a first-line imaging examination approach used for early breast tumor screening. Computational techniques based on deep-learning methods, such as convolutional neural network (CNN), are routinely used as classifiers for rapid automatic breast tumor screening in mammography examination. Classifying multiple feature maps on two-dimensional (2D) digital images, a multilayer CNN has multiple convolutional-pooling layers and fully connected networks, which can increase the screening accuracy and reduce the error rate. However, this multilayer architecture presents some limitations, such as high computational complexity, large-scale training dataset requirements, and poor suitability for real-time clinical applications. Hence, this study designs an optimal multilayer architecture for a CNN-based classifier for automatic breast tumor screening, consisting of three convolutional layers, two pooling layers, a flattening layer, and a classification layer. In the first convolutional layer, the proposed classifier performs the fractional-order convolutional process to enhance the image and remove unwanted noise for obtaining the desired object’s edges; in the second and third convolutional-pooling layers, two kernel convolutional and pooling operations are used to ensure the continuous enhancement and sharpening of the feature patterns for further extracting of the desired features at different scales and different levels. Moreover, there is a reduction of the dimensions of the feature patterns. In the classification layer, a multilayer network with an adaptive moment estimation algorithm is used to refine a classifier’s network parameters for mammography classification by separating tumor-free feature patterns from tumor feature patterns. Images can be selected from a curated breast imaging subset of a digital database for screening mammography (CBIS-DDSM), and K-fold cross-validations are performed. The experimental results indicate promising performance for automatic breast tumor screening in terms of recall (%), precision (%), accuracy (%), F1 score, and Youden’s index.

Application of Big Data to Support Evidence-Based Public Health Policy Decision-Making for Hearing

Ideally, public health policies are formulated from scientific data; however, policy-specific data are often unavailable. Big data can generate ecologically-valid, high-quality scientific evidence, and therefore has the potential to change how public health policies are formulated. Here, we discuss the use of big data for developing evidence-based hearing health policies, using data collected and analyzed with a research prototype of a data repository known as EVOTION (EVidence-based management of hearing impairments: public health pOlicy-making based on fusing big data analytics and simulaTION), to illustrate our points. Data in the repository consist of audiometric clinical data, prospective real-world data collected from hearing aids and an app, and responses to questionnaires collected for research purposes. To date, we have used the platform and a synthetic dataset to model the estimated risk of noise-induced hearing loss and have shown novel evidence of ways in which external factors influence hearing aid usage patterns. We contend that this research prototype data repository illustrates the value of using big data for policy-making by providing high-quality evidence that could be used to formulate and evaluate the impact of hearing health care policies.

The Academic Viewpoint on Patient Data Ownership in the Context of Big Data: Scoping Review

BACKGROUND

The ownership of patient information in the context of big data is a relatively new problem, which is not yet fully recognized by the medical academic community. The problem is interdisciplinary, incorporating legal, ethical, medical, and aspects of information and communication technologies, requiring a sophisticated analysis. However, no previous scoping review has mapped existing studies on the subject.

OBJECTIVE

This study aims to map and assess published studies on patient data ownership in the context of big data as viewed by the academic community.

METHODS

A scoping review was conducted based on the 5-stage framework outlined by Arksey and O'Malley and further developed by Levac, Colquhoun, and O'Brien. The organization and reporting of results of the scoping review were conducted according to PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses and its extensions for Scoping Reviews). A systematic and comprehensive search of 4 scientific information databases, PubMed, ScienceDirect, Scopus, and Springer, was performed for studies published between January 2000 and October 2019. Two authors independently assessed the eligibility of the studies and the extracted data.

RESULTS

The review included 32 eligible articles authored by academicians that correspond to 3 focus areas: problem (ownership), area (health care), and context (big data). Five major aspects were studied: the scientific area of publications, aspects and academicians' perception of ownership in the context of big data, proposed solutions, and practical applications for data ownership issues in the context of big data. The aspects in which publications consider ownership of medical data are not clearly distinguished but can be summarized as ethical, legal, political, and managerial. The ownership of patient data is perceived primarily as a challenge fundamental to conducting medical research, including data sales and sharing, and to a lesser degree as a means of control, problem, threat, and opportunity also in view of medical research. Although numerous solutions falling into 3 categories, technology, law, and policy, were proposed, only 3 real applications were discussed.

CONCLUSIONS

The issue of ownership of patient information in the context of big data is poorly researched; it is not addressed consistently and in its integrity, and there is no consensus on policy decisions and the necessary legal regulations. Future research should investigate the issue of ownership as a core research question and not as a minor fragment among other topics. More research is needed to increase the body of knowledge regarding the development of adequate policies and relevant legal frameworks in compliance with ethical standards. The combined efforts of multidisciplinary academic teams are needed to overcome existing gaps in the perception of ownership, the aspects of ownership, and the possible solutions to patient data ownership issues in the reality of big data.