Fast and accurate 3-D spine MRI segmentation using FastCleverSeg

Accurate and efficient segmenting of vertebral bodies, muscles, and discs is crucial for analyzing various spinal diseases. However, traditional methods are either laborious and time-consuming (manual segmentation) or require extensive training data (fully automatic segmentation). FastCleverSeg, our proposed semi-automatic segmentation approach, addresses those limitations by significantly reducing user interaction while maintaining high accuracy. First, we reduce user interaction by requiring the manual annotation of only two or three slices. Next, we automatically Estimate the Annotation on Intermediary Slices (EANIS) using traditional computer vision/graphics concepts. Finally, our proposed method leverages improved voxel weight balancing to achieve fast and precise volumetric segmentation in the segmentation process. Experimental evaluations on our assembled diverse MRI databases comprising 179 patients (60 male, 119 female), demonstrate a remarkable 25 ms (30 ms standard deviation) processing time and a significant reduction in user interaction compared to existing approaches. Importantly, FastCleverSeg maintains or surpasses the segmentation quality of competing methods, achieving a Dice score of 94%. This invaluable tool empowers physicians to efficiently generate reliable ground truths, expediting the segmentation process and paving the way for future integration with deep learning approaches. In turn, this opens exciting possibilities for future fully automated spine segmentation.

Patient assignment optimization in cloud healthcare systems: a distributed genetic algorithm

Integrating Internet technologies with traditional healthcare systems has enabled the emergence of cloud healthcare systems. These systems aim to optimize the balance between online diagnosis and offline treatment to effectively reduce patients' waiting times and improve the utilization of idle medical resources. In this paper, a distributed genetic algorithm (DGA) is proposed as a means to optimize the balance of patient assignment (PA) in cloud healthcare systems. The proposed DGA utilizes individuals as solutions for the PA optimization problem and generates better solutions through the execution of crossover, mutation, and selection operators. Besides, the distributed framework in the DGA is proposed to improve its population diversity and scalability. Experimental results demonstrate the effectiveness of the proposed DGA in optimizing the PA problem within the cloud healthcare systems.

ClinicalPath: A Visualization Tool to Improve the Evaluation of Electronic Health Records in Clinical Decision-Making

Physicians work at a very tight schedule and need decision-making support tools to help on improving and doing their work in a timely and dependable manner. Examining piles of sheets with test results and using systems with little visualization support to provide diagnostics is daunting, but that is still the usual way for the physicians' daily procedure, especially in developing countries. Electronic Health Records systems have been designed to keep the patients' history and reduce the time spent analyzing the patient's data. However, better tools to support decision-making are still needed. In this article, we propose ClinicalPath, a visualization tool for users to track a patient's clinical path through a series of tests and data, which can aid in treatments and diagnoses. Our proposal is focused on patient's data analysis, presenting the test results and clinical history longitudinally. Both the visualization design and the system functionality were developed in close collaboration with experts in the medical domain to ensure a right fit of the technical solutions and the real needs of the professionals. We validated the proposed visualization based on case studies and user assessments through tasks based on the physician's daily activities. Our results show that our proposed system improves the physicians' experience in decision-making tasks, made with more confidence and better usage of the physicians' time, allowing them to take other needed care for the patients.

Security and Privacy in Machine Learning for Health Systems: Strategies and Challenges

OBJECTIVES

Machine learning (ML) is a powerful asset to support physicians in decision-making procedures, providing timely answers. However, ML for health systems can suffer from security attacks and privacy violations. This paper investigates studies of security and privacy in ML for health.

METHODS

We examine attacks, defenses, and privacy-preserving strategies, discussing their challenges. We conducted the following research protocol: starting a manual search, defining the search string, removing duplicated papers, filtering papers by title and abstract, then their full texts, and analyzing their contributions, including strategies and challenges. Finally, we collected and discussed 40 papers on attacks, defense, and privacy.

RESULTS

Our findings identified the most employed strategies for each domain. We found trends in attacks, including universal adversarial perturbation (UAPs), generative adversarial network (GAN)-based attacks, and DeepFakes to generate malicious examples. Trends in defense are adversarial training, GAN-based strategies, and out-of-distribution (OOD) to identify and mitigate adversarial examples (AE). We found privacy-preserving strategies such as federated learning (FL), differential privacy, and combinations of strategies to enhance the FL. Challenges in privacy comprehend the development of attacks that bypass fine-tuning, defenses to calibrate models to improve their robustness, and privacy methods to enhance the FL strategy.

CONCLUSIONS

In conclusion, it is critical to explore security and privacy in ML for health, because it has grown risks and open vulnerabilities. Our study presents strategies and challenges to guide research to investigate issues about security and privacy in ML applied to health systems.

LargeNetVis: Visual Exploration of Large Temporal Networks Based on Community Taxonomies

Temporal (or time-evolving) networks are commonly used to model complex systems and the evolution of their components throughout time. Although these networks can be analyzed by different means, visual analytics stands out as an effective way for a pre-analysis before doing quantitative/statistical analyses to identify patterns, anomalies, and other behaviors in the data, thus leading to new insights and better decision-making. However, the large number of nodes, edges, and/or timestamps in many real-world networks may lead to polluted layouts that make the analysis inefficient or even infeasible. In this paper, we propose LargeNetVis, a web-based visual analytics system designed to assist in analyzing small and large temporal networks. It successfully achieves this goal by leveraging three taxonomies focused on network communities to guide the visual exploration process. The system is composed of four interactive visual components: the first (Taxonomy Matrix) presents a summary of the network characteristics, the second (Global View) gives an overview of the network evolution, the third (a node-link diagram) enables community- and node-level structural analysis, and the fourth (a Temporal Activity Map - TAM) shows the community- and node-level activity under a temporal perspective. We demonstrate the usefulness and effectiveness of LargeNetVis through two usage scenarios and a user study with 14 participants.

The UTrack framework for segmenting and measuring dermatological ulcers through telemedicine

Chronic dermatological ulcers cause great discomfort to patients, and while monitoring the size of wounds over time provides significant clues about the healing evolution and the clinical condition of patients, the lack of practical applications in existing studies impairs users' access to appropriate treatment and diagnosis methods. We propose the UTrack framework to help with the acquisition of photos, the segmentation and measurement of wounds, the storage of photos and symptoms, and the visualization of the evolution of ulcer healing. UTrack-App is a mobile app for the framework, which processes images taken by standard mobile device cameras without specialized equipment and stores all data locally. The user manually delineates the regions of the wound and the measurement object, and the tool uses the proposed UTrack-Seg segmentation method to segment them. UTrack-App also allows users to manually input a unit of measurement (centimeter or inch) in the image to improve the wound area estimation. Experiments show that UTrack-Seg outperforms its state-of-the-art competitors in ulcer segmentation tasks, improving F-Measure by up to 82.5% when compared to superpixel-based approaches and up to 19% when compared to Deep Learning ones. The method is unsupervised, and it semi-automatically segments real-world images with 0.9 of F-Measure, on average. The automatic measurement outperformed the manual process in three out of five different rulers. UTrack-App takes at most 30 s to perform all evaluation steps over high-resolution images, thus being well-suited to analyze ulcers using standard mobile devices.

Segmenting skin ulcers and measuring the wound area using deep convolutional networks

BACKGROUND AND OBJECTIVES

Bedridden patients presenting chronic skin ulcers often need to be examined at home. Healthcare professionals follow the evolution of the patients' condition by regularly taking pictures of the wounds, as different aspects of the wound can indicate the healing stages of the ulcer, including depth, location, and size. The manual measurement of the wounds' size is often inaccurate, time-consuming, and can also cause discomfort to the patient. In this work, we propose the Automatic Skin Ulcer Region Assessment ASURA framework to accurately segment the wound and automatically measure its size.

METHODS

ASURA uses an encoder/decoder deep neural network to perform the segmentation, which detects the measurement ruler/tape present in the image and estimates its pixel density.

RESULTS

Experimental results show that ASURA outperforms the state-of-the-art methods by up to 16% regarding the Dice score, being able to correctly segment the wound with a Dice score higher than 90%. ASURA automatically estimates the pixel density of the images with a relative error of 5%. When using a semi-automatic approach, ASURA was able to estimate the area of the wound in square centimeters with a relative error of 14%.

CONCLUSIONS

The results show that ASURA is well-suited for the problem of segmenting and automatically measuring skin ulcers.

A superpixel-driven deep learning approach for the analysis of dermatological wounds

BACKGROUND

The image-based identification of distinct tissues within dermatological wounds enhances patients' care since it requires no intrusive evaluations. This manuscript presents an approach, we named QTDU, that combines deep learning models with superpixel-driven segmentation methods for assessing the quality of tissues from dermatological ulcers.

METHOD

QTDU consists of a three-stage pipeline for the obtaining of ulcer segmentation, tissues' labeling, and wounded area quantification. We set up our approach by using a real and annotated set of dermatological ulcers for training several deep learning models to the identification of ulcered superpixels.

RESULTS

Empirical evaluations on 179,572 superpixels divided into four classes showed QTDU accurately spot wounded tissues (AUC = 0.986, sensitivity = 0.97, and specificity = 0.974) and outperformed machine-learning approaches in up to 8.2% regarding F1-Score through fine-tuning of a ResNet-based model. Last, but not least, experimental evaluations also showed QTDU correctly quantified wounded tissue areas within a 0.089 Mean Absolute Error ratio.

CONCLUSIONS

Results indicate QTDU effectiveness for both tissue segmentation and wounded area quantification tasks. When compared to existing machine-learning approaches, the combination of superpixels and deep learning models outperformed the competitors within strong significant levels.

Transforming Two Decades of ePR Data to OMOP CDM for Clinical Research

This paper presents the extract-transform-and-load (ETL) process from the Electronic Patient Records (ePR) at the Heart Institute (InCor) to the OMOP Common Data Model (CDM) format. We describe the initial database characterization, relational source mappings, selection filters, data transformations and patient de-identification using the open-source OHDSI tools and SQL scripts. We evaluate the resulting InCor-CDM database by recreating the same patient cohort from a previous reference study (over the original data source) and comparing the cohorts' descriptive statistics and inclusion reports. The results exhibit that up to 91% of the reference patients were retrieved by our method from the ePR through InCor-CDM, with AUC=0.938. The results indicate that the method that we employed was able to produce a new database that was both consistent with the original data and in accordance to the OMOP CDM standard.

dp-BREATH: Heat maps and probabilistic classification assisting the analysis of abnormal lung regions

BACKGROUND AND OBJECTIVE

Identifying abnormalities in chest CT scans is an important and challenging task, demanding time and effort from specialists. Different parts of a single lung image may present both normal and abnormal characteristics. Thus, detecting a single lung as healthy (normal) or not is inaccurate.

METHODS

In this work we propose dp-BREATH, a method capable of detecting abnormalities in pulmonary tissue regions and directing the specialist's attention to the lung region containing them. It starts by highlighting regions that may indicate pulmonary abnormalities based on the healthy pulmonary tissue behavior using a superpixel-based approach and a heat map visualization. This is achieved by modeling regions of healthy tissue using a statistical model. All regions considered abnormal are modeled and classified according to their probability of containing each of the studied abnormalities. Further, dp-BREATH provides a better recognition of radiological patterns, with the likelihood of a selected lung region to contain abnormalities.

RESULTS

We validate the statistical model of healthy and abnormal detection using a representative dataset of chest CT scans. The model has shown almost no overlap between healthy and abnormal regions, and the detection of abnormalities presented precision higher than 86%, for all recall values. Additionally, the fitted models describing pulmonary radiological patterns present precision of up to 87%, with a high separation for three of five radiological patterns.

CONCLUSIONS

dp-BREATH's heat map representation and its list of radiological patterns probabilities provided are intuitive methods to assist physicians during diagnosis.

Employing Domain Indexes to Efficiently Query Medical Data From Multiple Repositories

Content-based retrieval still remains one of the main problems with respect to controversies and challenges in digital healthcare over big data. To properly address this problem, there is a need for efficient computational techniques, especially in scenarios involving queries across multiple data repositories. In such scenarios, the common computational approach searches the repositories separately and combines the results into one final response, which slows down the process altogether. In order to improve the performance of queries in that kind of scenario, we present the Domain Index, a new category of index structures intended to efficiently query a data domain across multiple repositories, regardless of the repository to which the data belong. To evaluate our method, we carried out experiments involving content-based queries, namely range and k nearest neighbor (kNN) queries, 1) over real-world data from a public data set of mammograms, as well as 2) over synthetic data to perform scalability evaluations. The results show that images from any repository are seamlessly retrieved, sustaining performance gains of up to 53% in range queries and up to 81% in kNN queries. Regarding scalability, our proposal scaled well as we increased 1) the cardinality of data (up to 59% of gain) and 2) the number of queried repositories (up to 71% of gain). Hence, our method enables significant performance improvements, and should be of most importance for medical data repository maintainers and for physicians' IT support.

PRoSPer: perceptual similarity queries in medical CBIR systems through user profiles

In this paper, we present a novel approach to perform similarity queries over medical images, maintaining the semantics of a given query posted by the user. Content-based image retrieval systems relying on relevance feedback techniques usually request the users to label relevant/irrelevant images. Thus, we present a highly effective strategy to survey user profiles, taking advantage of such labeling to implicitly gather the user perceptual similarity. The profiles maintain the settings desired for each user, allowing tuning of the similarity assessment, which encompasses the dynamic change of the distance function employed through an interactive process. Experiments on medical images show that the method is effective and can improve the decision making process during analysis.

Using relevance feedback to reduce the semantic gap in content-based image retrieval of mammographic masses

This paper presents the use of relevance feedback (RFb) to reduce the semantic gap in content-based image retrieval (CBIR) of mammographic masses. Tests were conducted where the radiologists' classification of the lesions based on the BI-RADS categories were used with techniques of query-point movement to incorporate RFb. The measures of similarity of images used for CBIR were based upon Zernike moments. The performance of CBIR was measured in terms of precision and recall of retrieval. The results indicate improvement due to RFb of up to 41.6% in precision. In our experiments, the gain in the performance of CBIR with RFb was associated with the BI-RADS category of the query mammographic image, with large improvement in cases of lesions belonging to categories 4 and 5. The proposed method could find applications in computer-aided diagnosis (CAD) of breast cancer.

Using an image-extended relational database to support content-based image retrieval in a PACS

This paper presents a new Picture Archiving and Communication System (PACS), called cbPACS, which has content-based image retrieval capabilities. The cbPACS answers range and k-nearest- neighbor similarity queries, employing a relational database manager extended to support images. The images are compared through their features, which are extracted by an image-processing module and stored in the extended relational database. The database extensions were developed aiming at efficiently answering similarity queries by taking advantage of specialized indexing methods. The main concept supporting the extensions is the definition, inside the relational manager, of distance functions based on features extracted from the images. An extension to the SQL language enables the construction of an interpreter that intercepts the extended commands and translates them to standard SQL, allowing any relational database server to be used. By now, the system implemented works on features based on color distribution of the images through normalized histograms as well as metric histograms. Metric histograms are invariant regarding scale, translation and rotation of images and also to brightness transformations. The cbPACS is prepared to integrate new image features, based on texture and shape of the main objects in the image.

DBM-Tree: Trading height-balancing for performance in metric access methods

Metric Access Methods (MAM) are employed to accelerate the processing of similarity queries, such as the range and the k-nearest neighbor queries. Current methods, such as the Slim-tree and the M-tree, improve the query performance minimizing the number of disk accesses, keeping a constant height of the structures stored on disks (height-balanced trees). However, the overlapping between their nodes has a very high influence on their performance. This paper presents a new dynamic MAM called theDBM-tree (Density-Based Metric tree), which can minimize the overlap between high-density nodes by relaxing the height-balancing of the structure. Thus, the height of the tree is larger in denser regions, in order to keep a tradeoff between breadth-searching and depth-searching. An underpinning for cost estimation on tree structures is their height, so we show a non-height dependable cost model that can be applied for DBM-tree. Moreover, an optimization algorithm calledShrink is also presented, which improves the performance of an already builtDBM-tree by reorganizing the elements among their nodes. Experiments performed over both synthetic and real world datasets showed that theDBM-tree is, in average, 50% faster than traditional MAM and reduces the number of distance calculations by up to 72% and disk accesses by up to 66%. After performing the Shrink algorithm, the performance improves up to 40% regarding the number of disk accesses for range andk-nearest neighbor queries. In addition, theDBM-tree scales up well, exhibiting linear performance with growing number of elements in the database.