Protocol non-adherence in teledermatology, particularly during image acquisition, such as missing rulers or incorrect image orientations, leads to invalid examinations, delayed diagnoses, and increased patient burdens. This study addresses these challenges by evaluating modern neural network architectures for instance segmentation and proposing a novel orientation correction pipeline to improve adherence in two protocols: Approximation (identifying rulers/patient tags) and Panoramic (correcting body orientation errors). Using the Santa Catarina State Telemedicine System dataset (14,238 images for Approximation; 3,692 for Panoramic), models were benchmarked against previous study measuring computational efficiency and precision metrics. Results demonstrate YOLOv11's superiority, achieving state-of-the-art performance with$86.08 \text{AP}_{75}$, reducing segmentation errors by 14 % compared to Mask R-CNN, while maintaining computational efficiency (1.3 GB VRAM, 566-699 ms latency). For panoramic protocol images, our proposed pipeline mitigated orientation errors by methodically rotating misaligned human masks, improving weighted$F$-scores from 0.51 to 0.82 and significantly reducing misclassifications between valid and invalid poses. While Mask2Former's transformerbased design architecture exhibited higher precision, the computational demands (11.7 GB VRAM) hindered deployability in low-resource clinics. This study concludes that hybrid models like YOLOv11 optimally balance segmentation accuracy and operational efficiency, offering actionable insights for real-world clinical implementation bridging AI advancements with clinical pragmatism, improving protocol adherence to reduce invalid examinations by addressing orientation inconsistencies in teledermatology.

The increasing demand for accurate and efficient cytopathological diagnostics at early stages highlighted the need for computational solutions of handling real-world clinical variability. Our study evaluates state-of-the-art convolutional and transformer-based models on three distinct cytological stains:$\text{Pa}{-}$panicolaou, Feulgen and AgNOR. By comparing models trained on single-stain datasets with those trained on combined multistain dataset, we demonstrate that it improves generalization and operational efficiency.

While routine macroscopic placental assessment is performed in all deliveries, anatomopathological examination is reserved for selected cases due to logistical and resource constraints. Although artificial intelligence (AI) has shown promise in medical imaging, its application to postpartum placental screening remains underexplored in clinical settings. This study addresses this gap by developing a deep learning pipeline to improve segmentation of placental structures using macroscopic photographs captured in delivery rooms, with the aim of supporting anatomopathological referral decisions. A proprietary dataset was constructed under both standardized and routine clinical conditions, with annotations curated by pathologists across eight morphological classes. Three architectures—YOLOv11, U-Net with ResNet34, and U-Net with EfficientNet-B0—were evaluated for segmentation and classification performance. Results demonstrated moderate efficacy: YOLOv11 achieved an AP50 of 72% for instance segmentation, while U-Net variants showed higher classification accuracy (up to 85.18%) but lower segmentation precision (Dice scores: 24.70–61.97%). These findings support the feasibility of incorporating AI-based photographic analysis into routine clinical workflows for placental evaluation. Despite preliminary segmentation limitations, this work establishes a foundation for AI-driven placental screening, offering potential to optimize resource allocation by guiding anatomopathological referrals.

Postpartum placental assessment is essential for clarifying adverse pregnancy outcomes and informing clinical decisions; however, anatomopathological examination is generally reserved for selected cases due to structural and operational constraints. This study presents a deep learning-based pipeline for segmenting placental structures in real images taken in the delivery room, aiming to support clinical triage. A proprietary dataset was built using a standardized photographic protocol and annotated by pathologists across nine morphological classes. Five architectures were evaluated: U-Net with ResNet34, ResNet50, EfficientNet-B0, and EfficientNet-B7 backbones, in addition to YOLOv11 for instance segmentation. ResNet34 achieved the best overall performance (Dice: 81.2%, IoU: 70.6%, Accuracy: 85.3%), while YOLOv11 reached a AP50 of 73.2% in detecting key anatomical components. Despite the limitation imposed by the small dataset—which may affect the models’ generalization capability—the results demonstrate the feasibility of using AI for photographic placental triage, with potential to assist clinical decisions and optimize resource use in obstetric settings.

Planning local trajectories for nonholonomic vehicles in real-time that can converge, be feasible, and be smooth simultaneously is a key challenge in autonomous ground vehicle navigation. Many techniques have attempted to overcome this challenge by partially solving a subset of these goals. We propose an approach based on first allowing multiple local path planning algorithms to run in parallel and then choosing the right one based on selection criteria. This allows the use of different specialized types of local planning solutions to account for the execution time, convergence, and path smoothness. Simulations show that the parallel ensemble of local planners performs better than each of its composing local path planners running independently in terms of those goals. Our solution allows the vehicle’s behavior to adapt to different scenarios by automatic local planner selection, which maintains low system complexity and increases adaptability. Moreover, we present an efficient algorithm for local goal approximation of global goals that integrates sub-planning tasks as a single long-reach planning response, enabling the Ensemble to navigate unknown unstructured scenarios.

This systematic literature review investigates the state of the art in the application of artificial intelligence (AI), particularly convolutional neural networks (CNNs), in the diagnosis of pneumoconioses and asbestos-related diseases (ARDs). A total of 30 articles published between 2020 and 2025 were analyzed, selected from major scientific databases (IEEE Xplore, ScienceDirect, Springer Link, ACM Digital Library, Nature, Wiley Online Library). The analysis addressed the models used types of radiological images (chest X-rays and computed tomography), performance metrics, and limitations. A significant advancement was observed in the use of CNNs and 3D architectures, with an emphasis on automated screening and the interpretability of clinical patterns.

Digital Inline Holographic Microscopy (DIHM) is a lensless shadow imaging technique where a coherent light-source is utilized to illuminate samples and record interference patterns in a digitalizing device. Those patterns, called holograms, carry volumetric information regarding the inspected sample, requiring complex numerical methods to make them distinguishable as occurs in a conventional bright-field microscopic image. However, for many applications, the real-time analysis of holograms is computationally expensive due to the nature of numerical diffraction methods to reconstruct the signals into visual information. To mitigate this problem, in this paper we investigate the use of deep learning approaches to classify those interference patterns, directly from the raw holograms, without the requirement of phase-recovering methods for diffraction. In our approach, we investigated the use of distinct Convolutional Neural Networks (CNNs) architectures and its adaptability to correctly classify holograms in an experimental environment with a dataset generated from synthetic interference patterns produced by the Fresnel Diffraction Method. The computer-generated dataset was produced from 26 classes, resulting in a total of 520 samples after the data augmentation procedure. The obtained results demonstrated the feasibility of the proposed approach to properly classify samples with 96.8% of precision, directly from the holographic interference patterns, avoiding the need for computationally expensive diffraction methods.

Cervical cancer is the second most common cancer type in women. This is a deadly disease that could benefit from early detection methods. Cytology is a possible, noninvasive, alternative to biopsy for diagnosis of malignant lesions, where information about the cells can be obtained. This process consists of staining the slides being able to visualize the nuclei of the cells and then be properly diagnosed. In this work, silver-stained slides, a method known as Argyrophilic Nucleolar Organizer Regions (AgNOR), is employed. This method shows great potential for the diagnosis of the malignancy of lesions. However, it has not yet been much explored, mainly using computational methods. In this work, we propose the use of Convolutional Neural Networks on silver stained images to measure AgNOR quantitative image information. We will present the entire process, starting from pre-processing, passing through neural network training and results evaluation. The best results we obtained were 0.87 of Mean Intersection Over Union (mIoU) and 0.99 of Dice Similarity Coefficient (DSC) using an U-Net model with a ResNet18 as its backbone. The code and the dataset employed in this work are publicly available.

Although oral cancer is considered a global health issue with 350.000 people diagnosed over a year it can successfully be treated if diagnosed at early stages. Papanicolaou is an unexpensive and non-invasive method, generally applied to detect cervical cancer, but it can also be useful to detect cancer on oral cavities. The manual process of analyzing cells to detect abnormalities is a time-consuming cell analysis and is subject to variations in perceptions from different professionals. This paper compares models for three different deep learning approaches: segmentation, object detection and image classification. Our results show that the binary object detection with Faster R-CNN is the best approach for nuclei detection and localization (0.76 IoU). Since ResNet 34 had a good performance on abnormal nuclei classification (0.88 accuracy, 0.86 F_1 score) we concluded that these two models can be used in combination to make a localization and classification pipeline. This work reinforces that the automated analysis of oral cytology to make a pipeline for nuclei classification and localization using deep learning can help to minimize the subjectivity of the human analysis and also to detect cancer at early stages.

A structured reporting system is a set of standard-compliant practices that guides the physicians in order to create structured documents, organizing and simplifying their workload. In spite of this advantages, many physicians still use conventional methods to write their findings, like free-text reports. One main reason for this is the little convention on how to disseminate a structured reporting routine in the report environment. This work proposes a systematic approach to migrate a system routine from free-text reports to structured reports, focusing on the DICOM Structured Reporting guidelines. To evaluate this proposal, a structured reporting system was created in the context of the Santa Catarina State Integrated Telemedicine and Telehealth System (STT/SC), in Brazil, and, in a case study covering obstetric ultrasonography reports, was evaluated by a group of experts using the AdEQUATE model, showing a high user perception from the system. The results are a set of defined premises and steps that turns a telemedicine system into a complete structured reporting environment.

Telepathology systems (TPS) can improve and accelerate the process of diagnosis, particularly in those complex cases where the second expert/specialized opinion is required. To support scenarios involving complex cases, a TPS must provide collaborative virtual microscopy and communication between pathologists to accelerate and improve the process of reporting, while guaranteeing a good degree of image quality in order to enable the evaluation of microscopic details. The development of TPSs in large telemedicine environments has not yet been addressed properly. With this problematic in mind, we developed a TPS within the Santa Catarina State Integrated Telemedicine and Telehealth System (STT/SC), that allows pathologists to report pathology cases employing a collaborative environment with audio/video conferencing and that simulates a virtual multi-headed microscopy. To prove the concept of our system we performed a case study with pathologists from different institutions countrywide and applied an evaluation using the AdEQUATE Model. Our system scored good results, especially regarding its effectiveness, learnability, and comfort.

Telemedicine and Telehealth Stations (TTSs) are used to provide physicians with tools that simplify their workload, increasing the range of areas covered by the services offered by those kinds of equipments. To support this scenario, those stations must have a high external quality in order to improve user experience and accession and to speed up their workflow due to an easy-to-use platform. Based on that, this work presents an external quality evaluation for a developed TTS in the context of the State Integrated Telemedicine and Telehealth System (STT/SC). To accomplish that, we applied the AdEQUATE model, using descriptive statistics to analyze it's results and perceived a good quality system, especially for characteristics like comfort, authenticity and learnability.

DICOM structured reports (DICOM SR) are a set of standard-compliant practices dedicated to provide a pragmatic way to create documents when medical data organization and storage are an issue. In practice, it can be seen as a set of guidelines that can be used by physicians in order to create report templates that can be manipulated, filled and stored as structured data. Nowadays, systems that implements DICOM SR are used to provide physicians with tools that facilitate the report construction by organizing and simplifying their workload. To support scenarios involving features related to the creation and manipulation of structured documents, those systems must have a high-quality degree. While the external quality degree is important to secure a good user perception and adhesion to the system, having a high internal quality degree is a must to ensure external quality, functionality, reliability, maintainability and performance. Aiming to evaluate such scenario, this work presents an ISO compliant internal quality evaluation for the structured report system of the Santa Catarina State Integrated Telemedicine and Telehealth System (STT/SC), in Brazil. Using the ISO/IEC 9126-3 norm to do the evaluation, we concluded that the system has a high internal quality degree, regarding usability, maintainability, efficiency, reliability and functionality as dictated by the ISO/IEC 9126-1 norm.

In the literature, PACS and Telemedicine are located in different or complementary works. However, when integrated as part of the same solution, both contribute to the healthcare process as a whole. This work describes the experience on integrating a PACS infrastructure to a statewide telemedicine system, allowing to the latter to benefit from concepts defined by the former in terms of acquisition, transmission and storage of digital images. The obtained results attest the viability of such approach, allowing the construction of a robust, scalable, and secure solution attending scheduled procedures, image acquisition, content transmission, content storage, content evaluation, and the return of examination reports.

Poison Control Centres (PCCs) provide information and advise about poisoning, poisoning case management, and toxicovigilance activities. They are supported by information systems that help physicians and toxicologists to attend cases of patient exposure to poisonous agents. Still, there is no consensus on how to evaluate software quality of such systems. This work performs a case study applying a customized software quality evaluation model -- the AdEQUATE model -- in order to assess the software quality of a poison control information system widely used in Brazil - the DATATOX information system. Analysing the specificities of DATATOX we customized AdEQUATE, a model for evaluating the quality of telemedicine systems based on the ISO/IEC 25010 standard through the use of a questionnaire. The questionnaire has been answered by 49 end users, and the acquired data were analysed employing descriptive statistics. Overall, the DATATOX information system has been perceived as having a good quality, specifically for characteristics like effectiveness, efficiency, and satisfaction. Nevertheless, few characteristics have been considered of low quality, including reliability, compatibility, and age. As a complement, the work attest the feasibility on using AdEQUATE's questionnaires as an evaluation tool for software quality evaluation.

A laboratory information system (LIS) manages information about patients, examinations, results, and quality control. In this context, software quality is important in order to prevent risks for the patient. Studies show that many LISs have a low quality degree, and although the ISO/IEC 25010 standard defines software quality in general, so far there is no specific customization for a quality evaluation model tailored for such systems. This paper presents a case study applying a customized evaluation model - the AdEQUATE model - in order to evaluate the software quality of a LIS widely used in the State of Santa Catarina, Brazil - the LACEN system - from the perspective of its end users. Analyzing the characteristics of the LACEN system, we customized the AdEQUATE model - a model for quality evaluation of telemedicine systems based on the ISO/IEC 25010 standard through the use of a questionnaire. The questionnaire has been answered by 69 users (including physicians and biochemists), and its answers were analyzed using descriptive statistics. In general, the quality of the LACEN system has been considered as good, specifically for characteristics like effectiveness, efficiency, satisfaction, freedom from risk, functional suitability, and security. On the other hand, few characteristics have been considered of low quality, including performance efficiency and compatibility. Also, the collected data shows that it is feasible to use AdEQUATE's systematically derived questionnaire for detailed observations of quality degrees of the ISO/IEC 25010 standard characteristics and subcharacteristics.

Telemedicine and Telehealth are disciplines dedicated to provide health care and health care education when distance is an issue. Nowadays, Telemedicine and Telehealth Stations (TTSs) are used to provide physicians and educators with tools that concentrate and integrate different technologies, contributing to organize and simplify their workload. To support scenarios involving features related to teleconsultation and tele-education, TTSs must consider the adoption of hardware platforms robust enough to reach minimal requirements while providing service deliverability and continuity. Despite being usually built for execution on desktops and laptops, it is possible that TTSs be reorganized to use Single-Board Computers (SBCs) as a most portable-and cost-effective solution. Aiming to evaluate such possibility, this work presents a performance study of four of the most popular SBCs available on the market, focusing the execution of web conferences in the context of the Santa Catarina State Integrated Telemedicine and Telehealth System (STT/SC), in Brazil. Using descriptive statistics, we concluded that the Raspberry Pi 2 SBC is the best option to be adopted, considering its results regarding CPU consumption, memory utilization, and video delay.

Panretinal Laser Photocoagulation (PRP) using the slit-lamp is a widely used treatment for sight-threatening retina diseases. It is a challenging procedure, therefore a number of mapping techniques have been proposed to provide surgeons with reliable intra-operative navigation assistance and mapping. However, available methods are limited by tracking errors, caused by lens distortions, illumination variations and abrupt patient movement. To handle these issues, we propose a strategy for non-rigid fine adjustment of retina maps acquired using the slit-lamp. We describe a novel direct framework for alignment of multiple overlapping images via a Thin-Plate Spline model. Our method is able to cope with illumination variations between overlapping images and does not require a priori geometrical information. Experiments conducted on maps acquired from several human patients attest the method's ability to correct ghosts and blurring, improving the consistency and accuracy of the maps.

Histopathological staining is a technique widely used to highlight biological states or pharmacological activities in uman tissue. A quantitative analysis of the resulting images can produce biomarkers for diseases or even other specific conditions, thus providing valuable information for diagnosis and prognosis. Since biomarkers require measurements to be made in an objective and consistent way, software systems are employed to provide this quantitative analysis. For measurements to be reproducible, the same methods must be available across different laboratories. In this paper, we present a tool that allows users to perform quantitative analyses over the web, thus providing an efficient environment not only for individual cases to be evaluated, but also for users to share a common ground when making measurements. The classification method used by the tool to segment stained pixels is performed by a similarity function based on the polynomial version of the Mahalanobis distance, which is nonlinear and provides very robust classification for m-dimensional feature spaces. Furthermore, the similarity function can be generalized in the tool, so that images can be classified by reusing parameters of previous cases. The results of our web-based approach were compared with established ground-truth data sets, producing sensitivity, specificity and fitness values of 97.09%, 98.70%, and 97.90%.

This article introduces a multi-level visualization scheme for poetry. In each level, different elements are represented and their attributes are coded in different ways. The first level displays the phonemes in a verse, tracing them back to the written word. The second level shows the poem itself in a metric grid where the verse syllables are placed. The third level shows a set of poems whose verses and syllables can be visually compared. Finally, the fourth level is capable of showing hundreds or thousands of verses at once, turning each verse into a tile whose color indicates a meaningful property. We argue that each level has its strengths and its weaknesses and we evaluate them with our collaborators from the poetry community.

Nowadays the studies of how to store massive amounts of data is a constant concern for the entire scientific community. In this paper is presented a data storage approach for DICOM medical image in distributed environments, using the distributed file systems CEPH, FhGFS, PVFS and Lustre versus a traditional RDBMS. The main contribution behind this work is open an important discussion about the research for better long-term data persistence methods, improving the management levels and still, bringing practical benefits to the medical systems. Beyond that, the experiments results have show that our approach using a distributed file system CEPH had the best performance to storage 1000 and 2500, reaching up to 907.49% better than a conventional approach using RDBMS.

This work presents a systematic review of state-of-the-art usability heuristics in the area of clinical laboratories, and compiles evaluates usability heuristics customized for clinical laboratory systems by unifying and mapping them. In addition, the resulting set of heuristics is evaluated through a case study comparing results from the application of heuristics and usability tests.

This work presents an information retrieval architecture developed for the Santa Catarina State Telemedicine System. This architecture employs DICOM Structured Reporting, controlled vocabularies for data catalogization and a specially developed search engine for data indexing and storing. Results of our case study show that searches can be performed much faster with the proposed search mechanism and that the precision of results is acceptable in most cases. In some searches, irrelevant items within the 15 first results were identified. This occurred partially because search terms found in the additional free text observations inserted into the findings reports were treated with the same relevance as formally diagnostically relevant items of the DICOM SR structure and, partially because the semantics of negations associated to search terms in the findings reports were not taken into consideration.

Fiber tracking is a mature technique that aids neurosurgeons by providing the location of neuronal fibers in a patient's brain. Although the interactivity of tools have been typically limited, the exploitation of the parallel nature of fiber tracking is rapidly changing this landscape. The parallel environment offered by GPUs and modern CPUs enables fiber tracking to be executed very quickly on consumer-grade computers. The adjustment of fiber tracking parameters can currently be made on the fly, and the new results can be both computed and displayed in real-time on the screen. In order to further improve the interactivity of fiber tracking tools, this article proposes the application of histogram equalization, a traditional digital image processing procedure, to the fiber tracking parameters. The non-linear scale obtained by this process allows the user to interactively explore sensitive regions of the parameter space. Results show that this scale can be computed in real-time and that it can be cleanly integrated into fiber tracking tools.

This paper presents a technological solution with the purpose of collaborative medical image visualization with support of decision in telemedicine environments. A fully web-based modular platform was developed which contains visualization, marking and manipulation synchronized between multiple users with communication established through video call or text chat. Pure HTML5 was used ensuring the portability of the tool, meaning that any device with a modern web browser and internet connection can access it. Finally, the platform usability is confirmed by the excellent score resulted from a System Usability Scale evaluation.

The management of data acquired in a daily basis, together with historical data kept available during variable time frames, became a practical concern to the healthcare practice. Data organized according to the DICOM (Digital Imaging and Communications in Medicine) standard contribute to this scenario, demanding solutions for storage and retrieval of metadata and images that lie beyond the customary file system approach. As a way of experimentation, this work defines and implements a data model using a NoSQL, row-store database, aiming to evaluate its suitability on managing DICOM content at tag level, as well as its performance when compared to a commonly approach based on a relational schema. The obtained results attest the capacity of the proposed model in storing full-content DICOM images, using tags as data units. In terms of performance, however, the data model is outperformed by the relational approach.

The slit lamp device is a very popular equipment for inspecting the human retina due to its magnification, stereoscopy and easy control. In this context, computer assistance can provide information overlay to surgeons, improving navigation, accuracy and safety in procedures where the slit-lamp is used. Toward this goal, we describe a robust method for tracking and mapping the human retina using slit-lamp images. Inspired in [1], the proposed method is essentially a direct visual tracking method using local illumination compensation, as well as an efficient pixel selection scheme for achieving fast tracking speeds. In this paper, we describe a novel forward-backward strategy for achieving superior tracking resilience. Experiments conducted on several human patients confirm the practical value of the system.

The Santa Catarina State Telemedicine and Telehealth System — STT/SC stores well over 2 million examinations and every month about 20 thousand new imaging exams are sent to the system database. As a significant part of the findings associated to these medical data are stored as text documents, the precise extraction of information is a difficult task. In the healthcare domain it is common to find different terms, some of them appearing as composed expressions, used to represent the same concept, whereas the existence of simple modifiers can turn an expression into another, different concept. This work presents an information retrieval architecture developed for the STT/SC. It consists in a module integrated into the STT/SC and was developed in order to support both experienced and inexperienced users to perform queries. Experiments were performed to evaluate the accuracy of the proposed system in real situations. Results show that the search can be performed more much faster and with acceptable precision.

Gradient-based optimization is a very efficient strategy to solve the direct visual tracking (DVT) problem using transformation models with many degrees of freedom (DOF). Even though popular DVT methods use the sum of squared differences as similarity function, this approach is not robust to illumination variations often verified in practice. One technique to compensate illumination variations is through an illumination model, which, in turn, increases the total number of parameters to be computed. High quality augmented reality and robotic systems demand fast tracking speeds, which can be impaired by the computational complexity added by the illumination model. In this paper, we propose a robust DVT method capable of tracking in extreme illumination conditions. Building upon the sum of conditional variance as similarity function, we propose a novel tracking method that significantly reduces the computational effort compared to similar methods proposed in the literature. We provide extensive experiments and quantitative analysis using challenging videos to attest the advantages of the proposed method.

OBJECTIVE: To design, build, and evaluate a storage model able to manage heterogeneous digital imaging and communications in medicine (DICOM) images. The model must be simple, but flexible enough to accommodate variable content without structural modifications; must be effective on answering query/retrieval operations according to the DICOM standard; and must provide performance gains on querying/retrieving content to justify its adoption by image-related projects. METHODS: The proposal adapts the original decomposed storage model, incorporating structural and organizational characteristics present in DICOM image files. Tag values are stored according to their data types/domains, in a schema built on top of a standard relational database management system (RDBMS). Evaluation includes storing heterogeneous DICOM images, querying metadata using a variable number of predicates, and retrieving full-content images for different hierarchical levels. RESULTS AND DISCUSSION: When compared to a well established DICOM image archive, the proposal is 0.6-7.2 times slower in storing content; however, in querying individual tags, it is about 48.0% faster. In querying groups of tags, DICOM decomposed storage model (DCMDSM) is outperformed in scenarios with a large number of tags and low selectivity (being 66.5% slower); however, when the number of tags is balanced with better selectivity predicates, the performance gains are up to 79.1%. In executing full-content retrieval, in turn, the proposal is about 48.3% faster. CONCLUSIONS: DCMDSM is a model built for the storage of heterogeneous DICOM content, based on a straightforward database design. The results obtained through its evaluation attest its suitability as a storage layer for projects where DICOM images are stored once, and queried/retrieved whenever necessary.

Laser photocoagulation is currently the standard treatment for sight-threatening diseases worldwide, namely diabetic retinopathy and retinal vein occlusions. The slit lamp biomicroscope is the most commonly used device for this procedure, specially for the treatment of the eye periphery. However, only a small portion of the retina can be visualized through the biomicroscope, complicating the task of localizing and identifying surgical targets, increasing treatment duration and patient discomfort. In order to assist surgeons, we propose a method for creating intraoperative retina maps for view expansion using a slit-lamp device. Based on the mosaicking method described by Richa et al, 2012, the proposed method is a combination of direct and feature-based methods, suitable for the textured nature of the human retina. In this paper, we describe three major enhancements to the original formulation. The first is a visual tracking method using local illumination compensation to cope with the challenging visualization conditions. The second is an efficient pixel selection scheme for increased computational efficiency. The third is an entropy-based mosaic update method to dynamically improve the retina map during exploration. To evaluate the performance of the proposed method, we conducted several experiments on human subjects with a computer-assisted slit-lamp prototype. We also demonstrate the practical value of the system for photo documentation, diagnosis and intraoperative navigation.

The need for procedures and methodologies used in clinical information standardization is continuous, requiring the current telemedicine networks to evolve and adapt by using mature standards and guidelines as the basis for clinical documents transmitted through the network. In this work, we present a different approach for large-scale use of DICOM SR in telemedicine networks using structured vocabularies as its foundation, resulting in a reliable, efficient and data mining ready application applied in a real world telemedicine network. This approach is applied in a large-scale network where 60 physicians report daily an average of 400 reports; resulting in more than 80,000 reports written and stored in the last nine months. We conclude that the use of this approach can reduce eventual syntax or conceptual mistakes by using proper structured vocabularies in addition to DICOM SR.

We present a method to efficiently simulate fluid flows through tubular structures such as pipes and blood vessels. The method relies on the structure's centerline and for that purpose we present an algorithm to generate an approximate centerline of a tubular structure defined by an ordinary polygonal mesh. As an additional benefit, this center-line provides approximations of measures about the tubular structure such as length of regions of interest and cross-section diameters. The fluid is simulated with the smoothed particle hydrodynamics method and the fluid-structure interaction is simplified to provide interactive rates. The particles can be colored according to physical quantities to better visualize the evolution of the fluid's state in narrow and otherwise obstructed areas.

Computer aided diagnostic comprises wide research areas with stability and reliability software requirements. We propose a set of libraries aiming to provide a development framework for medical applications. This framework's libraries, mostly written from scratch (e.g Core library), were developed for both medical imaging systems and general physical simulations. Their internal interactions tend to be transparent to the developer and easy to use at the same time while providing flexibility to interact with other codes. The robustness of the code is shown in complex applications like the medical workstation here exemplified, while its flexibility and fast prototyping characteristics are exemplified with some simple demonstrations.

This paper presents a semiautomatic method for the identification of immunohistochemical (IHC) staining in digitized samples. The user trains the system by selecting on a sample image some typical positive stained regions that will be used as a reference for the construction of a distance metric. In this learning process, the global optimum is obtained by induction employing higher polynomial terms of the Mahalanobis distance, extracting nonlinear features of the IHC pattern distributions. The results of the proposed method showed a high correlation to a pathologist's manual analysis, which was used as a golden standard, presenting a more robust discrimination between stained and non-stained areas with little bias.

Fiber tracking is a technique that determines the location of fiber bundles in a human brain. Sequential implementations of fiber tracking usually require a large execution time, which hinders both the interactivity currently experienced by physicians and its applications to large collections of data sets. In this paper we present the design and implementation of fiber tracking applications in two different parallel environments, GPUs and PC clusters. The performance results obtained in our experiments show that both HPC configurations are suitable for fiber tracking: GPUs for interactive applications targeted at physicians and PC clusters for batch processing by fiber tracking researchers.

Telemedicine systems currently have increased the volume of information stored in their databases. A centralized telemedicine system project covers datasets that ranges from personal patient's information, physicians and institutions to all images of examinations performed. As they store large amounts of examinations, the medical databases can reach several terabytes of volume. This paper presents a contribution characterized by an asynchronous replication model for medical distributed databases. The model, called postgresreplication, is an extension to the relational database postgreSQL. In our experiment, an engine has been created to manage all integration operations and information replication for the medical databases. The early results indicate that the model and its implementation have successfully reached a good performance level and interoperability.

This paper reports the development and design of the State of Santa Catarina's telemedicine network in Brazil. The resources concentration, like hospitals and clinical staff, in Brazilian large cities have been a problem in public healthcare policies improvement. Telemedicine technology for large scale telediagnostic, processing of routine outpatient examinations, the electronic delivery of examinations results integrated with the decision process of wetter to provide further treatment for a patient, is one strategy to overcome the difficulties imposed by healthcare concentration. The State needs to reduce costs like patient transportation, improve the quality of healthcare service and the origins control of these examinations, motivated the creation of this medical knowledge network. As a result it promoted better patient care, making faster diagnosis, creating a patient information history, reducing the examinations redundancy, maximizing this way, the social welfare and the health technological park on the Santa Catarina State.

Nowadays, there is a growing interest in telemedicine in Brazil, because the country is the fifth largest country in the world and there is a huge gap between the population size and the number of available doctors. Institutions and the government are considering solutions for the integration and availability of data produced by several different systems. In this paper we present a contribution that is characterized by an asynchronous replication strategy between distributed medical databases. The contribution, named PGR, is an enhancement to the PostgreSQL database software that introduces lightweight asynchronous operations. In addition, we have implemented a multi-master engine with partial replication and hybrid fragmentation in order to allow interoperability between different telemedicine systems. Our early results indicate that the model and its implementation have reached a successful level in terms of performance and interoperability.

The following article presents a computational methodology to assist the assessment of tympanic pathologies, particularly chronic otitis media, based on digital otoscopic images. A computational system which implements the methodology is described. This approach does not impose additional effort to the acquisition protocol and diagnosis of tympanic pathologies. In preliminary tests, it has shown precision and easiness of use. The presented methodology and system were tested by the Otitis Media Center of Brazil (Com.Br) with a case base of 660 patients, resulting in 1320 bilateral tympanic images.

The lack of content-based retrieval models capable to provide fast and relevant results is one of the reasons why CBMIR tools have not yet became present in radiological routines. This work presents a methodology for CBMIR in order to help solve this deficiency. Tests performed have shown a reduction of 95.5% of the execution time for retrieval processes. The relevance of the results is assured by the categorization of the cases in the Case Base. The class of cases is defined through pseudo-semantics and visual attributes.

We discuss the use of glove-driven interface in radiological workstations and present a prototype that aims to integrate common functions such as virtual manipulation and navigation control, with a basic gesture interface. The fast rate of and numerical data acquired by glove-driven interfaces, allows a much broader set of functionalities to be implemented in a highly demanding tridimensional visualization application. The prototype provides the basic gesture interface using a 5DT DataGlove5 Ultra and a Flock of Birds tracker, combining independent hand posture and trajectory recognition to send command messages to the host application. Using a synchronized state machine in the host application and the prototype, the gestures can activate special menus, and call up specific function or screens.

This work presents a new generic application protocol for DICOM conformant medical collaborative systems. It implements simple techniques for data flow and encapsulation between two or more clients and a single server, using an extension of the DICOM Communication Layer within a server or a client as a session manager. The protocol was tested working over two image viewer clients remotely distributed and the synchronization level shows that collaborative systems can be executed in real time.

This paper presents a new methodology to be used in measuring brain atrophy on imaging exams of patients with neurodegenerative diseases. The methods foresee the usage of a mean brain that serves as a reference volume. Color maps were created following a cold-hot color scheme where highlighted regions of atrophy based on the standard deviation and mean values voxel-wise of the mean brain. Direct measures of the atrophy grade were obtained by statistical comparisons between the mean brain volume and regions identified as atrophied. The proposed method was evaluated using 35 patients, 16 had been previously diagnosed as Alzheimer cases, 7 as MCI, 2 as vascular dementia and 10 as complainers. The mean brain used in this study was created using affine transformations to drive the necessary volume registration and were composed of healthy patients. All diagnostics followed a neuropsychological protocol

This paper presents a method for arterial diameter measurement for the patient customization of endovascular prosthesis and for the assistance in preoperative procedures. The use of three-dimensional reconstruction and organ representation through a traversable geometric representation, known as Doubly Linked Face List (DLFL), which guarantees some topological properties, allows the measurement of the diameter of the polygons formed by the intersection between any arbitrary plane and the organ to be performed in a very efficient way.

Alzheimer’s disease (AD) is a major public health challenge as the median age of the industrialized world’s population is increasing gradually. No cure for this disease has yet been found and the development of new treatments has become a topic of major research interest. This paper aims to propose a sequence of fully automated MRI-based image analysis steps to measure the development stage of atrophy in the brain. The results have been validated on a mixed group of 68 subjects by distinguishing between AD patients, MCIs and health controls using linear classifiers and ANNs. The best classifier identified unseen AD patients correctly in 80% of the cases and control subjects in 85%. Recognizing more than 8 out of 10 MCI subjects, the method also yields an early indication of AD. This simple yet powerful analysis can compete with othermore time-consuming and semi-automaticmethodologies. It could abet an AD diagnosis and provide a tool for measuring the success of therapies.

This paper presents a hybrid approach to perform content-based retrieval on medical image databases. It takes advantage of a pre-processed case base that is batch updated. DICOM information is used to perform pre-filtering to speed up the retrieval process and a image processing knowledge base is used to dynamically reconfigure the most appropriated image processing procedures to perform the image feature extraction. It shows that pre-filtering can speed up considerably the retrieval process and also that some image features produce very similar results what leads to future work on defining the needed digital image processing knowledge base.

The use of modern telecommunications technologies like telemedicine and the internet it’s essential to decrease the centralization of the Brazilian’s public healthcare service, and to improve the access to specialty care in small-tows and rural localities. In this paper, we describe the development and implementation of a public statewide telemedicine network in the Brazilian’s State of Santa Catarina.

We present a visualization and manipulation framework for three-dimensional medical environments. The MVC architecture of the framework allows the use of different API technologies such as OpenGL and Direct3D, different model representations, and controllers. This facilitates the implementation of medical applications offering resources commonly needed by such systems. Two case studies are shown as examples of such applications.

The possession of information about the structural organization of brain white matter tracts by the neurosurgical staff taking care of patients with brain tumors can potentially increase the quality of the interventions. We have developed a software system to execute in vivo white matter fiber tractography using three dimensional tensor fields measured by magnetic resonance diffusion tensor imaging (MR-DTI) that can be used in the normal clinical routine workflow, providing neurosurgeons with very important patient data for operation planning and intra-operative navigation in brain tumors.

This work describes a developed software technology to support the clinic accompaniment of newborns in a neonatal intensive care unit (NICU), where it can help the physicians, nurses and pharmacist directly to the treatment of these babies. With the objective to have the format of a EPR (electronic patient record) and to store information of the patients in an international standard (DICOM SR.), was developed a too! that, effectively, decrease a considerable time in the execution of the tasks of this department and minimizes errors in situations of high risk for the newborns. Having a impact, indirectly, in the improvement of the quality of the attendance of the NICU, giving time to the physicians to the clinical accompaniment. One of the main characteristics of the Cyclops Neonatal Intensive Care Unit is the calculation of parenteral nutrition, this represents one of the first forms of feeding for the interned newborn. This project was implanted in the Maternity Carmela Dutra, of Florianopolis, and was validated for more than 18 months, having a positive impact.

This paper presents a method for robust three-dimensional representation and visualization of radiological image volumes, such as computer tomographies and magnetic resonances, taking advantage of the use of a well-known data-structure: the octree.

The market for health care systems supporting physicians and improving their daily routine is steadily growing. The development of these systems requires handling medical knowledge and process knowledge. One promising knowledge management technique that has been applied in various medical assistant systems is case-based reasoning (CBR). This paper presents selected work from the history of "CBR in medicine", clarifying that several systems already have been developed in the past, but the medical domain also still provides new problems and challenges, where a CBR-approach can be profitably applied. Currently, we are developing a CBR-based component, able to retrieve medical documents in written text form and to convert them into new patterns like a DICOM Structured Report. The component will be integrated in our existing transparent gateway for distributed data access to enable the collaboration of several health care organizations.

The medical procedures at the patient bedside are out of the scope of current systems, which means that verification during the medical visit, the course of treatment, execution and confirmation of the medicine prescription, as well as its administration, still do not enjoy computerized support. However, the inclusion of new data to the patient record, an activity that still needs to be carried out through notations on paper and later typed, causes delays in the availability of this information. This paper presents a software technology in which the medical and nursing staff will be equipped with a handheld computer connected by radio to a central server which will provide access to the patients' data and will actively inform about task pending distribution. The server functions as a database for patient information, patient images from computerized tomography (CT), ultrasonography (US), magnetic resonance (RM) and medical findings (DICOM structured report). This prototype contemplates a server of medical images and findings that makes the research and recovery of these images stored in archives in DICOM standard (Digital Imaging and Communications in Medicine) possible. The use of a web navigator for the visualization of patient data facilitates access to the system, allowing an interface independent platform.

Computer based report systems have proven their utility in improving health care delivery. However the majority of these systems do not comply with any standard for codification, storage and transference of medical data, thus making difficult the interchange of data between different health institutions. This paper describes the development of a user-friendly report editor that produces documents compliant to the DICOM Structured Report-DICOM SR-standard. The strategy chosen to develop the editor was to combine a documentation template editor with a report instance editor that allows the use of previously built templates to ease the process of creating reports. The developed template editor helps the user though building report structures by indicating which information items are needed in order to document a particular type of examination and suggesting values for the information items. The report instance editor represents reports in an easily understandable graphical visualization, and allows the user to create and edit structured reports permitting the use of templates as start-points. It also contains mechanisms to let the user set other document's information required to make it compliant to the DICOM SR standard. The editor is integrated into a DICOM server/client environment and can also be used as a client to other DICOM-SR compliant PACS systems. The report editor developed was already presented to the clinical staff of a community hospital in southern Brazil, where it is intended to be installed, used, and evaluated by the medical staff.

This work considers a methodology to represent medical knowledge about radiological images in order to support medical experts in the interpretation of such images. For this reason they need to express their knowledge in such a way that it can be accessed by the computer. For this purpose a medical language is created that has several levels in order to bridge the gap between computer and human language. Some discussion about validation takes place.

A methodology is presented in order to aid in the surgical planning of abdominal aortic aneurysms. This method addresses image recognition, especially of the aorta and abdominal aneurysms. After the recognition and segmentation of these structures, it is possible to obtain a correct measurement of the same structures, generating more reliable data for manufacturing the required endoprostheses for the treatment of the pathology. With the 3D reconstruction of the anatomical structures, is also possible to make the visualization and analysis of the data more reliable. This work is part of the Cyclops project, which intends to develop intelligent tools for the analysis of medical images, accomplished as a partnership between the Universidade Federal de Santa Catarina (UFSC) and the University of Kaiserslautern, Germany.

The market for health care systems supporting physicians and improving their daily routine is dynamically growing. The development of these systems makes great demands on the handling of medical knowledge, like anatomical and process knowledge. In this paper, a special approach for the radiological domain is presented. Our framework includes mechanisms to store medical knowledge in different knowledge containers, whose importance varies from application to application, and to support the execution of the processes. Three applications are introduced by way of example. We analyze these application scenarios to find the knowledge-intensive tasks that can be supported by an assistant system. The implemented solutions are integrated in the daily work of our radiological partner hospitals.

The image diagnosis area is the biggest medical field in telemedicine, because it does not obligate a direct contact of the patient with the responsible radiologist during the building of the report. The persistent lack of specialists in places distant from urban centers makes telemedicine an important tool for the improvement of healthcare services. In this paper, we present a framework, called CyclopsDistMedDB (Cyclops Distributed Medical Database), for the integration of distributed DICOM medical record databases over wide areas. The system has a central module that is responsible for receiving the clients' requests about patient data (images, waveforms), performing the querying and retrieval of images, patient records, etc. from the specific DICOM databases containing the data requested, and delivering them to the clients. The data communication protocols adopted are DICOM, for retrieval of objects directly from the DICOM data servers, and CORBA (Common Object Request Broker Architecture), for the delivery of DICOM data to client applications.

Ischaemic stroke is a very common disease that affects blood vessels in the brain, causing cerebral tissue damage. Despite several studies on this subject, the early detection of acute ischaemic stroke is difficult, as well its treatment. In this paper, we present some features and initial results of our research, which includes a new image acquisition protocol and a software tool in order to help physicians to improve the accuracy of the therapy.

Presents the IIADPS (Integrated Image Access and Distributed Processing Service) model, a software environment that allows radiological medical teams to have access to distributed image processing facilities in conformance to the DICOM 3.0 (Digital Image Communications in Medicine) standard. The model allows one to remotely execute distributed image processing for diagnosis tasks on any low-cost equipment located on a local network or high-speed wide area network. The model was developed due to the enormous computational processing requirements normally posted to these tasks. Our approach allows low-cost equipment located in a network to be used, independent of the operating system, to execute distributed image processing, leading to a significant reduction in the processing time. This enables also hospitals which do not have radiological stations of high processing capability to use low-cost PC networks for this purpose.

In the German-Brazilian cooperation project Cyclops, we are building an integrated solution in order to capture the exponentially growing medical knowledge. The system is being developed in close cooperation with private radiological hospitals in Germany and Brazil. One aim is easy knowledge exchange between the medical partners in both countries. In this paper, we describe POKMAT (Process-oriented Knowledge MAnagement Tool) and its application to medical examination guidelines. We focus on the support of examination processes and automatic report generation in a DICOM environment. In the future, the system will be extended to an organizational memory of the practice.

Presents a digital anisotropically deformable brain atlas based on the model of the Talairach atlas. The atlas is implemented as a DICOM 3.0-compliant graphic software tool that can be used for teaching and diagnosis support purposes. The tool supports the elastic anisotropic adaptation of the model to the tomographic volume of a particular patient, enabling the identification of the Talairach quadrants and automatically showing the Brodmann areas related to each quadrant. The system is now being evaluated in the scope of a neurological research project.

We present a parallel matching component of an integrated system for the automatic analysis of MRI-breast images towards the early detection of breast cancer. The system operates on images using the method of dynamic contrast-enhanced MRI. Suspicious breast lesions are automatically marked with colours, thus directing the physician's attention towards the critical regions. A proper and careful decision procedure is needed to differentiate between increases of signal intensity triggered by noise and tissue dislocations (motion artifacts) and increases that are triggered by an accumulation of contrast agent in the related breast region. We present our component for image matching using self organising maps (SOM), which enables the system to work properly even with image sequences that are strongly deformed by the patients breathing movements. To reach the time constraint of 15 minutes in medical practice we decide to implement a parallel architecture for the neural network matcher, which works on all computers in the heterogeneous network of our medical partners. The system is tested on real patient data and is now being refined in cooperation with our partner hospital for Radiology and Nuclear Medicine in Mainz.

The segmentation of images with poor contrast characteristics is an important issue in medical computer vision. Often image segmentation results are either oversegmented, with "objects" divided into parts, or incorrectly segmented, with two or more anatomies segmented as one single object. This problem occurs in all types of segmentation approaches, but is of particular importance in the field of region-growing algorithms, which are used in many medical applications, presenting the definition of stable and reliable segmentation parameters. We present a new knowledge-based method, based on an extension of the inexact consistent labelling method, that enables the automated consistency checking of the results of region-growing segmentations and is capable to automatically "fitting" erroneous segmentations, when they are oversegmented, when there exists a reliable domain model that can be used to guide a tree search procedure in the space. This allows the use of oversensitive parameters when an exact segmentation is not reliable.

We describe a computational method to assist radiologists in performing better more reliable and simpler diagnosis of neurocysticercosis (NC). Based on this method we implemented a software system that counts and measures the calcifications related to NC in computed tomography (CT) scans, thus reducing errors regarding visual inspection and providing better quantitative data. During computation, the system segments grey scale images obtained by CT scans and resulting segments are submitted to classification using artificial neural networks (ANNs). The system marks NC findings, replacing automatically all areas in the original image classified as NC with specially coloured markings. Afterwards, the system starts correlating NC-findings in different slices and performing a 3D reconstruction based on NC-classified areas belonging to the same finding. As a final step, the system performs a 3D reconstruction of the patient's skull, encephalic mass and findings boundaries, generating a 3D representation of the patient's head and the localisation of NC findings. In this step the volumes of each NC finding are also calculated.

Most of the digital medical devices used in modern clinics and practice provide data corresponding to the DICOM Standard. Cyclops, a knowledge-based image analysis system developed at the University of Kaiserslautern, was suitable to analyze such images, but we needed a DICOM interface to connect to DICOM databases easily and transfer data to the system. All non-commercial software available were not up to our demands, so we implemented a DICOM network client by our own, which can connect to the ctn image server, based on an SQL database. We attached great value to a neat, object-oriented implementation, because it is easy to expand and we are flexible to future versions of the DICOR/I standard.

We describe a method for the automatic registration of transversal MRI-head-volumes. Under automatic registration we understand the determination of an adequate coordinate system for a specific head volume, which is adapted to this particular head, describing the position, direction and size of the head in the respective MRI-volume. After determining the coordinate system the specific head volumes are mapped against a model of an anatomical atlas.