O programa tem como área de concentração a Ciência da Computação, e será desenvolvido segundo 2 linhas de pesquisa: Sistemas de Computação e Computação Aplicada. https://tedebc.ufma.br/jspui/handle/tede/3739 2026-04-16T01:48:51Z 2026-04-16T01:48:51Z ARAÚJO, Alexandre de Carvalho https://tedebc.ufma.br/jspui/handle/tede/6873 2026-03-26T18:22:07Z 2025-09-29T00:00:00Z

Título: Segmentação automática do pâncreas e massas pancreáticas em tomografias computadorizadas usando arquiteturas encoder-decoder Autor: ARAÚJO, Alexandre de Carvalho Primeiro orientador: ALMEIDA, João Dallyson Sousa de Abstract: Cancer, the second leading cause of mortality worldwide, accounted for approximately 10 million deaths in 2020. Although relatively rare compared to other malignancies, pancreatic cancer presents one of the poorest prognoses, with a mortality rate of approximately 98%, ranking among the highest across all cancer types. Early-stage detection is the most critical determinant of patient prognosis. However, early diagnosis is particularly challenging due to the difficulty in identifying small lesions in medical imaging modalities such as abdominal ultrasound, computed tomography (CT), and magnetic resonance imaging (MRI). These imaging techniques constitute the primary tools for early detection, and the accurate identification of small pancreatic lesions at early stages significantly improves clinical outcomes. Therefore, there is a pressing need for technologies that can augment and enhance image-based diagnostic workflows. In this work, two deep learning-based methods were developed: one for automatic pancreas segmentation and another for pancreatic mass segmentation in CT scans. The proposed pancreas segmentation approach integrates the EfficientNetB7 backbone within a U-Net architecture, achieving promising results, with a mean Dice similarity coefficient (DSC) of 85.39 ± 2.39% on the NIH dataset and 85.96 ± 2.08% on the MSD dataset. For pancreatic mass segmentation, an ensemble of five encoder-decoder architectures was employed: U-Net, FPN, LinkNet, EDU-Net, and ETDPU-Net. The first three utilize EfficientNetB7 as their encoder backbone. The latter two—EDU-Net and ETDPU-Net—were designed as part of this thesis and incorporate deformable convolutions and Squeeze-and-Excitation (SE) attention mechanisms to enhance feature representation. The outputs of all five models are aggregated using a majority voting ensemble strategy to surpass the performance of individual models. The proposed mass segmentation framework achieved a DSC of 65.28% ± 5.57% on the MSD dataset, validating the efficacy and robustness of the method. Instituição: Universidade Federal do Maranhão Tipo do documento: Tese; Trabalho sob Sigilo. Motivo: Resultados apresentados na Tese ainda não foram publicados. Data Provável de Liberação: 2 anos.

2025-09-29T00:00:00Z CRUZ, Allan Kássio Beckman Soares da https://tedebc.ufma.br/jspui/handle/tede/6664 2025-12-10T11:48:54Z 2025-10-15T00:00:00Z

Título: Geração de documentos de design de jogos educacionais a partir de templates orientados ao Pensamento Computacional Autor: CRUZ, Allan Kássio Beckman Soares da Primeiro orientador: SOARES NETO, Carlos de Salles Abstract: This thesis investigates how to support the authoring of educational games oriented to Computational Thinking by tightly linking pedagogical objectives, game mechanics, and assessability. It proposes the GDD2PC model, which makes explicit the correlation between Computational Thinking pillars (abstraction, algorithmic thinking, decomposition, and pattern recognition) and families of mechanics, and derives an operational Game Design Document template tailored to teaching practice, including a concise onepage format. Both were instantiated in the Hefesto Game Lab tool, designed to reduce authoring effort, standardize document quality, and preserve flexibility across educational contexts. The research followed a Design Science Research process: requirements elicitation, model and template construction, tool implementation, and user-oriented evaluation. Technology acceptance was assessed with the TAM, complemented by pedagogical alignment and applicability. Results show mean perceived usefulness around 4.0, perceived ease of use near 4.1, and behavioral intention at 3.8 on a 1–5 Likert scale, with high instrument reliability (high Cronbach’s alpha). Evidence indicates that the approach reduces authoring ambiguity, speeds up iteration cycles, and strengthens alignment between Computational Thinking pillars and design decisions, contributing to a more efficient and replicable teaching practice. Instituição: Universidade Federal do Maranhão Tipo do documento: Tese

2025-10-15T00:00:00Z DINIZ, Joel de Conceição Nogueira https://tedebc.ufma.br/jspui/handle/tede/6651 2025-12-11T17:24:09Z 2025-01-01T00:00:00Z

Título: Ensemble de Redes Convolucionais para Inspeção Visual de Estruturas de Concreto Armado Autor: DINIZ, Joel de Conceição Nogueira Primeiro orientador: PAIVA, Anselmo Cardoso de Abstract: Pathologies in concrete structures can be visually detected on their surface, such as cracks or fissures, fragmentation of concrete, efflorescence, corrosion stains, and exposed steel bars (the latter two occurring in reinforced concrete). Therefore, these pathologies can be analyzed via images of reinforced concrete structures. This thesis proposes an ensemble of convolutional networks for visual inspection of reinforced concrete structures. This method speeds up the detection task and increases its effectiveness by saving time on the identification to be analyzed and by eliminating or reducing errors, such as those arising from human error during the massive execution of tedious tasks in the analysis. The task of identifying pathology can be performed using a convolutional neural network if the images are cropped to the specific pathology to be identified, or using a detection network if the images are broad and the pathology is inserted into a context with several classes of pathology, including areas without pathology. Another task that enables the identification and analysis of these pathologies is segmentation. The method was tested with detection and classification tasks. The neural network architectures used for detection were YOLO v11 and TOOD (Taskaligned One-stage Object Detection) for the single-stage approach, and Faster RCNN for the two-stage approach. The three networks were then combined for the Ensemble application, using Weighted Box Fusion. For the classification task, the network architectures used were DenseNet121, ResNet50, and MobileNetV3. Research was conducted to locate and select appropriate datasets for the proposed method. The dataset selected for classification is Ozgnel, and the one for detection is CODEBRIM. The detection task allows an artifact to be located and classified. Although this approach is efficient, the research tested using detection to locate pathologies without initially defining which class they belong to, and then using a specific classification neural network to define the types of pathologies and, above all, to eliminate false positives. The approach of combining a detection network and a dedicated classification network achieved the desired result, with Ensemble used to increase sensitivity in the detection phase, thereby increasing the number of artifacts located. This approach allows false positives to pass in the first phase, but they are eliminated in the last phase. Instituição: Universidade Federal do Maranhão Tipo do documento: Tese

2025-01-01T00:00:00Z FERREIRA, Marcos Melo https://tedebc.ufma.br/jspui/handle/tede/6579 2025-11-04T12:48:07Z 2025-09-19T00:00:00Z

Título: Classificação do estágio de glaucoma usando dados multimodais Autor: FERREIRA, Marcos Melo Primeiro orientador: BRAZ JUNIOR, Geraldo Abstract: Glaucoma is the leading cause of irreversible blindness worldwide. Its early diagnosis is challenging due to the absence of symptoms in the initial stages, the need for multiple exams to be analysed by specialised professionals, and the general lack of awareness about the disease among the population. Although the visual loss caused by glaucoma is irreversible, its progression can be slowed if the disease is detected in its early stages. In this context, deep learning methods have demonstrated promising results in medical image processing tasks, including classification and segmentation, offering potential support for clinical diagnosis. In this work, we developed a method for glaucoma stage classification that combines fundus photographs and OCT volumes. The method employs a multimodal convolutional architecture and explores various fusion strategies, both at the feature map and prediction levels, aiming to integrate multimodal information effectively. Additionally, specific regions of interest were investigated — the optic nerve in fundus photographs and the retinal layers in OCT volumes — to improve data representation and enhance classification accuracy. The experiments demonstrated that multimodal models outperformed unimodal approaches, achieving a Kappa score of 0.88, which indicates a high level of agreement of the proposed method with specialist assessments. Moreover, the results showed that fundus photography has a greater influence than OCT volumes in the classification process. At the same time, the targeted capture of retinal layers proved to be a promising strategy for further improving accuracy. Overall, the proposed method demonstrated significant potential as a clinical decision support tool, contributing to the advancement of automated diagnostic systems and enabling earlier and more accurate glaucoma detection. Instituição: Universidade Federal do Maranhão Tipo do documento: Tese

2025-09-19T00:00:00Z