Ciencias Exactas y Ciencias de la Salud
Permanent URI for this collectionhttps://hdl.handle.net/11285/551039
Pertenecen a esta colección Tesis y Trabajos de grado de las Maestrías correspondientes a las Escuelas de Ingeniería y Ciencias así como a Medicina y Ciencias de la Salud.
Browse
Search Results
- Validating immersive technologies for enhanced decision-making data in surgical robotics via agent-based learning(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2025-06) Nieto Gutiérrez, Nezıh; Muñoz Ubando, Luis Alberto; emimmayorquin; Cerón López, Arturo Eduardo; Ochoa Ruiz, Gilberto; Durán Sierra, Elvis de Jesús; School of Engineering and Sciences; Campus MonterreyThis thesis addresses the critical challenge of acquiring high-quality decision-making data for surgical robotics by uniting immersive technologies with advanced machine-learning frameworks. Existing data-collection methods often fail to capture the rich multimodal signals necessary for training competent robotic agents. To overcome these limitations, a synchronized pipeline was developed within the DaVinci Research Kit (dVRK) simulator that combines stereoscopic virtual-reality (VR) visualization, real-time haptic feedback, electroencephalography (EEG)–based cognitive monitoring, motion tracking, and task performance metrics. A novel Reinforcement-and-Imitation Diffusion Learning (RIDL) approach was introduced, employing a stable-diffusion pre-training stage for a diffusion-policy network followed by Proximal Policy Optimization (PPO) fine-tuning with expert demonstrations. Contributions include (1) a reproducible multimodal data-acquisition pipeline for surgical robotics, (2) an open-source implementation of the VR–haptic–EEG–RIDL framework, and (3) a comprehensive evaluation protocol integrating behavioral, neurophysiological, and subjective metrics. Future work will validate the system on physical dVRK hardware, integrate additional biosignals (e.g., heart-rate variability, biomechanics via body tracking), and explore metareinforcement learning for rapid policy adaptation. By elevating the quality of decisionmaking datasets and enabling neuroadaptive training, this research aims to accelerate the deployment of capable, human-centered surgical robots in clinical practice. For readers and practitioners drawn to the intersection of immersive interface design, cognitive neuroscience, and robotic autonomy, this work offers both a urnkey framework and a rigorous evaluation methodology. Expect detailed guidance on implementing synchronized VR/haptic/EEG data streams, reproducible code for iffusion-policy pretraining, and a suite of metrics that link neurophysiological engagement to task performance. Whether the goal is to replicate these experiments, extend the pipeline to new robotic platforms, or inform the design of next-generation neuroadaptive systems, this thesis provides the conceptual foundations, open-source tools, and empirical benchmarks necessary to advance the state of surgical robotics and human-centered automation.
- Evaluating Pre-trained Neural Networks in Deep Learning for Early Detection and Enhanced Screening of Cervical Pathology(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2024-12) González Ortiz, Orlando; Muñoz Ubando, Luis Alberto; emimmayorquin; Raymundo Avilés, Arturo; Cerón López Universidad, Arturo Eduardo; School of Engineering and Sciences; Campus Monterrey; Ochoa Ruiz, GilbertoThis document presents a research thesis for the Master in Computer Science (MCC) degree at Tecnologico de Monterrey. Cervical cancer remains a leading cause of mortality among women, particularly in low-resource regions where screening tools such as the Pap smear often fall short in early detection. This research explores the application of deep learning and pre-trained neural networks for the binary classification of cervical pathology, focusing on detecting dysplasia, specifically CIN2 and CIN3, as a potential prevention tool. We im- plemented multiple neural network models, including DenseNet, EfficientNet, MobileNet, and ResNet. The models were evaluated on two distinct datasets: one from the International Agency for Research on Cancer (IARC) and another from the Zambrano Hospital. To as- sess the generalization capacity of these models, we employed a sequential training approach where the first batch was trained with IARC data and tested on a Zambrano Hospital batch, with subsequent tests progressively incorporating prior results. Each experiment was repeated over 10 iterations to calculate confidence intervals for the performance metrics. Our results demonstrate that DenseNet and EfficientNet outperformed other models, achieving superior sensitivity and accuracy compared to conventional Pap smear tests. These findings indicate that deep learning models hold promise as an affordable, effective cervical cancer screening tool in low-resource communities. Future work will focus on augmenting datasets through collaboration with healthcare institutions and exploring generative models such as GANs to improve model robustness and generalization.
- 3-D Detection & tracking for semi-deformable objects(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2024) De Los Rios Alatorre, Gustavo; Muñoz Ubando, Luis Alberto; emimmayorquin; Hernández Gress, Neil; Ceballos Cancino, Héctor Gibrán; Raygosa Barahona, Rubén Renan; Maestro en Ciencias de la Computación; Campus MonterreyThis thesis introduces a computer vision system designed for real-time detection and pose estimation of semi-deformable objects in 3-D space, leveraging edge computing devices. The primary motivation for this research stems from the need to enhance the capabilities of vision-based systems, which in turn can aid in improving the efficiency and effectiveness of robotics systems in a variety of fields. For the context of the thesis the chosen field was agriculture, focusing on the recognition, tracking and pose estimation of bell peppers by harvesting robots, an application where traditional methods often fall short due to the nature of semi-deformable objects like fruits. A Jetson Nano was used as the main component, while an Intel DE10-Nano was considered as a complementary part of the system for performing image preprocessing tasks with the Azure Kinect being considered as the main camera sensor. The algorithm was successfully deployed in the Jetson Nano, successfully tracking and estimating the pose of a bell pepper in 3-D by performing the necessary rotations and deformations to a canonical model used by the system as a general means to estimate the pose of the pepper in the real world scene. The algorithm was also tested in a ROS 2 Gazebo simulation where an x-arm robot was used to simulate the vision part of a pick and place operation with a simulated bell pepper, using the proposed method to accurately identify and estimate the pose of the pepper in the simulation. Lastly, a set of different segmentation techniques using both deep learning and traditional methods are presented as a means to explore how these could better the current segmentation capacity of the system.
- Biomechanical and machine learning integration for the detection of knee injuries: exploring the utility of non-intrusive elements(Instituto Tecnológico y de Estudios Superiores de Monterrey, 2022-08-15) Rosales Gurmendi, Diana Sofía Milagros; Muñoz Ubando, Luis Alberto; emimmayorquin; School of Engineering and Sciences; Campus MonterreyJoints play an essential role in the human body, allowing for movement, stability, and overall functionality. The knee, on the other hand, is a complex joint that contributes to the structural support of the body and enables a wide range of movements. Consequently, it is more susceptible to injury, which negatively impacts a person's daily routines. While there are important markers aiding in the assessment and diagnosis of joint health, they are typically evaluated using traditional qualitative methods. In this study, we focus on utilizing computational Machine Learning (ML) tools for recognizing patterns in injured joints. Specifically, three potential biomarkers were evaluated: joint sound, range of motion, and qualitative pain assessment. The purpose of this work is to explore the contribution of these three biomarkers in identifying knee injuries. A total of 22 participants, including both male and female individuals aged 23 and above, with and without a history of joint injury, were recorded and compared. A handheld device designed to measure the angular displacement and crackling sound of the joint was utilized. The tests consisted of 5 flexion and extension cycles, each comprising 4 repetitions. Signals were pre-processed, and 13 features were extracted for each frame. Using a binary classifier, dimensionality reduction methods, and signal processing by segmentation, the results showed an accuracy of 82\% in classifying the data with a precision of 83\%. The Machine Learning model successfully identified distinctive patterns between healthy and injured knees. Furthermore, the significant importance of considering the knee joint sound, presence of pain, and range of motion in clinical diagnostic evaluations to determine the presence or absence of a joint injury was highlighted.