DSpace Communidade:http://www.repositorio.ufop.br/jspui/handle/123456789/5962024-03-28T18:40:13Z2024-03-28T18:40:13ZContributions to automating the analysis of conventional Pap smears.Diniz, Débora Nasserhttp://www.repositorio.ufop.br/jspui/handle/123456789/180712024-02-07T21:07:30Z2023-01-01T00:00:00ZTítulo: Contributions to automating the analysis of conventional Pap smears.
Autor(es): Diniz, Débora Nasser
Resumo: This thesis, organized as a compilation of articles, develops and presents contri-
butions to the automated analysis of conventional Pap smear slides. A conventional
Pap smear slide is a sample of cervical cells collected and prepared on a glass slide for
subsequent cytopathological analysis. The main contributions are to detect and classify
cervical cell nuclei to develop a decision support tool for cytopathologists. The first arti-
cle resulting from this research utilizes a hierarchical methodology using Random Forest
for the nucleus classification of the Herlev and Center for Recognition and Inspection
of Cells (CRIC) Searchable Image Database databases based on 232 handcrafted fea-
tures. In this article, we investigate balancing techniques, perform statistical analyses
using Shapiro-Wilk and Kruskal-Wallis tests, and introduce the CRIC Searchable Image
Database segmentation base. Our result defined the state-of-the-art in five metrics for
nucleus classification in five and seven classes and the state-of-the-art in precision and
F1-score for two-class classification. The second article introduces a method for nu-
cleus detection in synthetic Pap smear images from the Overlapping Cervical Cytology
Image Segmentation Challenge dataset proposed at the 11th International Symposium
on Biomedical Imaging (ISBI’14). In this second article, we investigate clustering al-
gorithms for image segmentation. We also explore four traditional machine learning
techniques (Decision Tree (DT), Nearest Centroid (NC), k-Nearest Neighbors (k-NN),
and Multi-layer Perceptron (MLP)) for classification and propose an ensemble method
using DT, NC, and k-NN. Our result defined the state-of-the-art recall using this dataset.
The third article proposes an ensemble method using EfficientNets B1, B2, and B6 to
classify images from the CRIC Searchable Image Database dataset. Here, we investigate
ten neural network architectures to choose those used in the ensemble method and present
a data augmentation methodology using image transformation techniques. Our result de-
fined the five state-of-the-art metrics for nucleus classification in two and three classes.
Furthermore, we introduce results for six-class classification. Lastly, the fourth article introduces the Cytopathologist Eye Assistant (CEA), an intuitive and user-friendly tool
that uses deep learning to detect and classify cervical cells in Pap smear images, support-
ing cytopathologists in providing diagnoses. We investigate You Only Look Once (YOLO)
v5 and YOLOR for performing both tasks (detection and classification) and also explore
the combination of using YOLOv5 for detection and the ensemble of EfficientNets from
the third article for classification. The article explores data balancing techniques, under-
sampling, and oversampling using Python’s Clodsa library. The CRIC Cervix database
was used for tool evaluation, considering four scenarios: original images, resized im-
ages, augmented resized images, and balanced resized images. The application of CEA
was validated by specialists with years of experience in cytopathology, highlighting the
tool’s ease of use and potential to address specific queries.
Descrição: Programa de Pós-Graduação em Ciência da Computação. Departamento de Ciência da Computação, Instituto de Ciências Exatas e Biológicas, Universidade Federal de Ouro Preto.2023-01-01T00:00:00ZWearable edge AI towards cyber-physical applications.Silva, Mateus Coelhohttp://www.repositorio.ufop.br/jspui/handle/123456789/180102024-01-19T21:34:47Z2023-01-01T00:00:00ZTítulo: Wearable edge AI towards cyber-physical applications.
Autor(es): Silva, Mateus Coelho
Resumo: The creation of novel technologies to support field work and research has a major
impact from technologies such as the Internet of Things (IoT), Edge Computing and
wearable computing. In this context, Artificial-Intelligence-based systems became
more common and a trend in recent work. Environments with low connectivity and
high latency in data transmission enforce the usage of Edge Computing technologies
in the treatment of acquired data. Nonetheless, there is no clarity in how to transport
Artificial Intelligence (AI) to Edge Computing in extreme environments, given the
complexity of the requirements. This gap is more clear in the context of wearable
computing, where the systems restrictions for developing systems are even harder.
Thus, this work presents a protocol for developing Edge AI appliances and some
case-study applications in the context of wearable devices. This study helps to
evaluate the creation of Wearable Edge AI context as a novel research field.
Descrição: Programa de Pós-Graduação em Ciência da Computação. Departamento de Ciência da Computação, Instituto de Ciências Exatas e Biológicas, Universidade Federal de Ouro Preto.2023-01-01T00:00:00ZSelf-supervised learning for arrhythmia classification.Silva, Guilherme Augusto Lopeshttp://www.repositorio.ufop.br/jspui/handle/123456789/177402023-11-13T20:37:22Z2023-01-01T00:00:00ZTítulo: Self-supervised learning for arrhythmia classification.
Autor(es): Silva, Guilherme Augusto Lopes
Resumo: Arrhythmias, heart diseases that are commonly diagnosed through electrocar-
diograms (ECG), require computational methods for detection and classification
to improve the physician’s diagnosis. Although there is abundant literature on the
subject, the high intra-patient variability and noise of ECG signals pose challenges
in developing practical machine-learning models. To address this, we propose a cus-
tomized adjustment of machine learning models through self-supervised learning with
human-in-the-loop. Our approach introduces a pretext task called ECGWavePuzzle,
which improves classification performance through better generalization. Evaluation
metrics on the MIT-BIH database demonstrate the effectiveness of our approach,
which improved the ECGnet global accuracy by over 10% and the Mousavi’s CNN
by over 13%. Additionally, the experimental results demonstrated that the proposed
approach improved the sensitivity and positive predictive value of the arrhythmic
classes for certain patients.
Descrição: Programa de Pós-Graduação em Ciência da Computação. Departamento de Ciência da Computação, Instituto de Ciências Exatas e Biológicas, Universidade Federal de Ouro Preto.2023-01-01T00:00:00ZComputational techniques to find and suppress bone from chest radiological images.Ziviani, Hugo Eduardohttp://www.repositorio.ufop.br/jspui/handle/123456789/177302023-11-13T19:07:32Z2023-01-01T00:00:00ZTítulo: Computational techniques to find and suppress bone from chest radiological images.
Autor(es): Ziviani, Hugo Eduardo
Resumo: The proposal of this work is to propose bone suppression techniques in chest images. The most
common, but inaccessible, way is through Dual Energy Subtraction (DES). This the technique
requires specific hardware to generate and receive di
erent energy levels capable of di
erentiating
materials by atomic number. This work uses GAN to perform bone suppression on X-ray images
and aimed to evaluate the performance of the cGAN, train a model to locate the thoracic box, and
assess two di
erent training techniques for boneless image translation. Based on deep learning
the main contribution of this work is to improve the bone shadow elimination delimiting the
learning region of the Deep Learning (DL) model. By the contextualization of the bones region,
was possible present a metric that measures the model accuracy in an interested region. With
this study was possible a more precise metric to evaluate the bone suppression quality. Using
the Japanese Society of Radiological Technology (JSRT) this study achieved a PSNR index of
31.604, and a similarity coe
cient, known as SSIM of 0.9402. When delimiting the learning
region, the results were: 31.9136 for PSNR and 0.9633 for SSIM.
Descrição: Programa de Pós-Graduação em Ciência da Computação. Departamento de Ciência da Computação, Instituto de Ciências Exatas e Biológicas, Universidade Federal de Ouro Preto.2023-01-01T00:00:00Z