polito.it
Politecnico di Torino (logo)

PACKET CONTENT PREDICTION IN A TELEPATHOLOGY SESSION

Nicy Malanda-Sendo

PACKET CONTENT PREDICTION IN A TELEPATHOLOGY SESSION.

Rel. Guido Marchetto, Alessio Sacco. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Informatica (Computer Engineering), 2020

[img]
Preview
PDF (Tesi_di_laurea) - Tesi
Licenza: Creative Commons Attribution Non-commercial No Derivatives.

Download (625kB) | Preview
Abstract:

Telepathology refers to practicing pathology from a distance. Telecommunications technology is used for facilitating the transmission of pathology image-rich data between two distant locations for diagnosis, research and education purposes. In order to perform telepathology, a pathologist must choose the video images that need to be analyzed and then render a diagnosis. The use of television microscopy, which preceded telepathology, didn’t require a pathologist to have a virtual or physical hands-on involvement in choosing the microscopic fields-of-view to analyze and diagnose. Today, With the wide application of prediction, especially in the telemedicine field, the research of prediction algorithm and theory has made a great progress. The goal of this thesis is to propose an approach that uses a machine learning (ML) method, Hidden Markov Model (HMM), to predict the packet content from the generated network traffic. Also, the thesis work is focusesd on the comparison HMM and others ML tecniques such as Support Machine Vectors (SVM), Autoregressive Integrated Moving Average (ARIMA), Decision-Tree (DT), Naive-Bayes (NB) and K-Nearest Neighbors (KNN). Additionally, this approach makes use of Micro-Manager, which is a new edge computing-based telepathology system that enables live histological image processing and real-time remote control of the microscope. During the telepathology session, two different experimental datasets were manually captured through Micro-Manager for training and testing the models. In particular, HMM achieves 87.28% and 82.84% of accuracy, for the first and second datasets respectively. We obtained 62.48% and 58.44% with SVM. ARIMA has gotten 43.72% and 8.37%. Finally, DT, NB and KNN reach each 68.76% and 66.52% of accuracy. These findings indicate that the proposed method, HMM, is the best algorithm analyzed for this study.

Relatori: Guido Marchetto, Alessio Sacco
Anno accademico: 2020/21
Tipo di pubblicazione: Elettronica
Numero di pagine: 34
Soggetti:
Corso di laurea: Corso di laurea magistrale in Ingegneria Informatica (Computer Engineering)
Classe di laurea: Nuovo ordinamento > Laurea magistrale > LM-32 - INGEGNERIA INFORMATICA
Aziende collaboratrici: Politecnico di Torino
URI: http://webthesis.biblio.polito.it/id/eprint/16727
Modifica (riservato agli operatori) Modifica (riservato agli operatori)