Modelo de clasificación automática de imágenes de resonancia magnética para el diagnóstico del cáncer de próstata.
Metadata
Show full item recordDocuments PDF
Abstract
In the current era the image is to medicine as the eye is to the human being, this due to the enormous importance of image processing and analysis, in many fields such as medicine, and it is in medical science that a high degree of content of applications for the great challenges of science in the 21st century, where the use of efficient tools is essential.
This research is focused on the development of an Automatic Learning System, Machine Learning (ML), using Magnetic Resonance Imaging (RM) of human beings and analyzing the structures involved, by extracting geometric characteristics such as the measurement of bearing of the lesion for the three images (T2, DWI and ADC), eccentricity, area, arc length, extreme points, centroid, convexity, radius, major axis length, minor axis length, orientation, solidity, extension, equivalent diameter , moments and moments of Hu of the image and fractal dimension by count of boxes, together with the data of the coordinates of the position of the lesion, the matrix that describes the orientation and scale of the image, the values of i, j and k being the column, row and coordinate of the cut to be found respectively, the vector with scalars of spacing x, y, z, and the position of the lesion TZ: Transitional zone, PZ: Peripheral Zone and AS: Fibromusculous Stroma ar, taking into account the PI-RADS evaluation categories for the probability of having clinically significant cancer, where the PI-RADS evaluation characteristics show a high possibility of presenting a significant cancer in the prostate.
The work is aimed at the support of professionals in oncology, focused on the diagnosis of prostate cancer. This system aims to be a timely help, in cancer centers, hospital centers, cancer expert oncologists and for all those interested in improving the diagnosis of prostate cancer.
The prostate in size is like a walnut and is located in front of the rectum and below the bladder. It is located only in men. Prostate cancer, also known as prostate carcinoma, is located as the most frequent malignant tumor in men and the second most common cause of cancer-related death in men, due to which the importance of studying this type is seen cancer, in order to achieve significant progress in terms of diagnosis.
The diagnosis of prostate cancer is carried out by an expert professional, who, with the help of tools such as magnetic resonance imaging (MRI), which provides excellent detail of the anatomy and other techniques and procedures, issues the diagnosis of prostate cancer. In this research project, a novel machine learning based tool is implemented, with a classification method in magnetic resonance imaging to support the diagnosis of prostate cancer fast and timely.
Abstract
In the current era the image is to medicine as the eye is to the human being, this due to the enormous importance of image processing and analysis, in many fields such as medicine, and it is in medical science that a high degree of content of applications for the great challenges of science in the 21st century, where the use of efficient tools is essential.
This research is focused on the development of an Automatic Learning System, Machine Learning (ML), using Magnetic Resonance Imaging (RM) of human beings and analyzing the structures involved, by extracting geometric characteristics such as the measurement of bearing of the lesion for the three images (T2, DWI and ADC), eccentricity, area, arc length, extreme points, centroid, convexity, radius, major axis length, minor axis length, orientation, solidity, extension, equivalent diameter , moments and moments of Hu of the image and fractal dimension by count of boxes, together with the data of the coordinates of the position of the lesion, the matrix that describes the orientation and scale of the image, the values of i, j and k being the column, row and coordinate of the cut to be found respectively, the vector with scalars of spacing x, y, z, and the position of the lesion TZ: Transitional zone, PZ: Peripheral Zone and AS: Fibromusculous Stroma ar, taking into account the PI-RADS evaluation categories for the probability of having clinically significant cancer, where the PI-RADS evaluation characteristics show a high possibility of presenting a significant cancer in the prostate.
The work is aimed at the support of professionals in oncology, focused on the diagnosis of prostate cancer. This system aims to be a timely help, in cancer centers, hospital centers, cancer expert oncologists and for all those interested in improving the diagnosis of prostate cancer.
The prostate in size is like a walnut and is located in front of the rectum and below the bladder. It is located only in men. Prostate cancer, also known as prostate carcinoma, is located as the most frequent malignant tumor in men and the second most common cause of cancer-related death in men, due to which the importance of studying this type is seen cancer, in order to achieve significant progress in terms of diagnosis.
The diagnosis of prostate cancer is carried out by an expert professional, who, with the help of tools such as magnetic resonance imaging (MRI), which provides excellent detail of the anatomy and other techniques and procedures, issues the diagnosis of prostate cancer. In this research project, a novel machine learning based tool is implemented, with a classification method in magnetic resonance imaging to support the diagnosis of prostate cancer fast and timely.