Summary of the Project
In recent years, the growing early diagnostic Methods of prostate cancer such as a prostate-specific antigen, biopsy, and subsequently the rate of cancer diagnosis has increased. However, the risk of prostate cancer in a man’s death from prostate cancer is 16% and 2.9%, respectively, which most are benign. The statistics provided show the need to reduce unnecessary biopsy cases. At present, the pathology results from the tissue obtained by biopsy as a standard of malignancy that this method can lead to hematuria, hematospermia and, rectal bleeding. The use of prostate- specific antigen is another method of assessing tumor malignancy, although it is non-invasive but has low accuracy and specificity. In recent years, DCE-MRI imaging has established itself as a non-invasive method for diagnosing and classifying prostate cancer. Currently, evaluation of prostate tumors using MRI involves one or more cases of anatomical and functional imaging such as T2 weighted imaging, Diffusion weighted imaging (DWI), and DCE-MRI. It should be noted that none of the MRI sequences alone can fully characterize prostate cancer, and each functional imaging has its advantages and disadvantages and requires a combination of functional and anatomical sequences to better describe the tissue head.
Nevertheless, DCE-MRI plays an important role in the analysis of prostate cancer, and DCE-MRI is a useful clinical tool for assessing the grading and localization of tumors for diagnostic purposes and finally to monitoring of tumor recurrence pattern. DCE-MRI image analysis is performed in three ways:
qualitative, quantitative and semi-quantitative. The use of qualitative results has high speed and low accuracy and the result obtained depends on the experience of the radiologist. By quantitative analysis of DCE-MRI images, the parameters of the pharmacokinetic model of the tissue can be obtained, which are directly related to the physiological characteristics of the tissue. In this method, by fitting the concentration curves of the contrast agent to the tissue model, the model parameters such as the coefficient of volume transferred between blood plasma and extravascular space (K trans ), which is used to measure vascular permeability, the relative volume of extracellular and extracellular area. Vascular (ve), the flow constant from the extracellular and extravascular space to the plasma (K ep ) and. Are obtained, which can indicate the state of tumor progression. Semi-quantitative analysis of images is performed on image intensity curves that can measure quantitative parameters of the appearance of image intensity curves such as time to peak curve, input flow slope, output flow slope, area under the curve and , parameters obtained from Semi-quantitative analysis is indirectly related to the physiological characteristics of the studied tissue.
Our goal in this project is to build a model based on machine learning using the features that derived from quantitative and semi-quantitative analysis to predict tumor malignancy without the use of invasive procedures such as biopsy for clinical using. The results obtained from the automatic model designed in this study are compared with the pathology results obtained as gold standard.