MRI estimation of contrast agent concentration using a neural network approach
The invention comprises systems, methods, and apparatus to correlate changes in MRI data with CA concentration, using an adaptive neural network In MRI techniques, CAs are used to estimate vascular properties such as blood flow, blood volume, and transfer constant of tissue microvessels However, the relationship between the contrast in the MRI image and the contrast agent concentration is not linear, instead depending on factors such as the nature of the sequence, the nature of the tissue, and the tissue concentration of contrast agent, and thus limiting the reliability of vascular properties using MRI.
1. A method comprising:
providing an adaptive neural network with training criterion image data of known radioactive contrast agent concentration in biological tissue,
providing the adaptive neural network with MRI image data from a patient before injection of a radioactive MRI contrast agent into the patient,
providing the adaptive neural network with MRI image data from the patient after injection of a radioactive MRI contrast agent into the patient, the adaptive neural network measuring any change between the MRI image data from a patient before injection of a radioactive MRI contrast agent into the patient and the MRI image data from the patient after injection of a radioactive MRI contrast agent into the patient,
and
comparing the training criterion data of known radioactive agent concentration with any change in MRI image data from the patient after injection of the radioactive contrast agent and measuring the concentration of radioactive contrast agent in the post-injection MRI image as a continuous variable, each by using the adaptive neural network;
wherein the MRI image data from the patient before injection and the MRI image data from the patient after injection are each prepared from MRI tissue magnetization according to at least one of the following steps:
a. inverting the magnetization of the tissue and measuring any recovery of the tissue magnetization with further MRI imaging, or
b. perturbing the magnetization of the tissue with a known tip-angle, and measuring any decrease of tissue magnetization with further MRI imaging, or
c. perturbing the magnetization of the tissue with a set of known tip-angles and measuring the changes in tissue magnetization with further MRI imaging;
wherein the radioactive MRI contrast agent injected into the patient comprises a gadolinium labeled agent; and
wherein the training criterion image data are derived from autoradiographic images of tissue radioactive contrast agent administration to a mammal.
2. The method of claim 1 , wherein the radioactive MRI contrast agent comprises gadolinium labeled serum albumin.
3. The method of claim 2 , wherein the radioactive contrast agent used to generate the training criterion image data of known radioactive contrast agent concentration comprises radioactive serum albumin.
4. A method comprising:
providing an adaptive neural network with training criterion image data of known radioactive contrast agent concentration and matching data: (a) of tissue relaxation rate R1 before and after administration of MRI contrast agent in an experimental animal, or (b) of tissue relaxation rate R1 before and after administration of MRI contrast agent in a human,
providing the adaptive neural network with MRI image data from a patient or experimental animal before injection of a radioactive MRI contrast agent into the patient or the experimental animal,
providing the adaptive neural network with MRI image data from the patient or experimental animal after injection of a radioactive MRI contrast agent into the patient or the experimental animal, the adaptive neural network measuring any change between the MRI image data from a patient before injection of a radioactive MRI contrast agent into the patient and the MRI image data from the patient after injection of a radioactive MRI contrast agent into the patient or experiment animal,
and
comparing the training criterion data of known radioactive agent concentration with any change in MRI image data from the patient or experimental animal after injection of the radioactive MRI contrast agent and measuring the concentration of radioactive contrast agent in the post-injection MRI image as a continuous variable, each by using the adaptive neural network;
wherein the MRI image data from the patient or experimental animal before injection and the MRI image data from the patient or experimental animal after injection are each prepared from MRI tissue magnetization according to at least one of the following steps:
a. inverting the magnetization of the tissue and measuring any recovery of the tissue magnetization with further MRI imaging, or
b. perturbing the magnetization of the tissue with a known tip-angle, and measuring any decrease of tissue magnetization with further MRI imaging, or
c. perturbing the magnetization of the tissue with a set of known tip-angles and measuring the changes in tissue magnetization with further MRI imaging;
wherein the radioactive MRI contrast agent injected into the patient or experimental comprises a gadolinium labeled agent; and
wherein the training criterion image data are derived from autoradiographic images of tissue radioactive contrast agent administration to a mammal.
5. The method of claim 4 , wherein the radioactive MRI contrast agent comprises gadolinium labeled serum albumin.
6. The method of claim 5 , wherein the radioactive contrast agent used to generate the training criterion image data of known radioactive contrast agent concentration comprises radioactive serum albumin.