Quality control system for cardiac positron emission tomography images

Abstract

The present invention relates to a quality control system for generation of a concordance assessment of medical imaging data from the medical imaging unit, the infusion profile data from the infusion profile data unit, and / or patient characteristic's data using a computer executable algorithm stored in memory. The quality control system comprises a data processing unit, a memory that stores computer executable algorithm, a medical imaging data unit, an infusion profile data unit, and patient characteristic's data. The data processing unit determines a concordance assessment of the data; wherein: a) the concordance assessment is generated by comparing the medical imaging data from the medical imaging unit, the infusion profile data from the infusion profile data unit, and the patient characteristic's data using a computer executable algorithm stored in memory; b) a user warning is generated for excessive mismatches between infusion profile data, imaging data, and / or patient characteristic's data; and c) the infusion profile data unit is Sr-Rb elution system infusion data.

Description

QUALITY CONTROL SYSTEM FOR CARDIAC POSITRON EMISSION TOMOGRAPHY IMAGESTECHNICAL FIELD OF THE INVENTION

[0001] The present invention relates to a computer coupled radiopharmaceutical infusion system. More particularly, the present invention relates to a quality control method for generation of concordance assessment of medical imaging data from the medical imaging unit, the infusion profile data from the infusion profile data unit, and / or patient characteristics using a computer executable algorithm stored in memory.BACKGROUND OF THE INVENTION

[0002] Radiopharmaceuticals are important for various therapeutic and diagnostic use. A radiopharmaceutical infusion system is used to infuse a radiotracer in the body to diagnosis the patient through radio imaging devices such as PET and SPECT. The infusion profiles from radioisotope generators and infusion systems are correlated with blood / arterial input functions obtained from dynamic PET imaging used for modelling of myocardial blood flow and thus patient diagnosis and stratification, whereas new infusion modes (e.g., constant activity, constant time) produce blood input functions with different profiles than traditional bolus infusion. These new blood input functions may be unfamiliar to clinicians familiar with traditional bolus infusions. It is therefore, unclear if imaging results obtained with new infusion profiles are reliable.

[0003] Moreover, beyond the unfamiliar shape from new infusion modes, patient motion, pinched infusion lines, etc. may yield blood input functions that do not reflect the infusion profile. Traditionally, clinicians and technologists do not always perform visual quality control of these results. The accuracy of the blood input function is paramount to downstream analysis of blood flow, including estimations of myocardial blood flow (MBF) and myocardial flow reserve (MFR). It would be useful to have a program that runs in the background that provides a warning for excessive mismatches between an infusion profile coming from an Rb-82 generator / infusion system, such as RUBY FILL® and the images obtained from a PET machine. A confidence score of a ‘hormal” blood input function given a specific infusion profile, generator characteristics, and patient characteristics would inspire confidence in clinicians that adopt RUBY-FILL® and the constant activity mode.

[0004] Furthermore, if the part of the body for deriving the blood input function (e.g., the left ventricle cavity) is outside the field of view or overly corrupted without the possibility of performing the study again, but kinetic modelling of the imaged body parts is desired, it would be useful to have a program that estimates the most probable blood input function given thecharacteristics of the patient and information from the infusion system, such as Rb-82 generator / infusion system. Trained conditional variational autoencoders for performing quality control can be effortlessly repurposed to estimate the missing imaging data, such as the blood input function, to perform blood flow analysis. The present invention provides an additional advantage to use an application of such a program in peripheral artery disease where dynamic images of the lower limbs are obtained and the left ventricle cavity is outside the field of view. In the absence of a large field of view scanner or a second acquisition of the heart (which would require an additional radiation dose of Rb-82), the program can estimate the blood input function for performing kinetic modelling and flow analysis of the lower extremities within a single imaging acquisition.SUMMARY OF THE INVENTION

[0005] The present invention relates to a quality control method for generation of concordance assessment of medical imaging data from the medical imaging unit, the infusion profile data from the infusion profile data unit, and / or patient characteristics using a computer executable algorithm stored in memory.

[0006] The present invention relates to generating missing or overly corrupted medical imaging data conditional on infusion profile data and / or patient characteristics using the same computer executable algorithm stored in memory.

[0007] An aspect of the present invention is to provide an artificial intelligence or machine learning (AI / ML) method to yield robust quality control algorithms that can provide a confidence score for the reliability of the data.

[0008] An aspect of the present invention is to provide an artificial intelligence or machine learning (AI / ML) method or algorithm to create internal representations of “good quality” imaging data based on a reference database (here, of ‘good infusion profiles” and ‘good imaging data”) wherein imaging data is of diagnostic quality without any major artifacts, such as movement, and the infusion of a radiopharmaceutical has been successfully administered.

[0009] An aspect of the present invention is to provide a coupling between the Rb-82 infusion system and a medical imaging software, which is necessary for an artificial intelligence or machine learning (AI / ML) algorithm to compare both signals and determine a confidence score of data reliability / quality.

[0010] An aspect of the present invention is to provide a quality control system and method for generation of concordance assessment of the data, comprising: a data processing unit, wherein the data processing unit is configured to determine a concordance assessment of the data;a memory that stores a computer executable algorithm; a medical imaging data unit; an infusion profile data unit; and patient characteristics, wherein the concordance assessment is generated by comparing the medical imaging data from the medical imaging unit, the infusion profile data from the infusion profile data unit, and / or patient characteristics using a computer executable algorithm stored in memory.

[0011] An aspect of the present invention is to provide a quality control system for generation of concordance assessment of the data comprises: a data processing unit, wherein the data processing unit is configured to determine a concordance assessment of the data; a memory that stores a computer executable algorithm; a medical imaging data unit; an infusion profile data unit; and patient characteristics; wherein: a) the concordance assessment is generated by comparing the medical imaging data from the medical imaging unit, the infusion profile data from the infusion profile data unit, and patient characteristics using a computer executable algorithm stored in memory; and b) the infusion profile data unit is Sr-Rb elution system infusion data; wherein the system further comprises a user warning, wherein the user warning is generated for excessive mismatches between an infusion profile data, imaging data, and / or patient characteristics.

[0012] An aspect of the present invention is to provide a quality control system for generation of a concordance assessment of the data, comprising: a data processing unit; a memory that stores a computer executable algorithm; a medical imaging data unit; patient characteristics; and an infusion profile data unit; wherein:a) the memory that stores the computer executable algorithm is an artificial intelligence or machine learning algorithm to analyze the compatibility between obtained signals; and b) a data processing unit configured to determine a concordance assessment of two signals and / or patient characteristics; wherein the two signals are medical imaging data from medical imaging data unit and infusion profile data from infusion profile data unit (or other infusion profile data) and the patient characteristics are those affecting the dynamic concentration of a radioactive isotope in blood and tissue and the said concordance assessment is generated by comparing the medical imaging data, the infusion profile data, and / or patient characteristics; and wherein the system further comprises a user warning, wherein the user warning is generated for excessive mismatches between an infusion profile data, imaging data, and / or patient characteristics.

[0013] An aspect of the present invention is to provide a quality control system for the generation of missing or overly corrupted imaging data, comprising: a data processing unit; a memory that stores computer executable algorithm; a medical imaging data unit; patient characteristics; and an infusion profile data unit; wherein: a) the memory that stores computer executable algorithm is an artificial intelligence or machine learning algorithm to generate the most probable imaging data (infusion profile) conditional on patient characteristics and infusion profile data; and b) the infusion profile data from the infusion profile data unit (or other infusion profile data) and the patient characteristics are used for the estimation of the dynamic concentration of a radioactive isotope in blood and tissue, wherein the imaging data (e.g., blood input function) is generated by an estimate from the artificial intelligence or machine learning algorithm conditional on the given patient characteristics and infusion profile data.BRIEF DESCRIPTION OF DRAWINGS

[0014] Fig. 1 depicts the process flow diagram to generate the concordance assessment.

[0015] Fig. 2 depicts the process flow diagram of Al model, represented by a conditional variational autoencoder to notify the user based on excessive reconstruction error.

[0016] Fig. 3 depicts the process flow diagram of Al model, represented by a conditional variational autoencoder to notify the user when the latent encoding of poor quality data deviates from the good quality data.

[0017] Fig. 4 depicts the process flow diagram of Al model, represented by a conditional variational autoencoder to notify the user based on excessive reconstruction error based on the most probable imaging data, conditional on patient characteristics and infusion profile data.

[0018] Fig. 5 depicts the process flow diagram of Al model, represented by a conditional variational autoencoder to generate the most probable imaging data, conditional on patient characteristics and infusion profile data.DETAILED DESCRIPTION

[0019] The present invention can be more readily understood by reading the following detailed description of the invention and included embodiments.

[0020] The term “about” as used herein in the invention refers to a measurable value such as a parameter, an amount, a temporal duration, and the like, and is meant to encompass variations of and from the specified value, in particular variations of ±10% or less, preferably ±5% or less from the specified value, such variations are appropriate to perform in the disclosed invention. It is to be understood that the value to which the modifier ‘hbout” refers is itself also specifically, and preferably, disclosed.

[0021] As used in the specification of the present invention, the singular forms “a”, “an”, and “the” include plural references unless the context clearly dictates otherwise. Thus, for example, a reference to “a system” or “a device” or “a process” or “a composition” includes one or more systems, one or more devices, one or more processes or compositions, with one or more steps or ingredients or elements of the type described herein and / or which will become apparent to those persons skilled in the art upon reading this disclosure and so forth.

[0022] As used herein, the term “imaging” refers to techniques and processes used to create images of various parts of the human body for diagnostic and treatment purposes within digital health. Some imaging techniques are referenced but are not limited to X-ray radiography, Fluoroscopy, Magnetic Resonance Imaging (MRI), Computed Tomography (CT), Medical Ultrasonography or Ultrasound Endoscopy, Elastography, Hybrid Molecular Imaging for Image Guided Surgery, Tactile Imaging, Thermography Medical Photography, and nuclear medicine functional imaging techniques, e.g. Positron Emission Tomography (PET), Dynamic Positron Emission Tomography or Single-Photon Emission Computed Tomography (SPECT). Imaging seeks to reveal internal structures of the body, as well as to diagnose and treat disease.

[0023] As used herein, the term ‘SPECT” refers to a Single -Photon Emission Computed Tomography, a nuclear medicine tomographic imaging technique using gamma rays and providing true 3D information. This information is typically presented as cross-sectional slices through the patient, but can be freely reformatted or manipulated as required. The technique requires the delivery of a gamma-emitting radioisotope (a radionuclide) into the patient, normally through injection into the bloodstream. A marker radioisotope is generally attached to a specific ligand to create a radioligand and / or radiopharmaceutical, whose properties bind it to certain types of tissues. This allows the radiopharmaceutical to be carried and bound to a region of interest in the body, where the ligand concentration is assessed by a SPECT camera. The radioisotopes typically used in SPECT imaging are iodine-123 (1-123), indium-il l (In- 111), technetium-99m (Tc-99m), xenon-133 (Xe-133), thallium-201 (Tl-201), krypton-87m (Kr-81m), and gallium-67 (Ga-67).

[0024] As used herein, the term “Positron Emission Tomography (PET)” refers to a functional imaging technique that uses radioactive substances known as radiotracers or radiopharmaceuticals to visualize and measure changes in metabolic processes, and in other physiological activities including blood flow, regional chemical composition, and absorption. Different radiotracers can be used for various imaging purposes, depending on the target process within the body. The radioisotopes typically used in PET imaging are carbon- 11 (C- 11), nitrogen-13 (N-13), oxygen-15 (0-15), fluorine-18 (F-18), rubidium-82 (Rb-82), copper- 64 (Cu-64), zirconium-89 (Zr-89), and gallium-68 (Ga-68).

[0025] As used herein, the term “processing unit” refers to a computer or central processing unit which executes or processes the instruction as per the programmed and stored algorithm.

[0026] As used herein, the term “infusion profile data” refers to Sr-Rb elution system data such as injection rate, flow rate, infusion rate, infusion type, injected dose, age of generator, or performance of saline push, or instantaneous measures thereof during the infusion.

[0027] As used herein, the term “patient characteristics” refers to patient information such as age, sex, height, weight, or Body Mass Index (BMI).

[0028] As used herein, the term “medical imaging data encoder” basically refers to an auto encoder which encodes the image into a lower-dimensional representation.

[0029] As used herein, the term “medical imaging data decoder” refers to decode the lower dimensional representation into image data.

[0030] As used herein, the term “latent variable” refers to a low dimensional representation of observable variables.

[0031] As used herein, the term “latent encoding” refers to the values of the latent variable for a given image and / or conditional variables.

[0032] As used herein, the term “variational autoencoder” is an enhanced form of an autoencoder that incorporates regularization techniques to mitigate overfitting and ensure desirable properties in the latent space for effective generative processes.

[0033] In an embodiment of the present invention includes a quality control method, for generation of concordance assessment of the data, comprises: a data processing unit; wherein the data processing unit is configured to determine a concordance assessment of the data; a memory that stores computer executable algorithm; a medical imaging data unit; an infusion profile data unit; and patient characteristics; wherein the concordance assessment is generated by comparing the medical imaging data from the medical imaging unit, the infusion profile data from the infusion profile data unit, and / or patient characteristics using a computer executable algorithm stored in memory.

[0034] In an embodiment of the present invention includes a quality control system for generation of concordance assessment of the data comprises: a data processing unit; wherein the data processing unit is configured to determine a concordance assessment of the data; a memory that stores computer executable algorithm; a medical imaging data unit; an infusion profile data unit; and patient characteristics; wherein: a) the concordance assessment is generated by comparing the medical imaging data from the medical imaging unit, the infusion profile data from the infusion profile data unit, and patient characteristics using a computer executable algorithm stored in memory; and b) the infusion profile data unit is Sr-Rb elution system infusion data; wherein the system further comprises a user warning, wherein the user warning is generated for excessive mismatches between an infusion profile data, imaging data, and / or patient characteristics.

[0035] In an embodiment of the present invention includes a quality control system for generation of a concordance assessment of the data, comprises: a data processing unit; a memory that stores a computer executable algorithm; a medical imaging data unit; patient characteristics; and an infusion profile data unit; wherein: a) the memory that stores the computer executable algorithm is an artificial intelligence or machine learning algorithm to analyze the compatibility between obtained signals; and b) the data processing unit is configured to determine a concordance assessment of two signals and / or patient characteristics, wherein the two signals are medical imaging data from the medical imaging data unit and infusion profile data from the infusion profile data unit and the patient characteristics are those affecting the dynamic concentration of a radioactive isotope in blood and tissue and the said concordance assessment is generated by comparing the medical imaging data, the infusion profile data, and / or patient characteristics; wherein the system further comprises a user warning on excessive mismatches between an infusion profile data, imaging data, and / or patient characteristics.

[0036] An embodiment of the present invention includes a quality control system for the generation of missing and / or overly corrupted imaging data, comprising: a data processing unit; a memory that stores a computer executable algorithm; a medical imaging data unit; patient characteristics; and an infusion profile data unit; wherein: a) the memory that stores computer executable algorithm is an artificial intelligence or machine learning algorithm to generate the most probable imaging data (infusion profile) conditional on patient characteristics and infusion profile data; b) the infusion profile data from infusion profile data unit (or other infusion profile data) and the patient characteristics are used for the estimation of the dynamic concentration of a radioactive isotope in blood and tissue, wherein the imaging data(e.g., blood input function) is estimated by the given patient characteristics and infusion profile data.

[0037] An embodiment of the present invention includes a quality control method for generation of concordance assessment of the data, the method comprising: providing a data processing unit, wherein the data processing unit is configured to determine a concordance assessment of the data; providing a memory that stores computer executable algorithm; providing a medical imaging data unit and generating medical imaging data; providing an infusion profile data unit and generating infusion profile data; and providing patient characteristics; wherein the concordance assessment is generated by comparing the medical imaging data generated by the medical imaging unit, the infusion profile data generated by the infusion profile data unit, and / or patient characteristics using the computer executable algorithm stored in the memory.

[0038] An embodiment of the present invention includes a quality control system comprising a data processing unit, wherein the data processing unit is programmed with the artificial intelligence (Al) algorithm.

[0039] An embodiment of the present invention includes a quality control system comprising an artificial intelligence (Al) algorithm, wherein the Al algorithm can be one-class support vector machines or other anomaly / novelty detection algorithms, Boltzmann machines, deep belief network, neural autoregressive density estimators, variational auto-encoders or extensions thereof, generative adversarial network or extensions thereof, or other deep learning approaches.

[0040] An embodiment of the present invention includes a quality control system comprising an artificial intelligence (Al) algorithm, wherein the Al algorithm is a generative model to predict imaging data conditional on infusion profile data and / or patient characteristics.

[0041] An embodiment of the present invention includes a quality control system wherein the medical imaging data unit can be selected from the group consisting of PET, SPECT, CT, MRI, and / or combinations thereof.

[0042] An embodiment of the present invention includes a quality control system wherein the infusion profile data unit is a Sr-Rb generator and infusion system.

[0043] An embodiment of the present invention includes a quality control system wherein the infusion profile data can be selected from the injection rate, flow rate, infusion rate, infusiontype, injected dose, age of generator, performance of saline push or instantaneous measures thereof during the infusion.

[0044] An embodiment of the present invention includes a quality control system further comprising a user warning, wherein the user warning can be selected from any of visual, text and audio signals.

[0045] An embodiment of the present invention includes a quality control system, wherein the concordance assessment is related to the infusion-imaging data, which are concordant based on a dataset of “good quality” patient-infusion-imaging data. For example, the dataset can be previously generated and objectively evaluated by an expert to verify that it is of ‘^ood quality.”

[0046] An embodiment of the present invention includes a quality control system, wherein mismatches in the infusion data, imaging data, and / or patient characteristics are related to motion of the patient, poor placement of the left ventricle region-of-interest, pinching of the infusion line, pinching of arterial vessels from crossed arms, bent arms, etc.

[0047] An embodiment of the present invention includes a quality control system, wherein the infusion profile is obtained in constant flow, constant activity, or constant time mode.

[0048] An embodiment of the present invention comprises a quality control system, wherein the medical imaging data is a full dynamic imaging series (3D + time) or an average ID time series extracted from a 3D region of interest (ROI). The average ID time series may be an arterial input function used for kinetic modelling or any other signal, which captures arterial input of tracer. The 3D region of interest is determined automatically or manually.

[0049] An embodiment of the present invention includes a quality control system, wherein the medical imaging data is a decay time, corrected or not.

[0050] An embodiment of the present invention includes a quality control system, wherein the patient characteristics can be selected from weight, height, body mass index, age, sex, etc.

[0051] An embodiment of the present invention includes a quality control system, wherein the concordance assessment is a binary score (e.g., yes, no), ternary score (e.g., yes, no, unsure), a continuous variable reflecting the probability of concordance, or an absolute error of the predicted imaging data conditional on infusion profile data and / or patient characteristics vs observed imaging data.

[0052] Fig. 1 of the present invention illustrates the process flow diagram to generate the concordance assessment by comparing the medical imaging data from the medical imaging unit, the infusion profile data from the infusion profile data unit, and patient characteristics data using a data processing unit by applying an artificial intelligence (Al) algorithm.

[0053] Fig. 2 of the present invention illustrates the process flow diagram, wherein precise Al model is represented by a conditional variational autoencoder to notify the user, based on excessive reconstruction error. Further, the infusion profile data is processed by a medical imaging data encoder to make high dimensional infusion profile data into conditional low dimensional patient and infusion profile data. Moreover, the medical imaging data is also processed into low dimensional medical imaging data by using the medical imaging encoder. The latent variable model is applied on low dimensional imaging data to make data in continuous lower dimensional space. Afterward, the continuous lower dimensional imaging, low dimensional patient and infusion profile data are decoded by a medical imaging data decoder to get reconstructed medical image data. If any reconstruction error is found in the reconstructed medical image data, a warning will be issued to the user.

[0054] Fig. 3 of the present invention illustrates the process flow diagram wherein the artificial intelligence (Al) is a conditional variational autoencoder to notify the user when the latent encoding of poor quality data deviates from those of good quality data, of the nature described above. The medical imaging data of the present invention is processed into low dimensional medical imaging data by using the medical imaging encoder. The latent variable model is applied on low dimensional imaging data to make data in continuous lower dimensional space. If the latent variable deviates, a warning will be issued to the user.

[0055] Fig. 4 of the present invention illustrates the process flow diagram, wherein the artificial intelligence (Al) is a conditional variational autoencoder to notify the user when the latent encoding of poor quality data deviates from those of good quality data, of the nature described above. The medical imaging data of the present invention is estimated by setting the low dimensional latent vector z=0 and reconstructing the data, conditional on patient characteristics and infusion profile data. If the estimated reconstructed medical image data deviates, a warning will be issued to the user.

[0056] Fig. 5 of the present invention illustrates the process flow diagram, wherein the artificial intelligence (Al) is a conditional variational autoencoder to estimate medical image data when it is not available. The medical imaging data of the present invention is estimated by setting the low dimensional latent vector z=0 and reconstructing the data, conditional on patient characteristics and infusion profile data.

[0057] Each embodiment disclosed herein is contemplated as being applicable to each of the other disclosed embodiments. Thus, all combinations of the various elements described herein are within the scope of the invention. This invention will be better understood by reference to the drawings.

Claims

What is claimed:

1. A quality control method for generation of concordance assessment of one or more sets of data, the method comprising: providing a data processing unit, wherein the data processing unit is configured to determine a concordance assessment of the one or more sets of data; providing a memory that stores a computer executable algorithm; providing a medical imaging data unit and generating medical imaging data; providing an infusion profile data unit and generating infusion profile data; and providing patient characteristic’s data; wherein the concordance assessment is generated by comparing the medical imaging data generated by the medical imaging unit, the infusion profile data generated by the infusion profile data unit, and / or the patient characteristic’s data using the computer executable algorithm stored in the memory.

2. A quality control system for generation of a concordance assessment of one or more sets of data, comprising: a data processing unit; a memory that stores a computer executable algorithm; a medical imaging data unit; an infusion profile data unit; and a set of patient characteristic’s data; wherein: a) the memory that stores computer executable algorithm is an artificial intelligence or machine learning algorithm to analyze the compatibility between obtained signals; and b) the data processing unit is configured to determine a concordance assessment of two signals and / or patient characteristics, wherein the two signals are medical imaging data from the medical imaging data unit and infusion profile data from the infusion profile data unit and the patient characteristics are those affecting the dynamic concentration of a radioactive isotope in blood and tissue and the said concordance assessment is generated by comparing the medical imaging data, the infusion profile data, and / or patient characteristic’s data; and wherein the system further comprises a user warning, wherein the user warning is generated for excessive mismatches between the infusion profile data, the medical imaging data, and / or the patient characteristic’s data.

3. The method according to claim 1, wherein a data processing unit is programmed with the artificial intelligence (Al) algorithm.

4. The system according to claim 3, wherein the Al algorithm comprises one-class support vector machines or other anomaly / novelty detection algorithms, Boltzmann machines, deep belief network, neural autoregressive density estimators, variational auto-encoders or extensions thereof, generative adversarial network or extensions thereof, or other deep learning approaches.

5. The system according to claim 3, wherein the Al algorithm is a generative model to predict imaging data conditional on the infusion profile data and / or the patient characteristic’s data.

6. The method according to claim 1, wherein a medical imaging data unit is selected from the group consisting of PET, SPECT, and / or combinations thereof.

7. The method according to claim 1, wherein the infusion profile data unit is a Sr-Rb generator and infusion system.

8. The system according to claim 1, wherein the infusion profile data is selected from one or more of injection rate, flow rate, infusion rate, infusion type, injected dose, age of generator, performance of saline push, or instantaneous measures thereof during the infusion.

9. The system according to claim 2, wherein the user warning comprises one or more of visual, text and audio signals.

10. The method according to claim 1, wherein the concordance assessment is related to the infusion-imaging data, which are concordant based on a dataset of “good quality” patientinfusion-imaging data.

11. The system according to claim 2, wherein the mismatches in the infusion profile data, imaging data, and / or patient characteristic’s data are related to one or more of motion of the patient, poor placement of the left ventricle region-of-interest, pinching of the infusion line, pinching of arterial vessels from crossed arms, and bent arms.

12. The method according to claim 1, wherein the infusion profile is obtained in constant flow, constant activity, or constant time mode.

13. The method according to claim 1, wherein the medical imaging data is a full dynamic imaging series or an average ID time series extracted from a 3D region of interest and wherein the average ID time series is an arterial input function used for kinetic modelling or a signal that captures arterial input of tracer and the 3D region of interest is determined automatically or manually.

14. The method according to claim 1, wherein medical imaging data is decay time-corrected.

15. The method according to claim 1, wherein the patient characteristic’s data is selected from one or more of weight, height, body mass index, age, sex, and combinations thereof.

16. The method according to claim 1, wherein the concordance assessment is a binary score, ternary score, a continuous variable reflecting the probability of concordance, or an absolute error of the predicted imaging data conditional on infusion profile data and / or patient characteristic’s data vs observed imaging data.

17. A quality control system for the generation of missing and / or overly corrupted imaging data comprising: a data processing unit; a memory that stores a computer executable algorithm; a medical imaging data unit; patient characteristic’s data; and an infusion profile data unit; wherein: a) the memory that stores computer executable algorithm is an artificial intelligence or machine learning algorithm to generate the most probable imaging data (infusion profile) conditional on patient characteristic’s data and infusion profile data; and b) the infusion profile data from the infusion profile data unit (or other infusion profile data) and the patient characteristics are used for the estimation of the dynamic concentration of a radioactive isotope in blood and tissue, wherein the imaging data (e.g., blood input function) is estimated based on the given patient characteristic’s data and the infusion profile data.

18. The method according to claim 17, wherein the generation of missing and / or overly corrupted medical imaging data is conditional on infusion profile data and / or patient characteristic’s data using the same computer executable algorithm stored in memory.

19. The method according to claim 17, wherein the artificial intelligence (Al) is a conditional variational autoencoder to notify the user when the latent encoding of poor quality data deviates from those of good quality data.

20. The method according to claim 17, wherein the medical imaging data is estimated by setting the low dimensional latent vector z=0 and reconstructing the data, conditional on the patient characteristic’s data and the infusion profile data, wherein if the estimated reconstructed medical image data deviates, a warning will be issued to the user.

Images