Blind estimation of the arterial input function in MRI to model fluid transport in tissues and tumours

Measurement of tissue blood flow (perfusion) and related fluid transport parameters using contrast-enhanced MRI is a valuable technique in the clinic, but these parameters are difficult to quantify without knowing the concentration of the contrast agent in the arterial blood (the arterial input function, AIF). There have been many approaches described to estimate the AIF when it is not possible to measure it directly. One such approach is called blind estimation where the input is estimated from a series of tissue “outputs” only. However, most studies to date suffer from a lack of reference data (i.e., a known AIF) for validation purposes. We have two large reference data sets in Leeds as well as lots of clinical data which could be used to assess a blind estimation tool. In this project the student will develop such a tool, test its validity using our reference data then assess its application in other data sets to estimate blood flow and related fluid transport parameters. These parameters will be used to assess early response to treatment and predict outcomes in a range of clinical applications building upon the work of a previous CDT student.​

MR images obtained from a reference data set. A patient with a large tumour in the left breast imaged before (top left) and after 1 cycle of chemotherapy (top right). In this case the AIF (bottom left) is measured directly from the descending aorta before (blue data points) and after treatment (red data points). By measuring the AIF, we were able to estimate blood flow to the tumour (from the data bottom right) and show that despite a 30% reduction in tumour volume in response to the treatment, blood flow to the tumour remained the same. Following the completion of 6 cycles of chemotherapy the patient underwent surgery and the excised tumour was assessed to have not responded to that treatment.