E1799. Arterial Input Function Selection on DCE MRI Perfusion in Spine Metastases: Automatic Aortic Selection vs Phase-Delay Correction
  1. Felipe Camelo; Memorial Sloan Kettering Cancer Center; Weill Cornell Medicine
  2. Julio Arevalo Perez; Memorial Sloan Kettering Cancer Center
  3. John Lyo; Memorial Sloan Kettering Cancer Center
  4. Andrei Holodny; Memorial Sloan Kettering Cancer Center
  5. Kyung Peck; Memorial Sloan Kettering Cancer Center
DCE-MRI analysis relies on a pharmacokinetic model, which assumes that contrast uptake is simultaneous in arteries and tissues of interest. While this is virtually true in highly vascularized tissues like the brain parenchyma, the spine bone marrow is less perfused than the brain and this assumption of simultaneous contrast uptake can affect DCE-MRI analyses, specifically the fractional plasma volume (Vp) parameter. Quantitative analysis requires the detection of the arterial input function (AIF), which is automatically obtained from the aorta in spine studies. Practically, the accuracy of Vp has been shown to be affected by delayed response between AIF and tissue of interest. This study evaluated cases of patients with various metastatic spinal lesions where conventional MRI was deemed sufficient for the diagnosis of malignancy, but DCE-MRI analysis showed a false negative on the Vp map. We hypothesized that shifting the AIF tracer curve would affect the accuracy of Vp metastatic lesions.

Materials and Methods:
We retrospectively analyzed 18 patients without prior radiation treatment to the spine, who completed DCE-MRI perfusion imaging with discordant results between Vp and conventional imaging and who underwent vertebral biopsy confirming the malignancy. We manually shifted the AIF curve by delaying it in multiple phases, defined as the scan-specific phase temporal resolution, and analyzed DCE-MRI parameters with the new AIF curves. A Mann-Whitney U test at a significance level of corrected p = .01 was used to assess differences between phase shifts.

All patients were found to require at least one phase shift delay in the calculated AIF to both better visualize metastatic spinal lesions on Vp maps and increase the value of Vp. The average Vp values were 1.78 +/- 1.88 for zero phase shifts (P0), 4.72 +/- 4.31 for one phase shift (P1), and 5.59 +/- 4.41 for two phase shifts (P2). Mann-Whitney U tests were performed and obtained p-values between P0 and P1 of 0.003, as well as between P0 and P2 of 0.0004.

This study demonstrates that the spinal metastatic lesions could show contrast uptake at a delayed rate compared to the aorta. Based on these findings, radiologists and technologists should take this phenomenon into consideration, especially when images from Vp maps are not concordant with suspicious conventional MRI findings. It is important that both analysts and software developers are aware of these pitfalls that can erroneously increase the number of false negative perfusion results. Artificially delaying the AIF curve, simulating the delay between the aorta the capillaries within the bone marrow, could be a potential solution to improve Vp.