The radiological detection of BMs is essential for optimizing a patient’s treatment. This statement is even more valid when stereotactic radiosurgery (SRS), a non-invasive image guided treatment that can target BM as small as 1-2 mm, is delivered as part of that care. The timing of image acquisition after contrast administration can influence the diagnostic sensitivity of contrast enhanced MRI for BM. Objective: Investigate the effect of time delayed acquisition after administration of intravenous Atavist® (Gadobutrol 1mmol/ml) on the detection of BM. Methods: This is a prospective IRB approved study of 50 patients with BM who underwent post-contrast MRI sequences immediately after injection of 0.1 mmol/kg Gadavist® as part of clinical care (t0), followed by axial T1 sequences after a 10 minutes (t1) and 20 minute delay (t2). MRI studies were blindly compared by 3 neuro-radiologists. Results: Single measure intraclass correlation coefficients were very high (0.914, 0.904 and 0.905 for t0, t1 and t2 respectively), corresponding to a reliable inter-observer correlation. The t2 delayed sequences showed a significant and consistently higher diagnostic sensitivity for BM by every participating neuroradiologist as well as for the entire cohort (p=0.016, p=0.035 and 0.034 respectively). A disproportionately high representation of BM detected on the delayed studies was located within posterior circulation territories (compared to predictions based on tissue volume and blood-flow volumes). Conclusion: Considering the safe and potentially high yield nature of delayed MRI sequences, it should supplement the basic MRI sequences in all patients in need of precise delineation of their intracranial disease.
Convolutional Neural Networks for Direct Inference of Pharmacokinetic Parameters: Application to Stroke Dynamic Contrast-Enhanced MRI
