Harmonic Imaging: tissues and microbubbles

The phase inversion mode was used in a large multi-centre study of liver metastases using Levovist. It showed that more lesions were demonstrated in phase inversion than on baseline B-mode scanning and that the sensitivity for metastatic disease was greatly increased; the specificity was also somewhat increased. The smallest lesion that could be detected was improved and was somewhat better than on contrast enhanced 3 phase helical CT. In addition there were some lesions detected on PIM that were not seen on CT, raising the question whether they were actually false positives. However, some patients in this study received a further independent diagnostic technique (MRI, intraoperative ultrasound or histology) and PIM ultrasound was closer to this ultimate reference standard than CT thus suggesting that these additional lesions were true positives that had been missed on CT.

These techniques all used a high MI which destroyed the microbubbles and therefore was difficult to apply to real time scanning. This was frustrating because the true haemodynamics of lesion blood supply in the arterial phase could not be interrogated easily. For this reason lower MI non-destructive techniques were actively investigated. The key element in this approach is too avoid high MIs which inevitably elicit tissue harmonics, and to develop methods to image the microbubbles using low or very low MIs. The phase inversion technique, though originally developed for high MI work with agents like Levovist, lends itself very well to this approach especially with SonoVue and, though the final images remain a mixture between such tissue harmonics as are elicited and the (dominant) microbubble harmonics, it is a widely used method. More sophisticated approaches use more complex multi-pulse techniques involving alterations of amplitude as well as phase which enable the detection of non-linearities in the fundamental frequency band (they detect non-linearities arising from temporal asymmetries in compression and rarefaction of the microbubbles) and this means they are more sensitive or can operate at deeper depths. These methods allow the presentation of the microbubble information in the colour layer by analogy with colour Doppler, with the fundamental tissue information (albeit usually somewhat degraded on account of the low transmit powers required) as an underlying greyscale image that is useful for reference purposes. Obviously these low MI methods are not useful for tissue harmonic imaging!

With both these families of low MI approaches, real time scanning can be used and this has the major advantage of allowing the haemodynamics of the transit of the microbubbles to be examined. The arterial phase is important in many organs; in the liver the portal phase is also available, though it does not seem to add much diagnostic information. However, the sinusoidal (late) phase seems to have the same critical information as the Kupffer cell phase of agents such as Levovist in indicating the vascular volume of the lesion compared to the liver or spleen. Not only does the vascularity of the lesion appear but the pattern of arterial supply is evident: hypovascular malignancies are supplied from the periphery via one or two tortuous or helical vessels whereas focal nodular hyperplasia receives a central supply. Haemangiomas also receive an arterial supply but it is less intense and the arteries discharge into the capacious angiomatous vessels at the periphery of the lesion. The contrast then percolates slowly through the angiomatous vasculature so that the lesion fills completely or partially in a centripetal fashion. The analogy with CT or MRI is obvious and, as with both these other techniques, typical lesions are definitive diagnostically but there remains a proportion that is atypical and cannot be diagnosed in this way.