Ultrasonic scanning of straight micro tools in soft biological tissues: Methodology and implementation

The acoustical behaviour of straight micro tools (SMT), such as sub-millimetre needles or electrodes, is delicate and their automated maintaining in the image plane displayed to the practitioner must handle the extremely varying signal they backscatter. Because of the complex nature of the scattering by a narrow rod target, the localization process has to face two main tasks before being able to conduct 3D location: the reduction of the apparent diameter when this latter is visible, and the detection of the instrument and its real depth of penetration. In this paper, a methodology based on these steps is proposed, along with preliminary implementations of each of the steps. This methodology exploit the acoustic duality of those SMT and treat different acoustic aspects separately: first the apparent diameter is reduced through a deconvolution process, associating the backscattering to a purely linear acoustic process without attenuation; second, the imaging is treated as a pure attenuation process, and the signal is summed post-ROI to achieve the detection. Finally, the complete localisation is achieved by combining the two first steps with a previously tested global 3D straight line location technique: the Parallel Integral Projection. Experiments are conducted in vitro with 700 μm graphite rods and 150 μm in diameter tungsten electrode crossing gelatine and cryogel phantoms, and ex vivo on a porcine liver to assess the efficiency of the proposed implementations and of the whole methodology. The volumes were acquired with a Kretztechnic 530D scanner customized so as to provide the raw RF data. Results show that the combination of the different step-methods reduces progressively the apparent diameter, and that the detection step is essential to the proper localization of the micro tool. The final resolution improvement is at least of 3 and estimated to be typically of 12, for a final estimated apparent diameter reaching the average 400 μm resolution cell in the porcine liver.