PMMA (Stryker Corporation, Howmedica®, Kalamazoo, MI, USA) mixed according to specifications on the package insert was made into a 1.3-mm diameter, 8-mm long cylindrical construct comparable in size to conventional biopsy markers by extruding the PMMA from a 15-gauge hole punched into the hub of a needle attached to a 3-cc syringe.
Based on earlier developed techniques , a non-contact three-dimensional (3D) coherence scanning interferometer optical profiler (Zygo Corporation, Middlefield, CT, USA) measured the areal surface roughness (Sa) of four commercial metallic breast biopsy markers: TriMark® cork (Hologic®, Marlborough, MA, USA), Tumark® Q (Hologic®), UltraClip™ ribbon (Becton, Dickinson & Co., Franklin Lakes, NJ, USA), and SenoMark™ O clip (Becton, Dickinson & Co.). Optical surface characterization to measure Sa was performed using a consistent magnification of × 20 for all markers. Overall shape and curvature of the markers were removed from the surface characterization using 4th order polynomial curve fitting of the optical measurement data .
Scanning electron microscopy (SEM) (Hitachi S-4700, Hitachi High-Tech in America, Schaumburg, IL, USA) images captured the surface irregularities of PMMA and the metallic biopsy markers. Porosity was determined using micro-computed tomography (SkyScan 1272, Bruker Corporation, Allentown, PA, USA) using provided computed tomography analyzer software (CTAn, Bruker Corporation) based on thresholding and regions of interest on 15-μm slice thicknesses .
Ultrasound of the four commercial markers and the PMMA marker was performed in a gel phantom and ex vivo in pork belly meat using a clinical system (Logiq E9, General Electric Healthcare, Wauwatosa, WI, USA) with 9-L and ML6-15 linear array transducers, both generally used in breast ultrasound, and a C1-6 curvilinear transducer typically used in abdominal ultrasound (General Electric Healthcare). To minimize experimental bias, the markers were placed at roughly the same depth between 1 and 2 cm deep and spaced minimally apart so that they could be scanned simultaneously. For the phantom study, two gel phantoms were stacked on top of each other to minimize backscatter from the tabletop.
Scanning parameters such as ultrasound transmit frequency, color scale, and gain were adjusted to optimize twinkling. For radiological assessment, a twinkling score was defined from 0 (least twinkling and least confident detection) to 4 (most twinkling and most confident detection) . In general, a twinkling score of 3 or 4 would provide sufficient confidence for a breast radiologist to place an 125I seed next to it for localization without definite visualization of the marker on B-mode imaging. A twinkling score of 2 and below would require additional imaging features or information before an 125I seed would be used to localize it.