In summary, we found an overall trend towards reduced impedance at all meridian segments compared to adjacent controls at 10 and 100 kHz frequency, although this decrease was statistically significant only at the LI meridian segment. The differences in impedances between channel segments were partially attributable to SQ echogenic density. Greater SQ echogenic densities were associated with reduced impedances in both within-site and between-site analyses. This relationship remained significant in multivariable analyses which also accounted for gender, penetration depth, SQ layer thickness, and other ultrasound-derived measures.
Compared to our previous study, this study focused on the subcutaneous tissue at depths of 1–2 cm and not the intermuscular connective tissue. The tight spacing, oblique depth angles, curved longitudinal trajectory, and depth of the intermuscular fascia presented a major challenge to the precise placement of needles in the intermuscular spaces. Furthermore, the intermuscular planes did not consistently coincide with the location of meridians as determined by the acupuncturists. For these reasons, the meridian segments in this study were located by the acupuncturists, while ultrasound images were used as objective markers for any underlying physical structures that may account for differences in electrical impedance.
In addition, compared to the previous study, this study used a widely accepted impedance gain-phase analyzer – Solartron 1260 – and a compatible biological interface - Solatron 1294 – to obtain impedances across multiple frequencies. The low electrical impedances associated with the uninsulated needles, however, limited the accurate range (based on manufacturer specifications) of the Solartron 1294 biological interface to between 10 and 100 kHz. This differs with the 3.3 kHz frequency used in the original study.
Despite these differences, our study identified lower impedances at the LI meridian segment compared to its adjacent control at both frequencies. Mean impedances at the other two meridian sites showed a trend toward reduced levels for both frequencies although neither reached statistical significance. This result was similar to the findings identified at the Pericardium meridian (significantly reduced impedance) and Spleen meridian (no significant difference) in our previous study. Why this significant relationship occurred at the LI and PC meridian site and not the others is unclear. It is possible that the 1.3 cm separation distance between meridian and control segments was insufficient for the lower extremities.
Subcutaneous echogenicity was significantly associated with impedance in within-site analyses. In between-site analyses, electrical impedance was invariably below 450 Ohms for SQ echogenic densities above 23%. This inverse association between SQ echogenicity and impedance remained significant even after accounting for SQ thickness, gender, presence of large vessels and other potential confounders. Furthermore, this relationship existed at all tested body sites, strongly suggesting the generalizability of this association to other body areas. Additionally, this relationship may not be restricted to the SQ layer alone. Bivariable analyses revealed that dermal echogenicity was also inversely related with impedance. Meanwhile, the insignificant association between PM zone echogenicity and electrical impedance may be attributed to the insertion of needle electrodes to the depths of SQ thickness. In such cases, the needle did not frequently penetrate the PM layer and it is theoretically possible for the impedances to be registered at higher levels when electrical currents are shunted away from the inner voltage sensing electrodes towards deeper PM layer (thus explaining the higher impedances seen with greater PM echogenicity seen in ).
Based on biophysical and histological studies, these echogenic bands/areas are highly correlated with the location of collagen fibers and more specifically the bands of coalesced collagen fiber. 
. Consequently, our results suggest that collagen may be the macromolecule responsible for the observed lower impedances. Other studies support the evidence that collagen is characterized by low electrical impedance. A recent MREIT (Magnetic Resonance Electrical Impedance Tomography) study, for instance, performed conductivity imaging of the lower extremity in four healthy volunteers and observed significantly greater conductivities in the crural fascia and intermuscular septum (both areas known to contain large amounts of collagen) compared to muscle 
. The authors, however, ascribed this observation to presence of “conductive fluid in those regions”. Another study measured electrical impedance in irradiated and normal muscles in rat hind legs and found reduced impedances in irradiated legs proportional to the amount of radiation administered. The authors concluded that the “presence of interstitial fibroplasia [as a result of irradiation] may be providing a short around the muscle tissue, reducing the intracellular structural effects of muscle‥”, although direct, quantitative evidence for this hypotheses was not given 
. A third study evaluated the electrical impedance of normal and chronically infarcted left ventricular myocardium from sheep and reported a highly significant correlation between impedance and collagen content based on histological analyses 
From a theoretical standpoint, other authors have conjectured that the highly ordered, crystalline arrangement of collagen would confer it with various semiconductive properties, although no direct, objective data were provided in such cases 
. To our knowledge, this study is the first to demonstrate a statistically significant association between subcutaneous collagenous bands and electrical impedance in vivo
. The ultrasound images provided the means to not only quantify SQ collagenous bands but also account for other subcutaneous structures that may influence impedance measures and thus potentially confound the observed relationship. If the low impedance property of collagen and collagenous bands is further corroborated, the use of impedance-based devices for diagnostic purposes (such as BIA) would have an additional biological foundation particularly for conditions where the collagen and extracellular matrix are altered. This may include scarring from radiation or injury, inherited collagen disorders, or cancers where distortions of extracellular matrix by malignant cells are increasingly being recognized 
From the perspective of acupuncture, this study suggests that SQ collagenous bands may underlie the reduced impedance described at acupuncture meridians. Interestingly, SQ echogenicity was the only recorded variable associated with impedance in both within-site and between-site analyses. In other words, this association was not only generalizable to multiple body sites but also specific enough to account for local differences. Nevertheless, the physiological significance of this relationship remains unclear. The magnitude of differences between channel segments was small and unlikely to be physiologically significant. The use of uninsulated needles may partly explain this small difference since the shaft acts to short-circuit all the contacted tissue and thus average across multiple layers. The four electrode arrangement may also attenuate significant differences between channels, especially if currents are shunted away from current sensing electrodes towards an adjacent low-impedance pathway or if the segments between the current and measuring electrodes have high impedances (zone of negative sensitivity) 
. A more focused, depth-specific approach using insulated needles will be needed to appropriately evaluate the impedance of collagen fibers and to investigate its physiological importance, if any exists.
There are additional limitations to this study. First, due to added technical distortions stemming from the IEC-601 compliant biomedical inputs (Solartron 1294), information on impedance phase was essentially lost. Such information would greatly clarify how the meridians differed and how collagenous bands are associated with reduced impedance. Second, the needle electrodes were not inserted in muscle. The effects of perimuscular fascia and muscle were not directly assessed. It is possible that these deeper layers also play a role in local electrical impedance. Third, the meridian segments were located by acupuncturists and not by an objective standard. As a result, the reported differences (or lack thereof) in impedance between channel segments are subject to bias. Finally, the study is limited by small sample size. A larger study sample may have generated more robust results that can be interpreted with better confidence.
Despite these limitations, this study adds substantial insights into the electrical impedance associated with acupuncture meridians and connective tissue. Collagenous bands, represented by increased ultrasound echogenicity, are significantly associated with lower electrical impedance and may be the common denominating factor for the reduced impedances reported at certain acupuncture meridians at both subcutaneous and intermuscular depths. If so, the SQ collagenous bands may present an objective basis for the hitherto elusive acupuncture meridians. Furthermore, the study's findings may provide important insights into the nature of acupuncture-related treatments and into the relevance of collagen in bioelectrical measurements.