Semi-Automatic Ultrasonography Detection of Local Twitch Response Intensities During Dry Needling: Implications on Joint Range of Motion

The aim of this report was to evaluate the reliability of semi-automatically tracking local twitch responses (LTR) intensities during dry needling by assessing ultrasound signals with an optical flow method designed to estimate movement of ROIs, in addition to evaluating the relationship between LTR intensities and muscle elasticity.

Experimental protocol and measurements For each subject, the ROM of the medial gastrocnemius (MG) and hamstring (HA) was determined in both lower limbs using weight bearing and the active knee extension test, respectively, using a universal goniometer. For the knee extension test, 180° were subtracted from the obtained range, and the values were presented as negative. Both tests present excellent intra- and inter-rater reliability.

All participants were placed in a prone position with the ankle locked at 90º. Latent MTrPs were identified in both MG using according the presence of a palpable taut band and hypersensitive spot in the palpable taut band as diagnostic criteria. The ultrasound transducer was positioned over the MG muscle using a fixing device that was composed of a thermoplastic polymer, with enough space for the dry needling (DN) technique to be performed.

The deep DN was administered by a physical therapist with three years of experience in the technique. The ultrasound video (SonoSite Titan; Sonosite, Bothell, WA, USA) was recorded at 60 frames per second captured through an external capture device from Epiphan Systems Inc. (Ottawa, Ontario, Canada). According to the perception of the physical therapist and the ultrasound assessment, the filament needle was inserted into the MTrP to achieve three LTRs. Posterior to DN application, the ROMs of the MG and HA were measured.

The recorded ultrasound video was examined by the Camtasia software (Tech-Smith Corp, Okemos, MI, USA) to select three LTRs for latter analysis. Each LTR was tracked by the Lucas-Kanade optical flow (LKOF) algorithm with affine optic flow extension [12]. All analyses were performed using the Matlab® 2014 software (Mathworks Inc., Natick, MA, USA). For tracking muscle thickness, two ROIs (size 250 x 70 pixels) positioned on the superficial and deep aponeurosis were manually selected.