Self-adaptive transcranial ultrasound imaging using an effective skull model: experimental evaluation on a human skull at 3.15 MHz

Introduction. Transcranial ultrasound imaging is hindered by skull-induced aberrations that degrade focusing and image contrast. Existing correction methods typically rely on external imaging or detailed acoustic models, limiting their applicability in emergency settings. Methods. We propose an adaptive approach based on an effective skull model comprising two interfaces and an effective velocity, estimated directly from Full Matrix Capture data without external information. The estimated parameters are integrated into a Total Focusing Method reconstruction with refraction correction. The method is evaluated experimentally at 3.15 MHz by imaging wire targets located beyond a human skull fragment in four configurations: free-field reference, uncorrected, skull corrected with automatically estimated parameters, and skull corrected with optically scanned surfaces. Results. The effective model reduces the mean Euclidean localization error by 67% (2.16 to 0.71 mm). Comparison with the optical reference isolates interface estimation uncertainty as the dominant residual error source, while confirming that the effective velocity model itself produces comparable performance when provided with accurate interface positions. Discussion. These results demonstrate that a self-estimated model enables reliable localization in transcranial conditions. Improving surface estimation robustness is identified as a key factor for precise target reconstruction.