Introduction
Compound muscle action potential (CMAP) scans are detailed stimulus-response curves that provide information on the activation of motor units (MUs) and can be used to provide a motor unit number estimate (MUNE). As loss of MUs is the hallmark of amyotrophic lateral sclerosis (ALS), accurate MUNE will be a valuable biomarker for monitoring disease progression in patients with ALS. MScanFit, the current gold standard in CMAP scan-based MUNE, is efficient and widely used, but underestimates MUNE in muscles with a high number of MUs.

Methods
The current study introduces Wavolution, a waveform-based algorithm that estimates MUNE from CMAP scans. Whereas MScanFit assigns solely numerical values to MU properties in its fitting procedure, Wavolution assigns single MU action potentials (SMUAPs) to MUs to account for physiological phenomena such as phase cancellation and temporal dispersion present in CMAP recordings. Furthermore, Wavolution has a highly efficient way of simulating CMAP scans by matrix multiplication. The fitting procedure of simulated CMAP scans onto a target CMAP scan consists of two stages: (1) an initialization phase that generates preliminary MU pools based on amplitude and threshold density distributions of the input scan, and (2) an optimization phase, in which a genetic-like algorithm iteratively updates the simulated MU pools to minimize the error between simulated and target CMAP scans.

Results
Wavolution was validated on 920 simulated CMAP scans, ranging from 5 to 150 MUs with each MU number containing various noise levels ranging from 1.0 to 100.0 μV. Wavolution achieved significantly lower percentual MUNE discrepancy compared to MScanFit (13.2% vs. 25.5%) and had a significantly reduced computation time (67.8 vs. 92.9 seconds per scan). Wavolution achieved higher accuracy in the mean amplitude size of the MUs in the estimated MU pool than MScanFit when MU counts are high and reproduced reductions in maximum CMAP amplitude caused by phase cancellation and temporal dispersion, although it tended to overestimated their magnitude.

Discussion
Wavolution provides a physiologically transparent and computationally efficient approach to CMAP scan–based MUNE. By modeling SMUAP waveforms, it accounts for phase cancellation effects and reduces the ceiling effects of MScanFit in muscles with high MU counts, where muscles are still largely intact. A Windows application was developed to enable bulk processing, reducing operator workload in large-scale ALS trials. Future research should focus on extending the algorithm to make full use of the information available in CMAP recordings, validating Wavolution on experimental CMAP scans, and improving noise estimation. With further development, Wavolution could become a valuable tool for disease monitoring in ALS clinical trials.