Motor neuron disease (MND) is a devastating neurodegenerative disease. Diagnosis requires invasive, often painful, intramuscular myoelectric signal recordings from multiple body regions to identify the occurrence of electrophysiological events such as fasciculations [involuntary firing of a motor unit]. It has recently been shown that fasciculations can be seen by eye in ultrasound images collected at ~75Hz (1) and that assessment of images provides more sensitive detection than myoelectric signals in some muscles, e.g., biceps brachii (2). These studies are however based on manual, visual assessment of images to identify the occurrence of fasciculations. Here an automated, quantitative approach based on computationally tracking movement of features in collected images is assessed to determine the feasibility of using such techniques to automatically identify the occurrence of fasciculations. Development of such non-invasive, quantitative methods could facilitate earlier screening and diagnosis of suspected cases of MND. Nine human participants completed the study. Three participants, recruited through the MND Care and Research Centre at Royal Preston Hospital [RPH], had previously been diagnosed with MND. The study was approved by authorities at RPH and the universities local ethics committee. All participants provided informed written consent. In all participants B-mode ultrasound images simultaneously showing medial gastrocnemius [MG] and soleus [SO] were recorded. In those with MND, images were also collected from biceps brachii. The occurrence of fasciculations [localized tissue displacements lasting 0.2-1s (3)] were manually identified by two experienced operators. Features in each image were mathematically identified and tracked between images using methods described in (4). Resulting feature data were used in a mutual information [MI] analysis to provide a statistical classification of movement of tracked features between images. Correspondence between manually identified fasciculations and spikes in the MI metric was quantified by compiling receiver operating characteristics to identify the true positive rate at which the number of false positives and true negatives were equal [HEE]. For MG/SO image sequences HEE was 87% in healthy participants and 79% for participants with MND. HEE was 81% in sequences recorded from biceps. These strong preliminary results indicate excellent potential for the development of quantitative, automated detection of fasciculations in ultrasound images. The MI metric was not sensitive to the smallest manually identified movements and was influenced by the inherent noise in collected images. Further improvements in accuracy can therefore be expected with the application of more sophisticated analysis approaches.