Radiation therapy with ion beams provides a better conformation and effectiveness of the dose delivered to the tumor with respect to photon beams. This implies that a small uncertainty or variation in the crossed tissue shape and density may lead to a more important underdosage of the tumor and/or an overdosage of the surrounding healthy tissue. Although the online control of beam fluence and transverse position is well managed by an appropriate beam delivery system, the online measurement of the longitudinal position of the Bragg peak inside the patient is still an open issue. In this paper we propose a proof-of-concept study of a technique that would allow the online verification of the patient thickness along the beam direction, which could permit detecting a subset of possible range error causes, such as morphological variations. The nuclei 12 C and 4 He have the same magnetic rigidity: the two species could be accelerated together in an accelerator and a mixed particle beam delivered to the patient. In the same medium and with the same energy per nucleon, the range of 4 He2+ is about three times the 12 C6+ one. It is, thus, conceivable to achieve a dual goal with a single mixed beam: carbon, stopping into the tumor, is appointed to cure, while helium, emerging from the patient, to control: by detecting and measuring the residual range and position of He, it would be possible to determine the integrated relative stopping power of the patient and prove that it is the expected one. For the detection of helium particles, a plastic scintillator and an optical sensor are proposed. Being helium ions not available at CNAO, the detection system has been characterized using a proton beam. Nevertheless, since the light emitted by a proton is less than the one produced by a helium ion, the helium signal is expected to be more pronounced than the proton one (for the same number of particles). To predict the magnitude of the light signal measured by the sensor, two Monte Carlo models have been setup and validated by measuring the photons per pixel impinging on the sensor. To deal with the many optical issues and to reliably describe the physical process, some corrections have been included into the models. The predictions of both the models are in good agreement with the measurements (within the 20% in terms of absolute photons per pixel). The proposed detection system is able to measure the range of a proton beam with sub-millimetrical precision also in the presence of the background produced by carbon ion fragments and discrepancies in the expected range were detected with a resolution better than 1mm. Although many technical issues have still to be addressed for a real implementation in a clinical environment, the preliminary results of this study suggest that a surrogate of real-time verification of the beam range inside the patient during a treatment with carbon ions is possible by adding a small fraction of helium ions to the primary beam.