In particle therapy, proton or ion beams deposit a large fraction of their energy at the end of their paths, i.e. the delivered dose can be focused on the tumor, sparing nearby tissue due to a low entry and almost no exit dose. A novel imaging modality using protons promises to overcome some limitations of particle therapy and will allow the full exploitation of its potential. Being able to position the so-called Bragg peak accurately inside the tumor is a major advantage of charged particles, but incomplete knowledge about a crucial tissue property, the stopping power, limits its precision. The conversion of photon attenuation maps from computed tomography (CT) scans into relative stopping power introduces range uncertainties. A proton/helium-CT scanner provides direct information about the stopping power and has the potential to reduce range uncertainties significantly, but no proton-CT system has yet been shown to be suitable for clinical use. For a proton-CT scan the particles – typically protons or alpha particles - need to be energetic enough to traverse the patient completely, i.e. the Bragg peak is positioned in a detector. The trajectory of every outgoing proton, as well as the residual energy/range, is measured. The calculation of the proton trajectory inside the target region and the measured residual proton energy/range provide a 3D-map of the relative stopping power. During a scan, the patient needs to be rotated to obtain projection data from a set of different angles. A (clinical) prototype of an extremely high-granularity digitial tracking calorimeter has been designed and is being constructed in Bergen. The latest developments in Monolithic Active Pixel Sensors (MAPS) technology allow the fabrication of extremely-high granularity, low material budget and large area silicon detectors with integration times of microseconds and zero-suppression of the data on the sensor itself. The prototype is a silicon/absorber sandwich calorimeter with 41 sensitive layers of MAPS. A complete CT reconstruction of a simulated anthropomorphic paediatric head phantom shows that the concept of a single-sided detector setup and realistic pencil beam parameters gives a spatial resolution sufficient for proton therapy treatment planning. The expected performance based on simulations, first beam test results and the status of the construction will be presented, e.g. proton tracking accuracy, dE/dx capability, rate capability, radiation hardness and 3D spatial resolution after CT reconstruction.