The importance of measurement and modelling of the coronal magnetic fields has been appreciated for a long time, and in view of the fact that magnetic fields of Coronal Mass Ejections (CMEs) play a crucial role in determining their geo-effectiveness, this also has considerable societal impact. The coronal magnetic fields are, however, very hard to measure directly. Unlike those at play in X-ray, EUV and optical regimes, radio emission mechanisms are sensitive to the local magnetic fields and hence can potentially lead to their measurements. In practise, however, this potential has been hard to realize. To the best of our knowledge, in the past eighteen years, there have been only three published instances of radio detection of CME like structures. Only for two of these instances, it was possible to estimate coronal magnetic fields by fitting the measured spectra with gyrosynchrotron models. Using data from the Murchison Widefield Array (MWA), for two CMEs we have detected radio structures resembling the CME morphology. These structures are cospatial and cotemporal with the white light coronagraph images of the CME and we can convincingly demonstrate that this is not plasma emission. The maximum heliocentric distance where we can detect such emission is 4.74 solar radii, and the flux densities we measure are among the lowest reported. We note that these detections have been enabled by the confluence of the availability of data from the modern arrays like the MWA; and an automated interferometric solar radio imaging which we have developed. The MWA provides unprecedentedly dense sampling of Fourier (uv) plane, an essential pre-requisite for high imaging fidelity; and the imaging pipeline is tuned to extract the best imaging performance from these data. We present the hypothesis that gyrosynchrotron emission from CMEs intrinsically is not rare, it only appeared so because the dynamic range and imaging fidelity of available solar radio images was typically insufficient to convincingly detect this emission. If true, this then provides an exciting opportunity for routine estimation of the CME magnetic fields at coronal heights and a host of other coronal diagnostics, uniquely associated with gyrosynchrotron emissions.