A method of identifying regions of interest in patient/medical image data includes automatically identifying and segmenting 104 possible regions of interest (ROI) which may relate to known organs. The patient image data is displayed 106 to a user/clinician upon which verbal comments are received. The comments are received 110 via a voice input (213, figure 7) and may be analysed using voice recognition. An evaluation is undertaken to determine 112, in the context of the displayed image data and received comments, which of the displayed segmented ROIs are most likely to be the subject of the comments. The method may additionally perform automatic measurements 116 upon the determined ROI. A representation (e.g. a labelled view and ROI measurements) of the determined ROI is displayed 118 to the user. Accordingly, the method is able to automatically identify ROIs based upon a clinicians commentary. A system and apparatus are also disclosed. The method facilitates the automatic generation of medical reports with minimal manual intervention, allowing the clinician to focus upon clinical assessment of the image data.
