Presented herein are systems, methods, and architectures related to augmented reality (AR) surgical visualization of one or more dual-modality probe species in tissue. As described herein, near infrared (NIR) images are detected and rendered in real time. The NIR images are registered and/or overlaid with one or more radiological images (e.g., which were obtained preoperatively/perioperatively) by a processor [e.g., that uses an artificial neural network (ANN) or convolutional neural network (CNN) reconstruction algorithm] to produce a real-time AR overlay (3D representation). The AR overlay is displayed to a surgeon in real time. Additionally, a dynamic motion tracker tracks the location of fiducial tracking sensors on/in/about the subject, and this information is also used by the processor in producing (e.g., positionally adjusting) the AR overlay. The real-time AR overlay can improve surgery outcomes, for example, by providing additional real-time information about a surgical site via an intuitive visual interface.