DigitalFilm Tree

Medical scans are ubiquitous in healthcare. It follows that the plethora of problems which plague that part of the industry are equally pervasive. In our need-finding over the fall and winter quarters, we have encountered problems in patient wait times before and after the scans, communication all round (between radiologists, surgeons, patients and doctors), patient and surgeon visualization and general inefficiencies in the process. Our journey of exploration has led to the twofold problem of long wait times in medical imaging – specifically in the post-scan diagnosis period – and poor communication in the diagnosis segment of medical imaging. This is a particularly intriguing problem in that, for the most part, distinct solutions to one side would be easy to implement and effective, but often to the detriment of the other. We were thus presented with the challenge of having a bi-faceted problem with conflicting objectives; of course, we relished the challenge.

We had initially set out to address long diagnosis times, and prototyped the incorporation of Natural Language Processing (NLP), automated radiology report generation and a hand-tracking control system to help radiologists diagnose faster. However, this raised the aforementioned question of maintaining (or even improving) the quality of information relayed in diagnoses. Furthermore, we could hardly ignore the role of a shared understanding of diagnoses in improving information flow. Having determined we were to expedite and improve the quality of information flow in the medical imaging process - for radiologists, patients and doctors - we mapped a solution system which would encompass the desired functionality.

We implemented a system, Clarity, that assists patients in understanding and visualizing the diagnosis, and radiologists and doctors in communicating medical data between themselves and the patients. On the radiologists’ side, the system allows the segmentation of the areas of interest in the images. These areas are highlighted in all slices which capture it. Radiologists can attach the comments to the highlighted areas, which are later automatically combined into a report hence reducing the work time to deliver a diagnosis to a patient. The 2D CT and MRI slices along with radiologist’s comments, annotations and automatic measurements are converted to a 3D model and are passed to the patients and the doctors. Thus, patients and doctors can have a better understanding of the problem and visualize it in a convenient way. Patients will be able to view a 3D model of their own body with tissue color separation, see segmented problematic areas both in 3D and 2D and obtain an automatically simplified report from a radiologist. By giving patients the access to the 2D images, segmentations and annotations in 3D we believe we solve the problem of poor communication in the diagnosis segment of medical imaging.