AIFS/Medical Imaging is a medical image analytic platform delicately designed for radiologists and related medical workers for medical image viewing, labeling, AI modeling, and radionics analysis. It adopts a standard DICOM transmission protocol to facilitate image import/export between the institute’s PACS and AIFS/Medical Imaging. Within AIFS/Medical Imaging, users can load image sets to perform (1) manual labeling (classification tasks with class labels, detection tasks with rectangular bounding boxes, and segmentation tasks with polygons), (2) AIAA (AI Assisted Annotation; apply AI to label target objects in the image and followed by expert’s adjustment) and (3) interactive AI annotation (AI guesses the user’s intention and interactively accelerates the labeling process). The labeled image data can seamlessly dispatch to the AI trainer equipped with the autoML(automated Machine Learning) to train or update the AI model. The trained AI model can be used for AIAA in the aforementioned labeling scenario or deployed to an edge as a sole inference use case. All of these operations are done with access to a friendly user interface.

AIFS/Medical Imaging offers a professional medical image viewer that can address different viewing demands for images taken by various modalities (X-ray, CT, and MRI). According to labeling tasks, the image objects are displayed with bounding boxes (in the detection mode) or polygons (in the segmentation mode) with user-defined colors and shades. The segmented areas of interest can be further analyzed by the Radimoics module that produces a high number of features that are useful for predictive ML(Machine Learning) tasks such as the risk of malignancy, prognosis evaluation, treatment suggestion...etc.