Fig. 1

This flow chart summarizes the steps performed in the development of a deep learning-based framework model to preoperatively predicting MVI in HCC. The sequences and details of these steps are: (1) labeling of region of interest (ROI; red circle) on an arterial phase CT image; (2) covering the ROI with a square bounding box; (3) cropping ROI with a margin of 0.8; (4) performing data augmentation, including random rotation, random cropping and horizontal flipping, to increase variation of tumor appearance in the training set; (5) resizing all preprocessed images to 256x256x3; (6) utilizing preprocessed images as inputs for ResNet-18 model training; (7) inputting patients’ clinical factors to an fully connected layer with 9 units (FC-9); (8) concatenating output of the FC-9 layer into the last FC layer (with 2 units) of ResNet-18; (9) predicting MVI positive or negative as the outcome. In step (4), some examples of images in data augmentation were provided. For random rotation (images randomly rotating at − 10 to 10 degrees) and horizontal flipping, the rotating angle is shown below each image. For random cropping (images randomly cropped with sizes of 0.8–1.0 and aspect ratios of 0.95–1.05), the cropped size and aspect ratio is shown below each image. The red square box marks out the resulted image with the corresponding cropped size and aspect ratio. CT = computed tomography; ROI: region of interest; MVI = microvascular invasion; MTD = maximum tumor diameter; AFP = alpha-fetoprotein; HBsAg = Hepatitis B surface antigen; HCsAg = Hepatitis C surface antigen; FC = fully connected