Towards an Accurate MRI Acute Ischemic Stroke Lesion Segmentation Based on Bioheat Equation and U-Net Model

Figure 1

A schematic of U-Net architecture trained on thermal images. The input of the network is a normalized thermal 2D image, and the output is the segmentation map using Dice coefficient as a loss function with sigmoid activation function. The architecture includes a downsampling path and an upsampling path, with concatenation between the corresponding layers.