U-Net Prostate MRI Segmentation

Deep Learning Architecture Visualization · Encoder → Bottleneck → Decoder → Output

Input MRI
Encoder L1
Encoder L2
Encoder L3
Encoder L4
Bottleneck
Decoder L4
Decoder L3
Decoder L2
Decoder L1
Output Mask
Original MRI
512×512 · T2w
Segmentation Mask
Prostate Lesion
Speed 1×

Stage Info

Ready to process MRI...
Press Run Inference to begin

Architecture

U-Net with 4 encoder blocks, bottleneck, and 4 decoder blocks. Each encoder stage uses MaxPooling 2×2 for downsampling. Skip connections preserve spatial details for precise segmentation.

Metrics

— Dice Score
— IoU
— Confidence