The Quantum Compressor: A New Way to Find Brain Tumors ⚡ экспресс
Scientists used a quantum autoencoder to find tumors in brain MRI scans. First, the quantum circuit learns to compress healthy tissue in the images, and then flags anything that resists compression as an anomaly. The method achieved high accuracy (around 0.95 ROC-AUC at the slice level), outperforming classical counterparts. The most intriguing part: the key role is played not by the decoder, but by the structured compression in the encoder — much like an experienced archivist who effortlessly separates template documents from unusual ones.
A cook spends years honing the skill of reducing a recipe to three or four essentials and recreating the dish from them. If the ingredients are familiar, the expected meal lands on the plate. A foreign ingredient disrupts compression, and you get something unrecognizable. The Quantum Compressor does the same with medical scans. It slices the image into tiny patches, encodes them into an abstract code, and trains on healthy samples, lowering entropy — information junk. The scans rely on faint signals from hydrogen atoms.
After training, the algorithm easily packs and unpacks normal tissue. A tumor acts like an uninvited ingredient: compression throws a high error that can be measured. On full slices, accuracy reached 95% — better than classical compressors and simple statistical analysis.
🎯 The program needed just 4 qubits to work — like cooking a dinner party on a toy stove.
🎬 The medical scanners on Star Trek found diseases instantly — the first steps toward that future are being born in quantum labs.