Dhaka based startup Arioron announces a game-changing AI breakthrough in breast cancer detection. Their OncoDetect-BC-CT-Fusion system achieved 100% sensitivity and 99.8% accuracy on external benchmarks—surpassing global leaders like Vision Transformer (93.2%) and ResNet50 (82.9%).
Unlike lab-only models, this AI tackles real-world hospital data: noisy scans from old machines across diverse settings. Built as a "Council of Six" multi-model ensemble with fine-tuned SWIN Transformer tech, it spots subtle cancer patterns others miss.
Trained on 24,576 images from MIAS, DDSM, and INbreast datasets, it's hardware-agnostic—perfect for emerging markets like Bangladesh with scarce radiologists and outdated equipment. It nails 100% benign case detection too, minimizing false positives.
Arioron eyes hospital pilots and regulatory approval next. "From benchmarks to bedside impact," they say—poised to transform early detection in under-resourced healthcare.
