MedMLX brings 21 established medical imaging models to on-device Apple Silicon

[medRxiv] Sandoval and colleagues developed MedMLX, a collection of 21 established medical imaging models rewritten for Apple Silicon in MLX, Apple's machine learning framework, to run locally rather than on NVIDIA GPUs, according to the medRxiv preprint. The collection includes nnU-Net, TotalSegmentator, MedSAM2 and SAM 3, and covers organ and tumor segmentation, lung nodule detection, synthetic image generation and radiology report drafting from CT, MRI, PET/CT, radiographs and endoscopic images. Using public cases from 11 imaging datasets, the authors compared results with the original software run on an Intel CPU and an NVIDIA GPU; MedMLX agreed in 211 of 212 comparisons, with the single exception a brain MRI segmentation where the original software itself gave different results on the two devices. On an Apple M1 Max with 64 GB of memory, a memory-saving setting kept 19 of the 21 models under 16 GB, cutting memory use by a median of 49% while raising processing time by a median of 13%. The authors said the toolkit enables on-device AI with established imaging models on Apple Silicon rather than specialized NVIDIA hardware.