# Multi-modal low-dose medical imaging through instruction-guided unified AI

## TL;DR

This paper introduces MIRA-Net, a unified AI model that improves low-dose medical imaging across CT, PET, and MRI without sacrificing diagnostic quality.

## Contribution

The novel MIRA-Net architecture enables multi-modal image restoration with a single model using an adaptive instruction-guided framework.

## Key findings

- MIRA-Net matched or outperformed task-specific models in CT denoising, PET synthesis, and MRI super-resolution.
- The model maintained strong performance when trained across all modalities in a unified framework.
- Radiologists rated MIRA-Net outputs as more diagnostic with better anatomical clarity and noise control.

## Abstract

Ionizing radiation from PET/CT warrants dose reduction. However, lowering dose can degrade image quality and affect diagnosis. Many machine-learning approaches exist. Nevertheless, most are built for a single task and are difficult to deploy across multi-modal workflows. We sought to develop and evaluate a unified model that handles common restoration tasks across modalities.

We developed the Multi-modal Instruction-guided Restoration Architecture (MIRA-Net), a U-Net–based framework with an adaptive guidance module. The module estimates modality and degradation indicators from the input and produces a low-dimensional instruction that modulates feature processing throughout the network, selecting task-appropriate pathways within a single model. Performance was assessed on CT denoising, PET synthesis, and MRI super-resolution. Additionally, a double-blind reader study was conducted with board-certified radiologists.

Trained on individual tasks, MIRA-Net matched or exceeded strong task-specific baselines. When trained as a single unified model across CT, PET, and MRI, it maintained comparable performance without a meaningful drop from single-task training. Local clinical dataset validation demonstrated robust generalization with consistent performance metrics. In the reader study, MIRA-Net outputs were more often judged diagnostic and received higher scores for anatomical clarity, lesion conspicuity, and noise control.

MIRA-Net provides a high-fidelity solution for multi-modal medical image restoration. Its instruction-guided architecture successfully mitigates task interference, demonstrating an effective pathway to reducing radiation exposure without sacrificing diagnostic quality.