# GSplit LBI: Taming the Procedural Bias in Neuroimaging for Disease Prediction

## TL;DR

This paper introduces GSplit LBI, a dual-task algorithm that leverages procedural bias in neuroimaging data to improve disease prediction accuracy and feature interpretability, demonstrated on ADNI data.

## Contribution

The paper proposes GSplit LBI, a novel method that separates lesion features from procedural bias, enhancing stability and interpretability in neuroimaging-based disease prediction.

## Key findings

- Improved stability of lesion feature selection
- Enhanced classification accuracy on ADNI data
- Effective separation of procedural bias from lesion features

## Abstract

In voxel-based neuroimage analysis, lesion features have been the main focus in disease prediction due to their interpretability with respect to the related diseases. However, we observe that there exists another type of features introduced during the preprocessing steps and we call them "\textbf{Procedural Bias}". Besides, such bias can be leveraged to improve classification accuracy. Nevertheless, most existing models suffer from either under-fit without considering procedural bias or poor interpretability without differentiating such bias from lesion ones. In this paper, a novel dual-task algorithm namely \emph{GSplit LBI} is proposed to resolve this problem. By introducing an augmented variable enforced to be structural sparsity with a variable splitting term, the estimators for prediction and selecting lesion features can be optimized separately and mutually monitored by each other following an iterative scheme. Empirical experiments have been evaluated on the Alzheimer's Disease Neuroimaging Initiative\thinspace(ADNI) database. The advantage of proposed model is verified by improved stability of selected lesion features and better classification results.