Explainable Deep Learning for Personalized Age Prediction With Brain Morphology.

Angela Lombardi,Roberto Bellotti,João Manuel R S Tavares,Alfonso Monaco,Domenico Diacono,Sabina Tangaro,Nicola Amoroso

doi:10.3389/fnins.2021.674055

Abstract

Predicting brain age has become one of the most attractive challenges in computational neuroscience due to the role of the predicted age as an effective biomarker for different brain diseases and conditions. A great variety of machine learning (ML) approaches and deep learning (DL) techniques have been proposed to predict age from brain magnetic resonance imaging scans. If on one hand, DL models could improve performance and reduce model bias compared to other less complex ML methods, on the other hand, they are typically black boxes as do not provide an in-depth understanding of the underlying mechanisms. Explainable Artificial Intelligence (XAI) methods have been recently introduced to provide interpretable decisions of ML and DL algorithms both at local and global level. In this work, we present an explainable DL framework to predict the age of a healthy cohort of subjects from ABIDE I database by using the morphological features extracted from their MRI scans. We embed the two local XAI methods SHAP and LIME to explain the outcomes of the DL models, determine the contribution of each brain morphological descriptor to the final predicted age of each subject and investigate the reliability of the two methods. Our findings indicate that the SHAP method can provide more reliable explanations for the morphological aging mechanisms and be exploited to identify personalized age-related imaging biomarker.

Highlights

Brain age prediction has become a challenging topic in computational neuroscience, due to the strong link between aging processes and several brain disorders and diseases (Franke and Gaser, 2012; Gaser et al, 2013; Koutsouleris et al, 2014; Cole and Franke, 2017b; Wang et al, 2019)
Several works have shown that deep learning (DL) models improve performance and reduce model bias compared to other less complex machine learning (ML) methods (Couvy-Duchesne et al, 2020; Da Costa et al, 2020; Lombardi et al, 2020c); current DL approaches applied to neuroimaging typically do not provide an in-depth understanding of the underlying mechanisms and how they contributed to the outcome
We present an explainable DL framework to predict the age of a healthy cohort of subjects from ABIDE I database (Di Martino et al, 2014) by using morphological features extracted from their MRI scans

Summary

Introduction

Brain age prediction has become a challenging topic in computational neuroscience, due to the strong link between aging processes and several brain disorders and diseases (Franke and Gaser, 2012; Gaser et al, 2013; Koutsouleris et al, 2014; Cole and Franke, 2017b; Wang et al, 2019). Two. Explainable Deep Learning for Personalized Age main approaches are largely adopted to perform brain age prediction: on one hand, a number of selected features such as morphological descriptors, graph-based or other imaging-related features can be extracted from imaging to train different models (Erus et al, 2015; Amoroso et al, 2018, 2019; Bellantuono et al, 2020; Han et al, 2020); on the other hand, more complex models such as convolutional neural networks directly exploiting raw image as input have proven to be effective in brain age prediction even in broad age ranges (Cole et al, 2017, 2019; Feng et al, 2020; Levakov et al, 2020; Peng et al, 2021). These methods aim at estimating the contribution of individual features toward a specific prediction by perturbing a given instance and observing the effect of these perturbations on the output of the model

Results

Discussion

Conclusion