ترجمه مقاله نقش ضروری ارتباطات 6G با چشم انداز صنعت 4.0
- مبلغ: ۸۶,۰۰۰ تومان
ترجمه مقاله پایداری توسعه شهری، تعدیل ساختار صنعتی و کارایی کاربری زمین
- مبلغ: ۹۱,۰۰۰ تومان
The thresholded t-map produced by the General Linear Model (GLM) gives an effective summary of activation patterns in functional brain images and is widely used for feature selection in fMRI related classification tasks. As part of a project to build content-based retrieval systems for fMRI images, we have investigated ways to make GLM more adaptive and more robust in dealing with fMRI data from widely differing experiments. In this paper we report on exploration of the Finite Impulse Response model, combined with multiple linear regression, to identify the “locally best Hemodynamic Response Function (HRF) for each voxel” and to simultaneously estimate activation levels corresponding to several stimulus conditions. The goal is to develop a procedure for processing datasets of varying natures. Our experiments show that Finite Impulse Response (FIR) models with a smoothing factor produce better retrieval performance than does the canonical double gamma HRF in terms of retrieval accuracy.
1 Introduction
As a method for watching “how the brain works”, fMRI has become a powerful research tool in many aspects of neuroscience studies in the past decade [1]. More recently, classification of fMRI images, based on similarity between activation patterns, shows promising transition to clinical diagnosis [2,3,4]. These methods usually select features (that is to say, voxels or areas in the brain and their activation levels) and train models to best distinguish uncommon cases from so-called “typical” ones.
4 Conclusions and Discussions
The results of this study are: confirmation of one hypothesis (H1) , and some tantalizing clues regarding the other. Specifically, the FIR model, with MAP smoothing, which seems to be a more realistic way to describe the variations, across the brain, in the anatomy supplying blood, does also yield significantly better performance in the retrieval setting. This suggest that it may be worth the added effort to use smoothed FIR analysis when preparing data for retrieval across different experiments, and different laboratories.