This session focuses on and tries to appraise diffusion MRI (dMRI)-derived imaging biomarkers (dMRI-IB) for brain disease. To categorize dMRI-IB: 1) confirm the definition of IB; 2) proceed through dMRI-derived measures and applications; 3) introduce the possible candidate of dMRI-IB; and 4) discuss about barriers that dMRI-IB candidates should overcome. In addition, this session will address following issues: 1) why only limited dMRI measures can be considered for IB? 2) what is the problem for dMRI-IB candidates to become true IB? 3) what can we do for creating new IB?
Several decades have passed since diffusion MRI (dMRI) was established.
This session will address whether the dMRI derived measures can be considered as biomarkers.
To outline whether the dMRI derived measures can be used as biomarkers: I will
1) confirm the definition of biomarker [1, 2], especially an “imaging biomarker (IB)” [3, 4];
2) review the dMRI derived measures and their common clinical applications [5, 6, 7, 8];
3) introduce the candidates of dMRI-IB for brain diseases; and
4) discuss about the dMRI parameters that can be useful to define IB?
The definition of biomarker was proposed by FDA [1] in 2001 as follows:
“a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention.”
The definition of “imaging biomarker” was proposed by QIBA (RSNA) [4] in 2015 as follows:
“an objective characteristic derived from an in vivo image measured on a ratio or interval scale as an indicator of normal biological processes, pathogenic processes or a response to a therapeutic intervention”. Imaging biomarker can be divided four different purposes [9]:
diagnosis/prognosis, monitoring, predictive, response.
dMRI derived measures are now utilizing these four purposes.
Since dMRI was established [10], dMRI derived measures and their applications have been developed and widely spread from brain to whole body.
Well-known dMRI derived measurement methods are as follows:
DWI [11], IVIM [11], QSI [12], DTI [13], DSI [14], HARDI [15], QBI [16], CHARMED [17], DKI [18], AxCaliber [19], NODDI [20], etc. dMRI derived measurements are: ADC [11] by DWI; ADC, FA [21], and tractography by DTI [22, 23, 24, 25, 26, 27], by DSI [28], and by QBI [29]; MK by DKI [18]; MD, MK, and ZDP by QSI [30]; and NDIODI (orientation dispersion index), vin (intra-neurite tissue volume fraction), a (neurite density index), and viso (isotropic volume fraction) by NODDI [31].
Among these dMRI derived measurements, which one is recognized as an imaging biomarker?
For clinically typical brain diseases, such as ischemia, Alzheimer’s disease (AD) [35-37], and traumatic brain injury (TBI) [38-43], dMRI derived measures have been extensively studied.
For the diagnosis of ischemic stroke, clinical utility of DWI and ADC have been studied [32].For the treatment of acute phase of ischemic stroke, diffusion-perfusion mismatch has been also studied [33] and clinically utilized [34].
dMRI-IB candidates can be divided into two major categories [44].
Those are the “clinical research tool” and “decision making tool”.
To make it become a “clinical research tool”, we typically have to confirm only the possibility and applicability. On the other hands, “decision making tool” is required to fit the definition of IB through major study with attention to evidence-based medicine [45].
Therefore, to create and apply a new dMRI-IB, the candidate parameters should fulfill a stepwise process of “validation”, “qualification”, and “utilization”.
We need to recognize the difference between these two categories and select appropriate tool for our purpose.
As take home messages, learners can recognize:
1) some dMRI derived measures have already been used as biomarkers for certain brain diseases,
2) dMRI-IB candidates can be divided into two categories,
3) we need multi-center collaboration for creating new dMRI-IB to clarify “validation”, “qualification”, and “utilization”.
Continuous effort by interdisciplinary team (medical, informatics, engineering, etc.) may create a better dMRI-IB.