Interpreting DTI-ALPS variability (1.3 to 1.9) in young patients with suspected mTBI: Longitudinal analysis without a healthy control group

1

Hi everyone,

I am conducting a longitudinal dMRI study on a cohort of young individuals with suspected mild Traumatic Brain Injury (mTBI), evaluated at two timepoints:

  • Session 1: Acute/subacute symptomatic stage.

  • Session 2: Follow-up at approximately 3 months (recovery stage).

I do not have a dedicated healthy control group, so each participant acts as their own longitudinal control. Using both the automated gbarisano/alps pipeline (fed with QSIPrep outputs) and manual ROI validation, I observe index values ranging from 1.3 to 1.9 across the cohort. The automated and manual measurements correlate tightly.

Since this is a young population, I have a few methodological and interpretive questions:

  1. In young healthy individuals, baseline DTI-ALPS values often peak around 1.6–1.8. Are values around 1.3 consistent with acute glymphatic/perivascular disruption in symptomatic youth, while 1.8–1.9 represents spared/unaffected status?

  2. Without a normative reference group, what is the best statistical approach to evaluate normalization over the 3-month window? Is modeling ΔALPS against symptom recovery scores sufficient, or should I normalize against a reference tract (e.g., standard MD/FA in the corpus callosum)?

Thanks in advance for sharing your experience and advice.

2

1.3 in an acute symptomatic kid is plausibly disruption rather than noise, especially since your auto and manual ROIs agree with each other. the literature baseline for young adults sits around 1.6 to 1.8, so 1.3 is well below that. id be careful about calling 1.8-1.9 spared though, without controls you cant really say. for the stats, modeling delta ALPS against delta symptom scores with a mixed model is the honest approach here. id resist normalizing to callosal FA/MD, it eats the effect you care about. one thing to watch: ALPS test-retest reliability is only moderate, so check your within-subject measurement noise before interpreting small deltas.